API tools faq
paste
Advertisement
ziyuang

theano-test (238d2a84) output

ziyuang
Mar 10th, 2015
305
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 696.65 KB | None | 0 0
raw download clone embed print report
  1. ======================================================================
  2. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 64, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3. ----------------------------------------------------------------------
  4. Traceback (most recent call last):
  5. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6. self.test(*self.arg)
  7. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  8. gpuval = f(img, kern)
  9. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  10. storage_map=self.fn.storage_map)
  11. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  12. outputs = self.fn()
  13. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  14. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  15. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, False, True)), CudaNdarrayType(float32, (True, True, False, True)), <theano.gof.type.CDataType object at 0x7f85ac156110>, Scalar(float32)]
  16. Inputs shapes: [(2, 1, 1, 1), (1, 2, 2, 1), (1, 1, 2, 1), 'No shapes', ()]
  17. Inputs strides: [(1, 0, 0, 0), (0, 4, 2, 0), (0, 0, 1, 0), 'No strides', ()]
  18. Inputs values: [<CudaNdarray object at 0x7f85c7465a70>, <CudaNdarray object at 0x7f85d8901470>, <CudaNdarray object at 0x7f85c74659b0>, <PyCObject object at 0x7f85c5a5f918>, 1.0]
  19.  
  20. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  21. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  22.  
  23. ======================================================================
  24. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 2, 2), (1, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 65, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  25. ----------------------------------------------------------------------
  26. Traceback (most recent call last):
  27. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  28. self.test(*self.arg)
  29. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  30. gpuval = f(img, kern)
  31. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  32. storage_map=self.fn.storage_map)
  33. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  34. outputs = self.fn()
  35. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  36. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  37. Inputs types: [CudaNdarrayType(float32, (True, True, True, True)), CudaNdarrayType(float32, (False, True, False, True)), CudaNdarrayType(float32, (False, True, False, True)), <theano.gof.type.CDataType object at 0x7f85aa13f690>, Scalar(float32)]
  38. Inputs shapes: [(1, 1, 1, 1), (3, 1, 2, 1), (3, 1, 2, 1), 'No shapes', ()]
  39. Inputs strides: [(0, 0, 0, 0), (4, 0, 2, 0), (2, 0, 1, 0), 'No strides', ()]
  40. Inputs values: [<CudaNdarray object at 0x7f85dcd7bbb0>, 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fdc8>, 1.0]
  41.  
  42. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  43. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  44.  
  45. ======================================================================
  46. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 2, 2), (2, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 66, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  47. ----------------------------------------------------------------------
  48. Traceback (most recent call last):
  49. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  50. self.test(*self.arg)
  51. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  52. gpuval = f(img, kern)
  53. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  54. storage_map=self.fn.storage_map)
  55. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  56. outputs = self.fn()
  57. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  58. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  59. Inputs types: [CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, True, False, True)), CudaNdarrayType(float32, (True, False, False, True)), <theano.gof.type.CDataType object at 0x7f85af310dd0>, Scalar(float32)]
  60. Inputs shapes: [(1, 2, 1, 1), (1, 1, 2, 1), (1, 2, 2, 1), 'No shapes', ()]
  61. Inputs strides: [(0, 1, 0, 0), (0, 0, 2, 0), (0, 2, 1, 0), 'No strides', ()]
  62. Inputs values: [<CudaNdarray object at 0x7f85c74fec30>, <CudaNdarray object at 0x7f85e0743070>, <CudaNdarray object at 0x7f85ac721270>, <PyCObject object at 0x7f85c5a5f198>, 1.0]
  63.  
  64. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  65. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  66.  
  67. ======================================================================
  68. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 2, 2), (2, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 67, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  69. ----------------------------------------------------------------------
  70. Traceback (most recent call last):
  71. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  72. self.test(*self.arg)
  73. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  74. gpuval = f(img, kern)
  75. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  76. storage_map=self.fn.storage_map)
  77. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  78. outputs = self.fn()
  79. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  80. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  81. Inputs types: [CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (False, True, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85af40de50>, Scalar(float32)]
  82. Inputs shapes: [(1, 2, 1, 1), (3, 1, 2, 1), (3, 2, 2, 1), 'No shapes', ()]
  83. Inputs strides: [(0, 1, 0, 0), (4, 0, 2, 0), (4, 2, 1, 0), 'No strides', ()]
  84. Inputs values: [<CudaNdarray object at 0x7f85e728acb0>, 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f698>, 1.0]
  85.  
  86. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  87. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  88.  
  89. ======================================================================
  90. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 68, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  91. ----------------------------------------------------------------------
  92. Traceback (most recent call last):
  93. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  94. self.test(*self.arg)
  95. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  96. gpuval = f(img, kern)
  97. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  98. storage_map=self.fn.storage_map)
  99. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  100. outputs = self.fn()
  101. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  102. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  103. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, True, False, True)), <theano.gof.type.CDataType object at 0x7f85af3f4750>, Scalar(float32)]
  104. Inputs shapes: [(2, 1, 1, 1), (3, 2, 2, 1), (3, 1, 2, 1), 'No shapes', ()]
  105. Inputs strides: [(1, 0, 0, 0), (8, 4, 2, 0), (2, 0, 1, 0), 'No strides', ()]
  106. Inputs values: [<CudaNdarray object at 0x7f85dcd969b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5ffa8>, 1.0]
  107.  
  108. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  109. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  110.  
  111. ======================================================================
  112. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 69, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  113. ----------------------------------------------------------------------
  114. Traceback (most recent call last):
  115. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  116. self.test(*self.arg)
  117. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  118. gpuval = f(img, kern)
  119. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  120. storage_map=self.fn.storage_map)
  121. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  122. outputs = self.fn()
  123. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  124. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  125. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, False, True)), CudaNdarrayType(float32, (True, False, False, True)), <theano.gof.type.CDataType object at 0x7f85dcd3e190>, Scalar(float32)]
  126. Inputs shapes: [(2, 2, 1, 1), (1, 2, 2, 1), (1, 2, 2, 1), 'No shapes', ()]
  127. Inputs strides: [(2, 1, 0, 0), (0, 4, 2, 0), (0, 2, 1, 0), 'No strides', ()]
  128. Inputs values: [<CudaNdarray object at 0x7f85ac721c30>, <CudaNdarray object at 0x7f85aa602bb0>, <CudaNdarray object at 0x7f85ac721df0>, <PyCObject object at 0x7f85c5a5ff08>, 1.0]
  129.  
  130. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  131. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  132.  
  133. ======================================================================
  134. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 70, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  135. ----------------------------------------------------------------------
  136. Traceback (most recent call last):
  137. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  138. self.test(*self.arg)
  139. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  140. gpuval = f(img, kern)
  141. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  142. storage_map=self.fn.storage_map)
  143. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  144. outputs = self.fn()
  145. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  146. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  147. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85abd5c890>, Scalar(float32)]
  148. Inputs shapes: [(2, 2, 1, 1), (2, 2, 2, 1), (2, 2, 2, 1), 'No shapes', ()]
  149. Inputs strides: [(2, 1, 0, 0), (8, 4, 2, 0), (4, 2, 1, 0), 'No strides', ()]
  150. Inputs values: [<CudaNdarray object at 0x7f85ac5ab8f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f4b8>, 1.0]
  151.  
  152. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  153. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  154.  
  155. ======================================================================
  156. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 71, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  157. ----------------------------------------------------------------------
  158. Traceback (most recent call last):
  159. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  160. self.test(*self.arg)
  161. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  162. gpuval = f(img, kern)
  163. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  164. storage_map=self.fn.storage_map)
  165. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  166. outputs = self.fn()
  167. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  168. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  169. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85d8997110>, Scalar(float32)]
  170. Inputs shapes: [(2, 4, 1, 1), (3, 2, 2, 1), (3, 4, 2, 1), 'No shapes', ()]
  171. Inputs strides: [(4, 1, 0, 0), (8, 4, 2, 0), (8, 2, 1, 0), 'No strides', ()]
  172. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5ff80>, 1.0]
  173.  
  174. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  175. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  176.  
  177. ======================================================================
  178. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 72, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  179. ----------------------------------------------------------------------
  180. Traceback (most recent call last):
  181. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  182. self.test(*self.arg)
  183. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  184. gpuval = f(img, kern)
  185. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  186. storage_map=self.fn.storage_map)
  187. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  188. outputs = self.fn()
  189. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  190. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  191. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c748a610>, Scalar(float32)]
  192. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 2), (1, 1, 5, 3), 'No shapes', ()]
  193. Inputs strides: [(4, 0, 2, 1), (0, 16, 4, 2), (0, 0, 3, 1), 'No strides', ()]
  194. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b00a8>, 1.0]
  195.  
  196. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  197. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  198.  
  199. ======================================================================
  200. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 73, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  201. ----------------------------------------------------------------------
  202. Traceback (most recent call last):
  203. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  204. self.test(*self.arg)
  205. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  206. gpuval = f(img, kern)
  207. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  208. storage_map=self.fn.storage_map)
  209. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  210. outputs = self.fn()
  211. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  212. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  213. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c7422910>, Scalar(float32)]
  214. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 2), (3, 1, 5, 3), 'No shapes', ()]
  215. Inputs strides: [(0, 0, 2, 1), (16, 0, 4, 2), (15, 0, 3, 1), 'No strides', ()]
  216. Inputs values: [<CudaNdarray object at 0x7f85ac5ab770>, 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0878>, 1.0]
  217.  
  218. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  219. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  220.  
  221. ======================================================================
  222. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 74, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  223. ----------------------------------------------------------------------
  224. Traceback (most recent call last):
  225. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  226. self.test(*self.arg)
  227. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  228. gpuval = f(img, kern)
  229. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  230. storage_map=self.fn.storage_map)
  231. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  232. outputs = self.fn()
  233. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  234. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  235. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af3f4c90>, Scalar(float32)]
  236. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 2), (1, 2, 5, 3), 'No shapes', ()]
  237. Inputs strides: [(0, 4, 2, 1), (0, 0, 4, 2), (0, 15, 3, 1), 'No strides', ()]
  238. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0bc0>, 1.0]
  239.  
  240. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  241. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  242.  
  243. ======================================================================
  244. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 75, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  245. ----------------------------------------------------------------------
  246. Traceback (most recent call last):
  247. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  248. self.test(*self.arg)
  249. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  250. gpuval = f(img, kern)
  251. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  252. storage_map=self.fn.storage_map)
  253. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  254. outputs = self.fn()
  255. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  256. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  257. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85abdd97d0>, Scalar(float32)]
  258. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 2), (3, 2, 5, 3), 'No shapes', ()]
  259. Inputs strides: [(0, 4, 2, 1), (16, 0, 4, 2), (30, 15, 3, 1), 'No strides', ()]
  260. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b00d0>, 1.0]
  261.  
  262. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  263. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  264.  
  265. ======================================================================
  266. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 76, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  267. ----------------------------------------------------------------------
  268. Traceback (most recent call last):
  269. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  270. self.test(*self.arg)
  271. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  272. gpuval = f(img, kern)
  273. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  274. storage_map=self.fn.storage_map)
  275. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  276. outputs = self.fn()
  277. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  278. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  279. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcced5d0>, Scalar(float32)]
  280. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 2), (3, 1, 5, 3), 'No shapes', ()]
  281. Inputs strides: [(4, 0, 2, 1), (32, 16, 4, 2), (15, 0, 3, 1), 'No strides', ()]
  282. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0eb8>, 1.0]
  283.  
  284. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  285. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  286.  
  287. ======================================================================
  288. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 77, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  289. ----------------------------------------------------------------------
  290. Traceback (most recent call last):
  291. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  292. self.test(*self.arg)
  293. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  294. gpuval = f(img, kern)
  295. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  296. storage_map=self.fn.storage_map)
  297. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  298. outputs = self.fn()
  299. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  300. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  301. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e6c4ab90>, Scalar(float32)]
  302. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 2), (1, 2, 5, 3), 'No shapes', ()]
  303. Inputs strides: [(8, 4, 2, 1), (0, 16, 4, 2), (0, 15, 3, 1), 'No strides', ()]
  304. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85df3b11e8>, 1.0]
  305.  
  306. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  307. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  308.  
  309. ======================================================================
  310. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 78, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  311. ----------------------------------------------------------------------
  312. Traceback (most recent call last):
  313. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  314. self.test(*self.arg)
  315. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  316. gpuval = f(img, kern)
  317. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  318. storage_map=self.fn.storage_map)
  319. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  320. outputs = self.fn()
  321. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  322. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  323. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d3c31ed0>, Scalar(float32)]
  324. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 2), (2, 2, 5, 3), 'No shapes', ()]
  325. Inputs strides: [(8, 4, 2, 1), (32, 16, 4, 2), (30, 15, 3, 1), 'No strides', ()]
  326. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b490>, 1.0]
  327.  
  328. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  329. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  330.  
  331. ======================================================================
  332. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 79, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  333. ----------------------------------------------------------------------
  334. Traceback (most recent call last):
  335. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  336. self.test(*self.arg)
  337. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  338. gpuval = f(img, kern)
  339. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  340. storage_map=self.fn.storage_map)
  341. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  342. outputs = self.fn()
  343. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  344. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  345. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d93f1510>, Scalar(float32)]
  346. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 2), (3, 4, 5, 3), 'No shapes', ()]
  347. Inputs strides: [(16, 4, 2, 1), (32, 16, 4, 2), (60, 15, 3, 1), 'No strides', ()]
  348. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b710>, 1.0]
  349.  
  350. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  351. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  352.  
  353. ======================================================================
  354. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 80, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  355. ----------------------------------------------------------------------
  356. Traceback (most recent call last):
  357. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  358. self.test(*self.arg)
  359. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  360. gpuval = f(img, kern)
  361. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  362. storage_map=self.fn.storage_map)
  363. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  364. outputs = self.fn()
  365. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  366. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  367. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85a972bf10>, Scalar(float32)]
  368. Inputs shapes: [(2, 1, 2, 2), (1, 2, 6, 3), (1, 1, 7, 4), 'No shapes', ()]
  369. Inputs strides: [(4, 0, 2, 1), (0, 36, 6, 2), (0, 0, 4, 1), 'No strides', ()]
  370. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b940>, 1.0]
  371.  
  372. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  373. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  374.  
  375. ======================================================================
  376. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 81, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  377. ----------------------------------------------------------------------
  378. Traceback (most recent call last):
  379. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  380. self.test(*self.arg)
  381. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  382. gpuval = f(img, kern)
  383. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  384. storage_map=self.fn.storage_map)
  385. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  386. outputs = self.fn()
  387. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  388. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  389. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c73dd7d0>, Scalar(float32)]
  390. Inputs shapes: [(1, 1, 2, 2), (3, 1, 6, 3), (3, 1, 7, 4), 'No shapes', ()]
  391. Inputs strides: [(0, 0, 2, 1), (36, 0, 6, 2), (28, 0, 4, 1), 'No strides', ()]
  392. Inputs values: [<CudaNdarray object at 0x7f85d9465db0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3bbc0>, 1.0]
  393.  
  394. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  395. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  396.  
  397. ======================================================================
  398. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 82, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  399. ----------------------------------------------------------------------
  400. Traceback (most recent call last):
  401. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  402. self.test(*self.arg)
  403. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  404. gpuval = f(img, kern)
  405. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  406. storage_map=self.fn.storage_map)
  407. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  408. outputs = self.fn()
  409. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  410. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  411. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85abdcf650>, Scalar(float32)]
  412. Inputs shapes: [(1, 2, 2, 2), (1, 1, 6, 3), (1, 2, 7, 4), 'No shapes', ()]
  413. Inputs strides: [(0, 4, 2, 1), (0, 0, 6, 2), (0, 28, 4, 1), 'No strides', ()]
  414. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3be40>, 1.0]
  415.  
  416. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  417. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  418.  
  419. ======================================================================
  420. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 83, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  421. ----------------------------------------------------------------------
  422. Traceback (most recent call last):
  423. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  424. self.test(*self.arg)
  425. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  426. gpuval = f(img, kern)
  427. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  428. storage_map=self.fn.storage_map)
  429. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  430. outputs = self.fn()
  431. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  432. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  433. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e7277050>, Scalar(float32)]
  434. Inputs shapes: [(1, 2, 2, 2), (3, 1, 6, 3), (3, 2, 7, 4), 'No shapes', ()]
  435. Inputs strides: [(0, 4, 2, 1), (36, 0, 6, 2), (56, 28, 4, 1), 'No strides', ()]
  436. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85df384120>, 1.0]
  437.  
  438. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  439. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  440.  
  441. ======================================================================
  442. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 84, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  443. ----------------------------------------------------------------------
  444. Traceback (most recent call last):
  445. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  446. self.test(*self.arg)
  447. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  448. gpuval = f(img, kern)
  449. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  450. storage_map=self.fn.storage_map)
  451. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  452. outputs = self.fn()
  453. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  454. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  455. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85aa8377d0>, Scalar(float32)]
  456. Inputs shapes: [(2, 1, 2, 2), (3, 2, 6, 3), (3, 1, 7, 4), 'No shapes', ()]
  457. Inputs strides: [(4, 0, 2, 1), (72, 36, 6, 2), (28, 0, 4, 1), 'No strides', ()]
  458. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0198>, 1.0]
  459.  
  460. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  461. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  462.  
  463. ======================================================================
  464. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 85, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  465. ----------------------------------------------------------------------
  466. Traceback (most recent call last):
  467. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  468. self.test(*self.arg)
  469. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  470. gpuval = f(img, kern)
  471. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  472. storage_map=self.fn.storage_map)
  473. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  474. outputs = self.fn()
  475. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  476. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  477. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af27a590>, Scalar(float32)]
  478. Inputs shapes: [(2, 2, 2, 2), (1, 2, 6, 3), (1, 2, 7, 4), 'No shapes', ()]
  479. Inputs strides: [(8, 4, 2, 1), (0, 36, 6, 2), (0, 28, 4, 1), 'No strides', ()]
  480. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0580>, 1.0]
  481.  
  482. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  483. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  484.  
  485. ======================================================================
  486. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 86, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  487. ----------------------------------------------------------------------
  488. Traceback (most recent call last):
  489. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  490. self.test(*self.arg)
  491. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  492. gpuval = f(img, kern)
  493. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  494. storage_map=self.fn.storage_map)
  495. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  496. outputs = self.fn()
  497. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  498. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  499. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ab6a5290>, Scalar(float32)]
  500. Inputs shapes: [(2, 2, 2, 2), (2, 2, 6, 3), (2, 2, 7, 4), 'No shapes', ()]
  501. Inputs strides: [(8, 4, 2, 1), (72, 36, 6, 2), (56, 28, 4, 1), 'No strides', ()]
  502. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0120>, 1.0]
  503.  
  504. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  505. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  506.  
  507. ======================================================================
  508. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 87, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  509. ----------------------------------------------------------------------
  510. Traceback (most recent call last):
  511. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  512. self.test(*self.arg)
  513. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  514. gpuval = f(img, kern)
  515. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  516. storage_map=self.fn.storage_map)
  517. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  518. outputs = self.fn()
  519. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  520. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  521. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dcd4dc50>, Scalar(float32)]
  522. Inputs shapes: [(2, 4, 2, 2), (3, 2, 6, 3), (3, 4, 7, 4), 'No shapes', ()]
  523. Inputs strides: [(16, 4, 2, 1), (72, 36, 6, 2), (112, 28, 4, 1), 'No strides', ()]
  524. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0a30>, 1.0]
  525.  
  526. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  527. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  528.  
  529. ======================================================================
  530. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 88, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  531. ----------------------------------------------------------------------
  532. Traceback (most recent call last):
  533. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  534. self.test(*self.arg)
  535. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  536. gpuval = f(img, kern)
  537. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  538. storage_map=self.fn.storage_map)
  539. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  540. outputs = self.fn()
  541. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  542. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  543. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e06ecdd0>, Scalar(float32)]
  544. Inputs shapes: [(2, 1, 2, 2), (1, 2, 8, 3), (1, 1, 9, 4), 'No shapes', ()]
  545. Inputs strides: [(4, 0, 2, 1), (0, 48, 6, 2), (0, 0, 4, 1), 'No strides', ()]
  546. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee12c60>, 1.0]
  547.  
  548. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  549. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  550.  
  551. ======================================================================
  552. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 89, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  553. ----------------------------------------------------------------------
  554. Traceback (most recent call last):
  555. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  556. self.test(*self.arg)
  557. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  558. gpuval = f(img, kern)
  559. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  560. storage_map=self.fn.storage_map)
  561. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  562. outputs = self.fn()
  563. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  564. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  565. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85aa826a10>, Scalar(float32)]
  566. Inputs shapes: [(1, 1, 2, 2), (3, 1, 8, 3), (3, 1, 9, 4), 'No shapes', ()]
  567. Inputs strides: [(0, 0, 2, 1), (48, 0, 6, 2), (36, 0, 4, 1), 'No strides', ()]
  568. Inputs values: [<CudaNdarray object at 0x7f85aa837df0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dee121c0>, 1.0]
  569.  
  570. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  571. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  572.  
  573. ======================================================================
  574. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 90, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  575. ----------------------------------------------------------------------
  576. Traceback (most recent call last):
  577. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  578. self.test(*self.arg)
  579. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  580. gpuval = f(img, kern)
  581. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  582. storage_map=self.fn.storage_map)
  583. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  584. outputs = self.fn()
  585. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  586. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  587. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e06ec750>, Scalar(float32)]
  588. Inputs shapes: [(1, 2, 2, 2), (1, 1, 8, 3), (1, 2, 9, 4), 'No shapes', ()]
  589. Inputs strides: [(0, 4, 2, 1), (0, 0, 6, 2), (0, 36, 4, 1), 'No strides', ()]
  590. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4e90>, 1.0]
  591.  
  592. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  593. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  594.  
  595. ======================================================================
  596. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 91, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  597. ----------------------------------------------------------------------
  598. Traceback (most recent call last):
  599. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  600. self.test(*self.arg)
  601. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  602. gpuval = f(img, kern)
  603. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  604. storage_map=self.fn.storage_map)
  605. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  606. outputs = self.fn()
  607. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  608. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  609. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ddac4750>, Scalar(float32)]
  610. Inputs shapes: [(1, 2, 2, 2), (3, 1, 8, 3), (3, 2, 9, 4), 'No shapes', ()]
  611. Inputs strides: [(0, 4, 2, 1), (48, 0, 6, 2), (72, 36, 4, 1), 'No strides', ()]
  612. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4be8>, 1.0]
  613.  
  614. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  615. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  616.  
  617. ======================================================================
  618. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 92, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  619. ----------------------------------------------------------------------
  620. Traceback (most recent call last):
  621. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  622. self.test(*self.arg)
  623. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  624. gpuval = f(img, kern)
  625. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  626. storage_map=self.fn.storage_map)
  627. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  628. outputs = self.fn()
  629. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  630. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  631. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af111f10>, Scalar(float32)]
  632. Inputs shapes: [(2, 1, 2, 2), (3, 2, 8, 3), (3, 1, 9, 4), 'No shapes', ()]
  633. Inputs strides: [(4, 0, 2, 1), (96, 48, 6, 2), (36, 0, 4, 1), 'No strides', ()]
  634. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4fa8>, 1.0]
  635.  
  636. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  637. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  638.  
  639. ======================================================================
  640. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 93, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  641. ----------------------------------------------------------------------
  642. Traceback (most recent call last):
  643. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  644. self.test(*self.arg)
  645. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  646. gpuval = f(img, kern)
  647. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  648. storage_map=self.fn.storage_map)
  649. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  650. outputs = self.fn()
  651. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  652. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  653. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c73b4590>, Scalar(float32)]
  654. Inputs shapes: [(2, 2, 2, 2), (1, 2, 8, 3), (1, 2, 9, 4), 'No shapes', ()]
  655. Inputs strides: [(8, 4, 2, 1), (0, 48, 6, 2), (0, 36, 4, 1), 'No strides', ()]
  656. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4fd0>, 1.0]
  657.  
  658. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  659. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  660.  
  661. ======================================================================
  662. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 94, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  663. ----------------------------------------------------------------------
  664. Traceback (most recent call last):
  665. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  666. self.test(*self.arg)
  667. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  668. gpuval = f(img, kern)
  669. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  670. storage_map=self.fn.storage_map)
  671. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  672. outputs = self.fn()
  673. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  674. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  675. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e06c1390>, Scalar(float32)]
  676. Inputs shapes: [(2, 2, 2, 2), (2, 2, 8, 3), (2, 2, 9, 4), 'No shapes', ()]
  677. Inputs strides: [(8, 4, 2, 1), (96, 48, 6, 2), (72, 36, 4, 1), 'No strides', ()]
  678. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614878>, 1.0]
  679.  
  680. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  681. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  682.  
  683. ======================================================================
  684. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 95, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  685. ----------------------------------------------------------------------
  686. Traceback (most recent call last):
  687. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  688. self.test(*self.arg)
  689. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  690. gpuval = f(img, kern)
  691. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  692. storage_map=self.fn.storage_map)
  693. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  694. outputs = self.fn()
  695. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  696. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  697. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85abd28690>, Scalar(float32)]
  698. Inputs shapes: [(2, 4, 2, 2), (3, 2, 8, 3), (3, 4, 9, 4), 'No shapes', ()]
  699. Inputs strides: [(16, 4, 2, 1), (96, 48, 6, 2), (144, 36, 4, 1), 'No strides', ()]
  700. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a96143c8>, 1.0]
  701.  
  702. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  703. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  704.  
  705. ======================================================================
  706. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 96, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  707. ----------------------------------------------------------------------
  708. Traceback (most recent call last):
  709. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  710. self.test(*self.arg)
  711. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  712. gpuval = f(img, kern)
  713. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  714. storage_map=self.fn.storage_map)
  715. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  716. outputs = self.fn()
  717. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  718. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  719. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e7282c90>, Scalar(float32)]
  720. Inputs shapes: [(2, 1, 2, 2), (1, 2, 6, 4), (1, 1, 7, 5), 'No shapes', ()]
  721. Inputs strides: [(4, 0, 2, 1), (0, 48, 8, 2), (0, 0, 5, 1), 'No strides', ()]
  722. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614d78>, 1.0]
  723.  
  724. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  725. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  726.  
  727. ======================================================================
  728. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 97, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  729. ----------------------------------------------------------------------
  730. Traceback (most recent call last):
  731. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  732. self.test(*self.arg)
  733. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  734. gpuval = f(img, kern)
  735. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  736. storage_map=self.fn.storage_map)
  737. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  738. outputs = self.fn()
  739. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  740. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  741. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abdcf710>, Scalar(float32)]
  742. Inputs shapes: [(1, 1, 2, 2), (3, 1, 6, 4), (3, 1, 7, 5), 'No shapes', ()]
  743. Inputs strides: [(0, 0, 2, 1), (48, 0, 8, 2), (35, 0, 5, 1), 'No strides', ()]
  744. Inputs values: [<CudaNdarray object at 0x7f85ac722170>, 'not shown', 'not shown', <PyCObject object at 0x7f85a9614288>, 1.0]
  745.  
  746. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  747. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  748.  
  749. ======================================================================
  750. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 98, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  751. ----------------------------------------------------------------------
  752. Traceback (most recent call last):
  753. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  754. self.test(*self.arg)
  755. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  756. gpuval = f(img, kern)
  757. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  758. storage_map=self.fn.storage_map)
  759. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  760. outputs = self.fn()
  761. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  762. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  763. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85a9fabf90>, Scalar(float32)]
  764. Inputs shapes: [(1, 2, 2, 2), (1, 1, 6, 4), (1, 2, 7, 5), 'No shapes', ()]
  765. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 2), (0, 35, 5, 1), 'No strides', ()]
  766. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614170>, 1.0]
  767.  
  768. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  769. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  770.  
  771. ======================================================================
  772. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 99, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  773. ----------------------------------------------------------------------
  774. Traceback (most recent call last):
  775. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  776. self.test(*self.arg)
  777. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  778. gpuval = f(img, kern)
  779. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  780. storage_map=self.fn.storage_map)
  781. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  782. outputs = self.fn()
  783. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  784. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  785. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7503950>, Scalar(float32)]
  786. Inputs shapes: [(1, 2, 2, 2), (3, 1, 6, 4), (3, 2, 7, 5), 'No shapes', ()]
  787. Inputs strides: [(0, 4, 2, 1), (48, 0, 8, 2), (70, 35, 5, 1), 'No strides', ()]
  788. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf12b0>, 1.0]
  789.  
  790. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  791. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  792.  
  793. ======================================================================
  794. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 100, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  795. ----------------------------------------------------------------------
  796. Traceback (most recent call last):
  797. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  798. self.test(*self.arg)
  799. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  800. gpuval = f(img, kern)
  801. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  802. storage_map=self.fn.storage_map)
  803. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  804. outputs = self.fn()
  805. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  806. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  807. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d88dce10>, Scalar(float32)]
  808. Inputs shapes: [(2, 1, 2, 2), (3, 2, 6, 4), (3, 1, 7, 5), 'No shapes', ()]
  809. Inputs strides: [(4, 0, 2, 1), (96, 48, 8, 2), (35, 0, 5, 1), 'No strides', ()]
  810. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1c10>, 1.0]
  811.  
  812. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  813. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  814.  
  815. ======================================================================
  816. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 101, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  817. ----------------------------------------------------------------------
  818. Traceback (most recent call last):
  819. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  820. self.test(*self.arg)
  821. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  822. gpuval = f(img, kern)
  823. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  824. storage_map=self.fn.storage_map)
  825. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  826. outputs = self.fn()
  827. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  828. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  829. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85aa829410>, Scalar(float32)]
  830. Inputs shapes: [(2, 2, 2, 2), (1, 2, 6, 4), (1, 2, 7, 5), 'No shapes', ()]
  831. Inputs strides: [(8, 4, 2, 1), (0, 48, 8, 2), (0, 35, 5, 1), 'No strides', ()]
  832. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf12d8>, 1.0]
  833.  
  834. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  835. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  836.  
  837. ======================================================================
  838. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 102, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  839. ----------------------------------------------------------------------
  840. Traceback (most recent call last):
  841. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  842. self.test(*self.arg)
  843. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  844. gpuval = f(img, kern)
  845. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  846. storage_map=self.fn.storage_map)
  847. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  848. outputs = self.fn()
  849. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  850. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  851. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e072a2d0>, Scalar(float32)]
  852. Inputs shapes: [(2, 2, 2, 2), (2, 2, 6, 4), (2, 2, 7, 5), 'No shapes', ()]
  853. Inputs strides: [(8, 4, 2, 1), (96, 48, 8, 2), (70, 35, 5, 1), 'No strides', ()]
  854. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1aa8>, 1.0]
  855.  
  856. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  857. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  858.  
  859. ======================================================================
  860. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 103, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  861. ----------------------------------------------------------------------
  862. Traceback (most recent call last):
  863. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  864. self.test(*self.arg)
  865. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  866. gpuval = f(img, kern)
  867. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  868. storage_map=self.fn.storage_map)
  869. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  870. outputs = self.fn()
  871. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  872. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  873. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dcd71950>, Scalar(float32)]
  874. Inputs shapes: [(2, 4, 2, 2), (3, 2, 6, 4), (3, 4, 7, 5), 'No shapes', ()]
  875. Inputs strides: [(16, 4, 2, 1), (96, 48, 8, 2), (140, 35, 5, 1), 'No strides', ()]
  876. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1be8>, 1.0]
  877.  
  878. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  879. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  880.  
  881. ======================================================================
  882. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 104, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  883. ----------------------------------------------------------------------
  884. Traceback (most recent call last):
  885. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  886. self.test(*self.arg)
  887. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  888. gpuval = f(img, kern)
  889. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  890. storage_map=self.fn.storage_map)
  891. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  892. outputs = self.fn()
  893. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  894. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  895. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac156450>, Scalar(float32)]
  896. Inputs shapes: [(2, 1, 3, 2), (1, 2, 8, 4), (1, 1, 10, 5), 'No shapes', ()]
  897. Inputs strides: [(6, 0, 2, 1), (0, 64, 8, 2), (0, 0, 5, 1), 'No strides', ()]
  898. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433698>, 1.0]
  899.  
  900. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  901. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  902.  
  903. ======================================================================
  904. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 105, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  905. ----------------------------------------------------------------------
  906. Traceback (most recent call last):
  907. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  908. self.test(*self.arg)
  909. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  910. gpuval = f(img, kern)
  911. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  912. storage_map=self.fn.storage_map)
  913. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  914. outputs = self.fn()
  915. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  916. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  917. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcff8c50>, Scalar(float32)]
  918. Inputs shapes: [(1, 1, 3, 2), (3, 1, 8, 4), (3, 1, 10, 5), 'No shapes', ()]
  919. Inputs strides: [(0, 0, 2, 1), (64, 0, 8, 2), (50, 0, 5, 1), 'No strides', ()]
  920. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433a58>, 1.0]
  921.  
  922. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  923. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  924.  
  925. ======================================================================
  926. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 106, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  927. ----------------------------------------------------------------------
  928. Traceback (most recent call last):
  929. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  930. self.test(*self.arg)
  931. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  932. gpuval = f(img, kern)
  933. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  934. storage_map=self.fn.storage_map)
  935. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  936. outputs = self.fn()
  937. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  938. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  939. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d88dc1d0>, Scalar(float32)]
  940. Inputs shapes: [(1, 2, 3, 2), (1, 1, 8, 4), (1, 2, 10, 5), 'No shapes', ()]
  941. Inputs strides: [(0, 6, 2, 1), (0, 0, 8, 2), (0, 50, 5, 1), 'No strides', ()]
  942. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94331c0>, 1.0]
  943.  
  944. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  945. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  946.  
  947. ======================================================================
  948. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 107, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  949. ----------------------------------------------------------------------
  950. Traceback (most recent call last):
  951. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  952. self.test(*self.arg)
  953. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  954. gpuval = f(img, kern)
  955. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  956. storage_map=self.fn.storage_map)
  957. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  958. outputs = self.fn()
  959. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  960. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  961. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af33b290>, Scalar(float32)]
  962. Inputs shapes: [(1, 2, 3, 2), (3, 1, 8, 4), (3, 2, 10, 5), 'No shapes', ()]
  963. Inputs strides: [(0, 6, 2, 1), (64, 0, 8, 2), (100, 50, 5, 1), 'No strides', ()]
  964. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94338f0>, 1.0]
  965.  
  966. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  967. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  968.  
  969. ======================================================================
  970. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 108, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  971. ----------------------------------------------------------------------
  972. Traceback (most recent call last):
  973. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  974. self.test(*self.arg)
  975. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  976. gpuval = f(img, kern)
  977. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  978. storage_map=self.fn.storage_map)
  979. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  980. outputs = self.fn()
  981. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  982. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  983. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d88dc210>, Scalar(float32)]
  984. Inputs shapes: [(2, 1, 3, 2), (3, 2, 8, 4), (3, 1, 10, 5), 'No shapes', ()]
  985. Inputs strides: [(6, 0, 2, 1), (128, 64, 8, 2), (50, 0, 5, 1), 'No strides', ()]
  986. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433d00>, 1.0]
  987.  
  988. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  989. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  990.  
  991. ======================================================================
  992. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 109, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  993. ----------------------------------------------------------------------
  994. Traceback (most recent call last):
  995. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  996. self.test(*self.arg)
  997. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  998. gpuval = f(img, kern)
  999. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1000. storage_map=self.fn.storage_map)
  1001. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1002. outputs = self.fn()
  1003. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1004. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1005. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af4008d0>, Scalar(float32)]
  1006. Inputs shapes: [(2, 2, 3, 2), (1, 2, 8, 4), (1, 2, 10, 5), 'No shapes', ()]
  1007. Inputs strides: [(12, 6, 2, 1), (0, 64, 8, 2), (0, 50, 5, 1), 'No strides', ()]
  1008. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973ab98>, 1.0]
  1009.  
  1010. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1011. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1012.  
  1013. ======================================================================
  1014. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 110, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1015. ----------------------------------------------------------------------
  1016. Traceback (most recent call last):
  1017. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1018. self.test(*self.arg)
  1019. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1020. gpuval = f(img, kern)
  1021. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1022. storage_map=self.fn.storage_map)
  1023. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1024. outputs = self.fn()
  1025. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1026. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1027. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac1563d0>, Scalar(float32)]
  1028. Inputs shapes: [(2, 2, 3, 2), (2, 2, 8, 4), (2, 2, 10, 5), 'No shapes', ()]
  1029. Inputs strides: [(12, 6, 2, 1), (128, 64, 8, 2), (100, 50, 5, 1), 'No strides', ()]
  1030. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a6e8>, 1.0]
  1031.  
  1032. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1033. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1034.  
  1035. ======================================================================
  1036. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 111, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1037. ----------------------------------------------------------------------
  1038. Traceback (most recent call last):
  1039. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1040. self.test(*self.arg)
  1041. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1042. gpuval = f(img, kern)
  1043. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1044. storage_map=self.fn.storage_map)
  1045. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1046. outputs = self.fn()
  1047. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1048. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1049. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e06df410>, Scalar(float32)]
  1050. Inputs shapes: [(2, 4, 3, 2), (3, 2, 8, 4), (3, 4, 10, 5), 'No shapes', ()]
  1051. Inputs strides: [(24, 6, 2, 1), (128, 64, 8, 2), (200, 50, 5, 1), 'No strides', ()]
  1052. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973aaf8>, 1.0]
  1053.  
  1054. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1055. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1056.  
  1057. ======================================================================
  1058. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 112, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1059. ----------------------------------------------------------------------
  1060. Traceback (most recent call last):
  1061. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1062. self.test(*self.arg)
  1063. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1064. gpuval = f(img, kern)
  1065. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1066. storage_map=self.fn.storage_map)
  1067. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1068. outputs = self.fn()
  1069. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1070. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1071. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac59dbd0>, Scalar(float32)]
  1072. Inputs shapes: [(2, 1, 2, 3), (1, 2, 8, 4), (1, 1, 9, 6), 'No shapes', ()]
  1073. Inputs strides: [(6, 0, 3, 1), (0, 64, 8, 2), (0, 0, 6, 1), 'No strides', ()]
  1074. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a4b8>, 1.0]
  1075.  
  1076. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1077. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1078.  
  1079. ======================================================================
  1080. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 113, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1081. ----------------------------------------------------------------------
  1082. Traceback (most recent call last):
  1083. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1084. self.test(*self.arg)
  1085. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1086. gpuval = f(img, kern)
  1087. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1088. storage_map=self.fn.storage_map)
  1089. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1090. outputs = self.fn()
  1091. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1092. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1093. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcfece90>, Scalar(float32)]
  1094. Inputs shapes: [(1, 1, 2, 3), (3, 1, 8, 4), (3, 1, 9, 6), 'No shapes', ()]
  1095. Inputs strides: [(0, 0, 3, 1), (64, 0, 8, 2), (54, 0, 6, 1), 'No strides', ()]
  1096. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a878>, 1.0]
  1097.  
  1098. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1099. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1100.  
  1101. ======================================================================
  1102. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 114, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1103. ----------------------------------------------------------------------
  1104. Traceback (most recent call last):
  1105. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1106. self.test(*self.arg)
  1107. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1108. gpuval = f(img, kern)
  1109. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1110. storage_map=self.fn.storage_map)
  1111. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1112. outputs = self.fn()
  1113. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1114. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1115. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e072a190>, Scalar(float32)]
  1116. Inputs shapes: [(1, 2, 2, 3), (1, 1, 8, 4), (1, 2, 9, 6), 'No shapes', ()]
  1117. Inputs strides: [(0, 6, 3, 1), (0, 0, 8, 2), (0, 54, 6, 1), 'No strides', ()]
  1118. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973ab20>, 1.0]
  1119.  
  1120. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1121. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1122.  
  1123. ======================================================================
  1124. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 115, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1125. ----------------------------------------------------------------------
  1126. Traceback (most recent call last):
  1127. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1128. self.test(*self.arg)
  1129. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1130. gpuval = f(img, kern)
  1131. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1132. storage_map=self.fn.storage_map)
  1133. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1134. outputs = self.fn()
  1135. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1136. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1137. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7578ad0>, Scalar(float32)]
  1138. Inputs shapes: [(1, 2, 2, 3), (3, 1, 8, 4), (3, 2, 9, 6), 'No shapes', ()]
  1139. Inputs strides: [(0, 6, 3, 1), (64, 0, 8, 2), (108, 54, 6, 1), 'No strides', ()]
  1140. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f0a8>, 1.0]
  1141.  
  1142. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1143. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1144.  
  1145. ======================================================================
  1146. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 116, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1147. ----------------------------------------------------------------------
  1148. Traceback (most recent call last):
  1149. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1150. self.test(*self.arg)
  1151. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1152. gpuval = f(img, kern)
  1153. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1154. storage_map=self.fn.storage_map)
  1155. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1156. outputs = self.fn()
  1157. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1158. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1159. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ab699490>, Scalar(float32)]
  1160. Inputs shapes: [(2, 1, 2, 3), (3, 2, 8, 4), (3, 1, 9, 6), 'No shapes', ()]
  1161. Inputs strides: [(6, 0, 3, 1), (128, 64, 8, 2), (54, 0, 6, 1), 'No strides', ()]
  1162. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f828>, 1.0]
  1163.  
  1164. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1165. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1166.  
  1167. ======================================================================
  1168. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 117, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1169. ----------------------------------------------------------------------
  1170. Traceback (most recent call last):
  1171. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1172. self.test(*self.arg)
  1173. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1174. gpuval = f(img, kern)
  1175. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1176. storage_map=self.fn.storage_map)
  1177. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1178. outputs = self.fn()
  1179. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1180. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1181. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ac5ab550>, Scalar(float32)]
  1182. Inputs shapes: [(2, 2, 2, 3), (1, 2, 8, 4), (1, 2, 9, 6), 'No shapes', ()]
  1183. Inputs strides: [(12, 6, 3, 1), (0, 64, 8, 2), (0, 54, 6, 1), 'No strides', ()]
  1184. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fe68>, 1.0]
  1185.  
  1186. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1187. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1188.  
  1189. ======================================================================
  1190. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 118, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1191. ----------------------------------------------------------------------
  1192. Traceback (most recent call last):
  1193. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1194. self.test(*self.arg)
  1195. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1196. gpuval = f(img, kern)
  1197. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1198. storage_map=self.fn.storage_map)
  1199. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1200. outputs = self.fn()
  1201. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1202. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1203. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ddb45210>, Scalar(float32)]
  1204. Inputs shapes: [(2, 2, 2, 3), (2, 2, 8, 4), (2, 2, 9, 6), 'No shapes', ()]
  1205. Inputs strides: [(12, 6, 3, 1), (128, 64, 8, 2), (108, 54, 6, 1), 'No strides', ()]
  1206. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fcb0>, 1.0]
  1207.  
  1208. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1209. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1210.  
  1211. ======================================================================
  1212. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 119, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1213. ----------------------------------------------------------------------
  1214. Traceback (most recent call last):
  1215. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1216. self.test(*self.arg)
  1217. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1218. gpuval = f(img, kern)
  1219. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1220. storage_map=self.fn.storage_map)
  1221. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1222. outputs = self.fn()
  1223. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1224. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1225. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ddb1a8d0>, Scalar(float32)]
  1226. Inputs shapes: [(2, 4, 2, 3), (3, 2, 8, 4), (3, 4, 9, 6), 'No shapes', ()]
  1227. Inputs strides: [(24, 6, 3, 1), (128, 64, 8, 2), (216, 54, 6, 1), 'No strides', ()]
  1228. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f7b0>, 1.0]
  1229.  
  1230. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1231. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1232.  
  1233. ======================================================================
  1234. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 120, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1235. ----------------------------------------------------------------------
  1236. Traceback (most recent call last):
  1237. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1238. self.test(*self.arg)
  1239. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1240. gpuval = f(img, kern)
  1241. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1242. storage_map=self.fn.storage_map)
  1243. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1244. outputs = self.fn()
  1245. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1246. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1247. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, True, False)), CudaNdarrayType(float32, (True, True, True, False)), <theano.gof.type.CDataType object at 0x7f85d3c1c790>, Scalar(float32)]
  1248. Inputs shapes: [(2, 1, 1, 1), (1, 2, 1, 2), (1, 1, 1, 2), 'No shapes', ()]
  1249. Inputs strides: [(1, 0, 0, 0), (0, 4, 0, 1), (0, 0, 0, 1), 'No strides', ()]
  1250. Inputs values: [<CudaNdarray object at 0x7f85c75cc570>, <CudaNdarray object at 0x7f85af12b570>, <CudaNdarray object at 0x7f85dcfec930>, <PyCObject object at 0x7f85c5a5f0f8>, 1.0]
  1251.  
  1252. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1253. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1254.  
  1255. ======================================================================
  1256. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 124, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1257. ----------------------------------------------------------------------
  1258. Traceback (most recent call last):
  1259. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1260. self.test(*self.arg)
  1261. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1262. gpuval = f(img, kern)
  1263. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1264. storage_map=self.fn.storage_map)
  1265. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1266. outputs = self.fn()
  1267. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1268. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1269. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, True, True, False)), <theano.gof.type.CDataType object at 0x7f85ac6f6890>, Scalar(float32)]
  1270. Inputs shapes: [(2, 1, 1, 1), (3, 2, 1, 2), (3, 1, 1, 2), 'No shapes', ()]
  1271. Inputs strides: [(1, 0, 0, 0), (8, 4, 0, 1), (2, 0, 0, 1), 'No strides', ()]
  1272. Inputs values: [<CudaNdarray object at 0x7f85e06952f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3bee0>, 1.0]
  1273.  
  1274. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1275. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1276.  
  1277. ======================================================================
  1278. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 125, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1279. ----------------------------------------------------------------------
  1280. Traceback (most recent call last):
  1281. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1282. self.test(*self.arg)
  1283. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1284. gpuval = f(img, kern)
  1285. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1286. storage_map=self.fn.storage_map)
  1287. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1288. outputs = self.fn()
  1289. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1290. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1291. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, True, False)), CudaNdarrayType(float32, (True, False, True, False)), <theano.gof.type.CDataType object at 0x7f85c59ea810>, Scalar(float32)]
  1292. Inputs shapes: [(2, 2, 1, 1), (1, 2, 1, 2), (1, 2, 1, 2), 'No shapes', ()]
  1293. Inputs strides: [(2, 1, 0, 0), (0, 4, 0, 1), (0, 2, 0, 1), 'No strides', ()]
  1294. Inputs values: [<CudaNdarray object at 0x7f85ac5fe470>, <CudaNdarray object at 0x7f85c9036ef0>, <CudaNdarray object at 0x7f85d88dc1b0>, <PyCObject object at 0x7f85dcd3bfd0>, 1.0]
  1295.  
  1296. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1297. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1298.  
  1299. ======================================================================
  1300. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 126, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1301. ----------------------------------------------------------------------
  1302. Traceback (most recent call last):
  1303. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1304. self.test(*self.arg)
  1305. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1306. gpuval = f(img, kern)
  1307. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1308. storage_map=self.fn.storage_map)
  1309. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1310. outputs = self.fn()
  1311. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1312. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1313. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, False, True, False)), <theano.gof.type.CDataType object at 0x7f85dee0d7d0>, Scalar(float32)]
  1314. Inputs shapes: [(2, 2, 1, 1), (2, 2, 1, 2), (2, 2, 1, 2), 'No shapes', ()]
  1315. Inputs strides: [(2, 1, 0, 0), (8, 4, 0, 1), (4, 2, 0, 1), 'No strides', ()]
  1316. Inputs values: [<CudaNdarray object at 0x7f85a9e55870>, 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b350>, 1.0]
  1317.  
  1318. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1319. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1320.  
  1321. ======================================================================
  1322. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 127, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1323. ----------------------------------------------------------------------
  1324. Traceback (most recent call last):
  1325. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1326. self.test(*self.arg)
  1327. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1328. gpuval = f(img, kern)
  1329. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1330. storage_map=self.fn.storage_map)
  1331. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1332. outputs = self.fn()
  1333. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1334. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1335. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, False, True, False)), <theano.gof.type.CDataType object at 0x7f85a87198d0>, Scalar(float32)]
  1336. Inputs shapes: [(2, 4, 1, 1), (3, 2, 1, 2), (3, 4, 1, 2), 'No shapes', ()]
  1337. Inputs strides: [(4, 1, 0, 0), (8, 4, 0, 1), (8, 2, 0, 1), 'No strides', ()]
  1338. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3bf30>, 1.0]
  1339.  
  1340. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1341. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1342.  
  1343. ======================================================================
  1344. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 128, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1345. ----------------------------------------------------------------------
  1346. Traceback (most recent call last):
  1347. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1348. self.test(*self.arg)
  1349. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1350. gpuval = f(img, kern)
  1351. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1352. storage_map=self.fn.storage_map)
  1353. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1354. outputs = self.fn()
  1355. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1356. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1357. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d88e4050>, Scalar(float32)]
  1358. Inputs shapes: [(2, 1, 2, 2), (1, 2, 2, 4), (1, 1, 3, 5), 'No shapes', ()]
  1359. Inputs strides: [(4, 0, 2, 1), (0, 16, 8, 1), (0, 0, 5, 1), 'No strides', ()]
  1360. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973af30>, 1.0]
  1361.  
  1362. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1363. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1364.  
  1365. ======================================================================
  1366. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 129, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1367. ----------------------------------------------------------------------
  1368. Traceback (most recent call last):
  1369. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1370. self.test(*self.arg)
  1371. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1372. gpuval = f(img, kern)
  1373. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1374. storage_map=self.fn.storage_map)
  1375. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1376. outputs = self.fn()
  1377. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1378. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1379. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c72f2fd0>, Scalar(float32)]
  1380. Inputs shapes: [(1, 1, 2, 2), (3, 1, 2, 4), (3, 1, 3, 5), 'No shapes', ()]
  1381. Inputs strides: [(0, 0, 2, 1), (16, 0, 8, 1), (15, 0, 5, 1), 'No strides', ()]
  1382. Inputs values: [<CudaNdarray object at 0x7f85ddb1a6b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85a973a6c0>, 1.0]
  1383.  
  1384. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1385. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1386.  
  1387. ======================================================================
  1388. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 130, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1389. ----------------------------------------------------------------------
  1390. Traceback (most recent call last):
  1391. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1392. self.test(*self.arg)
  1393. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1394. gpuval = f(img, kern)
  1395. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1396. storage_map=self.fn.storage_map)
  1397. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1398. outputs = self.fn()
  1399. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1400. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1401. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af10d710>, Scalar(float32)]
  1402. Inputs shapes: [(1, 2, 2, 2), (1, 1, 2, 4), (1, 2, 3, 5), 'No shapes', ()]
  1403. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 1), (0, 15, 5, 1), 'No strides', ()]
  1404. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1dc8>, 1.0]
  1405.  
  1406. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1407. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1408.  
  1409. ======================================================================
  1410. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 131, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1411. ----------------------------------------------------------------------
  1412. Traceback (most recent call last):
  1413. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1414. self.test(*self.arg)
  1415. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1416. gpuval = f(img, kern)
  1417. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1418. storage_map=self.fn.storage_map)
  1419. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1420. outputs = self.fn()
  1421. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1422. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1423. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e6c3a3d0>, Scalar(float32)]
  1424. Inputs shapes: [(1, 2, 2, 2), (3, 1, 2, 4), (3, 2, 3, 5), 'No shapes', ()]
  1425. Inputs strides: [(0, 4, 2, 1), (16, 0, 8, 1), (30, 15, 5, 1), 'No strides', ()]
  1426. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1fa8>, 1.0]
  1427.  
  1428. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1429. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1430.  
  1431. ======================================================================
  1432. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 132, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1433. ----------------------------------------------------------------------
  1434. Traceback (most recent call last):
  1435. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1436. self.test(*self.arg)
  1437. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1438. gpuval = f(img, kern)
  1439. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1440. storage_map=self.fn.storage_map)
  1441. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1442. outputs = self.fn()
  1443. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1444. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1445. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c7578710>, Scalar(float32)]
  1446. Inputs shapes: [(2, 1, 2, 2), (3, 2, 2, 4), (3, 1, 3, 5), 'No shapes', ()]
  1447. Inputs strides: [(4, 0, 2, 1), (32, 16, 8, 1), (15, 0, 5, 1), 'No strides', ()]
  1448. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4418>, 1.0]
  1449.  
  1450. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1451. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1452.  
  1453. ======================================================================
  1454. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 133, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1455. ----------------------------------------------------------------------
  1456. Traceback (most recent call last):
  1457. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1458. self.test(*self.arg)
  1459. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1460. gpuval = f(img, kern)
  1461. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1462. storage_map=self.fn.storage_map)
  1463. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1464. outputs = self.fn()
  1465. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1466. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1467. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af40f7d0>, Scalar(float32)]
  1468. Inputs shapes: [(2, 2, 2, 2), (1, 2, 2, 4), (1, 2, 3, 5), 'No shapes', ()]
  1469. Inputs strides: [(8, 4, 2, 1), (0, 16, 8, 1), (0, 15, 5, 1), 'No strides', ()]
  1470. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4580>, 1.0]
  1471.  
  1472. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1473. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1474.  
  1475. ======================================================================
  1476. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 134, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1477. ----------------------------------------------------------------------
  1478. Traceback (most recent call last):
  1479. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1480. self.test(*self.arg)
  1481. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1482. gpuval = f(img, kern)
  1483. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1484. storage_map=self.fn.storage_map)
  1485. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1486. outputs = self.fn()
  1487. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1488. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1489. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7368cd0>, Scalar(float32)]
  1490. Inputs shapes: [(2, 2, 2, 2), (2, 2, 2, 4), (2, 2, 3, 5), 'No shapes', ()]
  1491. Inputs strides: [(8, 4, 2, 1), (32, 16, 8, 1), (30, 15, 5, 1), 'No strides', ()]
  1492. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee12710>, 1.0]
  1493.  
  1494. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1495. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1496.  
  1497. ======================================================================
  1498. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 135, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1499. ----------------------------------------------------------------------
  1500. Traceback (most recent call last):
  1501. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1502. self.test(*self.arg)
  1503. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1504. gpuval = f(img, kern)
  1505. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1506. storage_map=self.fn.storage_map)
  1507. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1508. outputs = self.fn()
  1509. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1510. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1511. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c5a204d0>, Scalar(float32)]
  1512. Inputs shapes: [(2, 4, 2, 2), (3, 2, 2, 4), (3, 4, 3, 5), 'No shapes', ()]
  1513. Inputs strides: [(16, 4, 2, 1), (32, 16, 8, 1), (60, 15, 5, 1), 'No strides', ()]
  1514. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0c60>, 1.0]
  1515.  
  1516. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1517. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1518.  
  1519. ======================================================================
  1520. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 136, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1521. ----------------------------------------------------------------------
  1522. Traceback (most recent call last):
  1523. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1524. self.test(*self.arg)
  1525. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1526. gpuval = f(img, kern)
  1527. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1528. storage_map=self.fn.storage_map)
  1529. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1530. outputs = self.fn()
  1531. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1532. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1533. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c901efd0>, Scalar(float32)]
  1534. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 6), (1, 1, 4, 7), 'No shapes', ()]
  1535. Inputs strides: [(4, 0, 2, 1), (0, 36, 12, 1), (0, 0, 7, 1), 'No strides', ()]
  1536. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0670>, 1.0]
  1537.  
  1538. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1539. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1540.  
  1541. ======================================================================
  1542. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 137, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1543. ----------------------------------------------------------------------
  1544. Traceback (most recent call last):
  1545. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1546. self.test(*self.arg)
  1547. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1548. gpuval = f(img, kern)
  1549. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1550. storage_map=self.fn.storage_map)
  1551. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1552. outputs = self.fn()
  1553. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1554. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1555. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e072afd0>, Scalar(float32)]
  1556. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 6), (3, 1, 4, 7), 'No shapes', ()]
  1557. Inputs strides: [(0, 0, 2, 1), (36, 0, 12, 1), (28, 0, 7, 1), 'No strides', ()]
  1558. Inputs values: [<CudaNdarray object at 0x7f85e7e0cf70>, 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0df0>, 1.0]
  1559.  
  1560. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1561. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1562.  
  1563. ======================================================================
  1564. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 138, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1565. ----------------------------------------------------------------------
  1566. Traceback (most recent call last):
  1567. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1568. self.test(*self.arg)
  1569. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1570. gpuval = f(img, kern)
  1571. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1572. storage_map=self.fn.storage_map)
  1573. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1574. outputs = self.fn()
  1575. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1576. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1577. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c7497e10>, Scalar(float32)]
  1578. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 6), (1, 2, 4, 7), 'No shapes', ()]
  1579. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 1), (0, 28, 7, 1), 'No strides', ()]
  1580. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0288>, 1.0]
  1581.  
  1582. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1583. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1584.  
  1585. ======================================================================
  1586. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 139, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1587. ----------------------------------------------------------------------
  1588. Traceback (most recent call last):
  1589. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1590. self.test(*self.arg)
  1591. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1592. gpuval = f(img, kern)
  1593. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1594. storage_map=self.fn.storage_map)
  1595. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1596. outputs = self.fn()
  1597. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1598. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1599. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e7e1ff10>, Scalar(float32)]
  1600. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 6), (3, 2, 4, 7), 'No shapes', ()]
  1601. Inputs strides: [(0, 4, 2, 1), (36, 0, 12, 1), (56, 28, 7, 1), 'No strides', ()]
  1602. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0490>, 1.0]
  1603.  
  1604. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1605. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1606.  
  1607. ======================================================================
  1608. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 140, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1609. ----------------------------------------------------------------------
  1610. Traceback (most recent call last):
  1611. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1612. self.test(*self.arg)
  1613. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1614. gpuval = f(img, kern)
  1615. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1616. storage_map=self.fn.storage_map)
  1617. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1618. outputs = self.fn()
  1619. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1620. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1621. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac13e910>, Scalar(float32)]
  1622. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 6), (3, 1, 4, 7), 'No shapes', ()]
  1623. Inputs strides: [(4, 0, 2, 1), (72, 36, 12, 1), (28, 0, 7, 1), 'No strides', ()]
  1624. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a96143a0>, 1.0]
  1625.  
  1626. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1627. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1628.  
  1629. ======================================================================
  1630. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 141, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1631. ----------------------------------------------------------------------
  1632. Traceback (most recent call last):
  1633. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1634. self.test(*self.arg)
  1635. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1636. gpuval = f(img, kern)
  1637. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1638. storage_map=self.fn.storage_map)
  1639. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1640. outputs = self.fn()
  1641. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1642. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1643. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dc2b29d0>, Scalar(float32)]
  1644. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 6), (1, 2, 4, 7), 'No shapes', ()]
  1645. Inputs strides: [(8, 4, 2, 1), (0, 36, 12, 1), (0, 28, 7, 1), 'No strides', ()]
  1646. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614800>, 1.0]
  1647.  
  1648. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1649. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1650.  
  1651. ======================================================================
  1652. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 142, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1653. ----------------------------------------------------------------------
  1654. Traceback (most recent call last):
  1655. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1656. self.test(*self.arg)
  1657. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1658. gpuval = f(img, kern)
  1659. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1660. storage_map=self.fn.storage_map)
  1661. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1662. outputs = self.fn()
  1663. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1664. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1665. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dc2b2610>, Scalar(float32)]
  1666. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 6), (2, 2, 4, 7), 'No shapes', ()]
  1667. Inputs strides: [(8, 4, 2, 1), (72, 36, 12, 1), (56, 28, 7, 1), 'No strides', ()]
  1668. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614b70>, 1.0]
  1669.  
  1670. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1671. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1672.  
  1673. ======================================================================
  1674. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 143, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1675. ----------------------------------------------------------------------
  1676. Traceback (most recent call last):
  1677. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1678. self.test(*self.arg)
  1679. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1680. gpuval = f(img, kern)
  1681. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1682. storage_map=self.fn.storage_map)
  1683. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1684. outputs = self.fn()
  1685. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1686. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1687. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7327d10>, Scalar(float32)]
  1688. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 6), (3, 4, 4, 7), 'No shapes', ()]
  1689. Inputs strides: [(16, 4, 2, 1), (72, 36, 12, 1), (112, 28, 7, 1), 'No strides', ()]
  1690. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fcd8>, 1.0]
  1691.  
  1692. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1693. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1694.  
  1695. ======================================================================
  1696. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 144, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1697. ----------------------------------------------------------------------
  1698. Traceback (most recent call last):
  1699. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1700. self.test(*self.arg)
  1701. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1702. gpuval = f(img, kern)
  1703. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1704. storage_map=self.fn.storage_map)
  1705. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1706. outputs = self.fn()
  1707. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1708. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1709. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddac4a50>, Scalar(float32)]
  1710. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 6), (1, 1, 5, 7), 'No shapes', ()]
  1711. Inputs strides: [(4, 0, 2, 1), (0, 48, 12, 1), (0, 0, 7, 1), 'No strides', ()]
  1712. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f468>, 1.0]
  1713.  
  1714. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1715. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1716.  
  1717. ======================================================================
  1718. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 145, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1719. ----------------------------------------------------------------------
  1720. Traceback (most recent call last):
  1721. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1722. self.test(*self.arg)
  1723. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1724. gpuval = f(img, kern)
  1725. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1726. storage_map=self.fn.storage_map)
  1727. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1728. outputs = self.fn()
  1729. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1730. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1731. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e7cd29d0>, Scalar(float32)]
  1732. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 6), (3, 1, 5, 7), 'No shapes', ()]
  1733. Inputs strides: [(0, 0, 2, 1), (48, 0, 12, 1), (35, 0, 7, 1), 'No strides', ()]
  1734. Inputs values: [<CudaNdarray object at 0x7f85abd28df0>, 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f378>, 1.0]
  1735.  
  1736. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1737. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1738.  
  1739. ======================================================================
  1740. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 146, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1741. ----------------------------------------------------------------------
  1742. Traceback (most recent call last):
  1743. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1744. self.test(*self.arg)
  1745. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1746. gpuval = f(img, kern)
  1747. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1748. storage_map=self.fn.storage_map)
  1749. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1750. outputs = self.fn()
  1751. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1752. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1753. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c743a090>, Scalar(float32)]
  1754. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 6), (1, 2, 5, 7), 'No shapes', ()]
  1755. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 1), (0, 35, 7, 1), 'No strides', ()]
  1756. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f490>, 1.0]
  1757.  
  1758. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1759. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1760.  
  1761. ======================================================================
  1762. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 147, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1763. ----------------------------------------------------------------------
  1764. Traceback (most recent call last):
  1765. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1766. self.test(*self.arg)
  1767. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1768. gpuval = f(img, kern)
  1769. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1770. storage_map=self.fn.storage_map)
  1771. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1772. outputs = self.fn()
  1773. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1774. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1775. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c5a39650>, Scalar(float32)]
  1776. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 6), (3, 2, 5, 7), 'No shapes', ()]
  1777. Inputs strides: [(0, 4, 2, 1), (48, 0, 12, 1), (70, 35, 7, 1), 'No strides', ()]
  1778. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433c60>, 1.0]
  1779.  
  1780. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1781. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1782.  
  1783. ======================================================================
  1784. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 148, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1785. ----------------------------------------------------------------------
  1786. Traceback (most recent call last):
  1787. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1788. self.test(*self.arg)
  1789. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1790. gpuval = f(img, kern)
  1791. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1792. storage_map=self.fn.storage_map)
  1793. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1794. outputs = self.fn()
  1795. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1796. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1797. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85df38ab10>, Scalar(float32)]
  1798. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 6), (3, 1, 5, 7), 'No shapes', ()]
  1799. Inputs strides: [(4, 0, 2, 1), (96, 48, 12, 1), (35, 0, 7, 1), 'No strides', ()]
  1800. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433aa8>, 1.0]
  1801.  
  1802. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1803. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1804.  
  1805. ======================================================================
  1806. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 149, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1807. ----------------------------------------------------------------------
  1808. Traceback (most recent call last):
  1809. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1810. self.test(*self.arg)
  1811. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1812. gpuval = f(img, kern)
  1813. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1814. storage_map=self.fn.storage_map)
  1815. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1816. outputs = self.fn()
  1817. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1818. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1819. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c9039310>, Scalar(float32)]
  1820. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 6), (1, 2, 5, 7), 'No shapes', ()]
  1821. Inputs strides: [(8, 4, 2, 1), (0, 48, 12, 1), (0, 35, 7, 1), 'No strides', ()]
  1822. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433440>, 1.0]
  1823.  
  1824. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1825. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1826.  
  1827. ======================================================================
  1828. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 150, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1829. ----------------------------------------------------------------------
  1830. Traceback (most recent call last):
  1831. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1832. self.test(*self.arg)
  1833. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1834. gpuval = f(img, kern)
  1835. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1836. storage_map=self.fn.storage_map)
  1837. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1838. outputs = self.fn()
  1839. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1840. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1841. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dcd4d7d0>, Scalar(float32)]
  1842. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 6), (2, 2, 5, 7), 'No shapes', ()]
  1843. Inputs strides: [(8, 4, 2, 1), (96, 48, 12, 1), (70, 35, 7, 1), 'No strides', ()]
  1844. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433fa8>, 1.0]
  1845.  
  1846. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1847. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1848.  
  1849. ======================================================================
  1850. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 151, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1851. ----------------------------------------------------------------------
  1852. Traceback (most recent call last):
  1853. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1854. self.test(*self.arg)
  1855. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1856. gpuval = f(img, kern)
  1857. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1858. storage_map=self.fn.storage_map)
  1859. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1860. outputs = self.fn()
  1861. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1862. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1863. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85aa16be10>, Scalar(float32)]
  1864. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 6), (3, 4, 5, 7), 'No shapes', ()]
  1865. Inputs strides: [(16, 4, 2, 1), (96, 48, 12, 1), (140, 35, 7, 1), 'No strides', ()]
  1866. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b8a0>, 1.0]
  1867.  
  1868. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1869. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1870.  
  1871. ======================================================================
  1872. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 152, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1873. ----------------------------------------------------------------------
  1874. Traceback (most recent call last):
  1875. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1876. self.test(*self.arg)
  1877. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1878. gpuval = f(img, kern)
  1879. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1880. storage_map=self.fn.storage_map)
  1881. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1882. outputs = self.fn()
  1883. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1884. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1885. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac5ab750>, Scalar(float32)]
  1886. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 8), (1, 1, 4, 9), 'No shapes', ()]
  1887. Inputs strides: [(4, 0, 2, 1), (0, 48, 16, 1), (0, 0, 9, 1), 'No strides', ()]
  1888. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b8c8>, 1.0]
  1889.  
  1890. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1891. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1892.  
  1893. ======================================================================
  1894. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 153, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1895. ----------------------------------------------------------------------
  1896. Traceback (most recent call last):
  1897. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1898. self.test(*self.arg)
  1899. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1900. gpuval = f(img, kern)
  1901. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1902. storage_map=self.fn.storage_map)
  1903. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1904. outputs = self.fn()
  1905. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1906. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1907. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af28b590>, Scalar(float32)]
  1908. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 8), (3, 1, 4, 9), 'No shapes', ()]
  1909. Inputs strides: [(0, 0, 2, 1), (48, 0, 16, 1), (36, 0, 9, 1), 'No strides', ()]
  1910. Inputs values: [<CudaNdarray object at 0x7f85d37bf3f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b9e0>, 1.0]
  1911.  
  1912. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1913. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1914.  
  1915. ======================================================================
  1916. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 154, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1917. ----------------------------------------------------------------------
  1918. Traceback (most recent call last):
  1919. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1920. self.test(*self.arg)
  1921. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1922. gpuval = f(img, kern)
  1923. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1924. storage_map=self.fn.storage_map)
  1925. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1926. outputs = self.fn()
  1927. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1928. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1929. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85abd28510>, Scalar(float32)]
  1930. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 8), (1, 2, 4, 9), 'No shapes', ()]
  1931. Inputs strides: [(0, 4, 2, 1), (0, 0, 16, 1), (0, 36, 9, 1), 'No strides', ()]
  1932. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b4e0>, 1.0]
  1933.  
  1934. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1935. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1936.  
  1937. ======================================================================
  1938. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 155, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1939. ----------------------------------------------------------------------
  1940. Traceback (most recent call last):
  1941. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1942. self.test(*self.arg)
  1943. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1944. gpuval = f(img, kern)
  1945. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1946. storage_map=self.fn.storage_map)
  1947. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1948. outputs = self.fn()
  1949. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1950. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1951. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e7277450>, Scalar(float32)]
  1952. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 8), (3, 2, 4, 9), 'No shapes', ()]
  1953. Inputs strides: [(0, 4, 2, 1), (48, 0, 16, 1), (72, 36, 9, 1), 'No strides', ()]
  1954. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e0a37058>, 1.0]
  1955.  
  1956. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1957. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1958.  
  1959. ======================================================================
  1960. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 156, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1961. ----------------------------------------------------------------------
  1962. Traceback (most recent call last):
  1963. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1964. self.test(*self.arg)
  1965. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1966. gpuval = f(img, kern)
  1967. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1968. storage_map=self.fn.storage_map)
  1969. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1970. outputs = self.fn()
  1971. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1972. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1973. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af3f4b10>, Scalar(float32)]
  1974. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 8), (3, 1, 4, 9), 'No shapes', ()]
  1975. Inputs strides: [(4, 0, 2, 1), (96, 48, 16, 1), (36, 0, 9, 1), 'No strides', ()]
  1976. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9e2d8>, 1.0]
  1977.  
  1978. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  1979. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  1980.  
  1981. ======================================================================
  1982. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 157, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  1983. ----------------------------------------------------------------------
  1984. Traceback (most recent call last):
  1985. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  1986. self.test(*self.arg)
  1987. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  1988. gpuval = f(img, kern)
  1989. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  1990. storage_map=self.fn.storage_map)
  1991. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  1992. outputs = self.fn()
  1993. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  1994. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  1995. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d3c31690>, Scalar(float32)]
  1996. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 8), (1, 2, 4, 9), 'No shapes', ()]
  1997. Inputs strides: [(8, 4, 2, 1), (0, 48, 16, 1), (0, 36, 9, 1), 'No strides', ()]
  1998. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9e558>, 1.0]
  1999.  
  2000. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2001. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2002.  
  2003. ======================================================================
  2004. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 158, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2005. ----------------------------------------------------------------------
  2006. Traceback (most recent call last):
  2007. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2008. self.test(*self.arg)
  2009. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2010. gpuval = f(img, kern)
  2011. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2012. storage_map=self.fn.storage_map)
  2013. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2014. outputs = self.fn()
  2015. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2016. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2017. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dee983d0>, Scalar(float32)]
  2018. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 8), (2, 2, 4, 9), 'No shapes', ()]
  2019. Inputs strides: [(8, 4, 2, 1), (96, 48, 16, 1), (72, 36, 9, 1), 'No strides', ()]
  2020. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9e800>, 1.0]
  2021.  
  2022. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2023. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2024.  
  2025. ======================================================================
  2026. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 159, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2027. ----------------------------------------------------------------------
  2028. Traceback (most recent call last):
  2029. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2030. self.test(*self.arg)
  2031. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2032. gpuval = f(img, kern)
  2033. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2034. storage_map=self.fn.storage_map)
  2035. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2036. outputs = self.fn()
  2037. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2038. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2039. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af22ce50>, Scalar(float32)]
  2040. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 8), (3, 4, 4, 9), 'No shapes', ()]
  2041. Inputs strides: [(16, 4, 2, 1), (96, 48, 16, 1), (144, 36, 9, 1), 'No strides', ()]
  2042. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9ea80>, 1.0]
  2043.  
  2044. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2045. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2046.  
  2047. ======================================================================
  2048. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 160, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2049. ----------------------------------------------------------------------
  2050. Traceback (most recent call last):
  2051. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2052. self.test(*self.arg)
  2053. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2054. gpuval = f(img, kern)
  2055. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2056. storage_map=self.fn.storage_map)
  2057. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2058. outputs = self.fn()
  2059. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2060. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2061. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c73b4d90>, Scalar(float32)]
  2062. Inputs shapes: [(2, 1, 3, 2), (1, 2, 4, 8), (1, 1, 6, 9), 'No shapes', ()]
  2063. Inputs strides: [(6, 0, 2, 1), (0, 64, 16, 1), (0, 0, 9, 1), 'No strides', ()]
  2064. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9ecb0>, 1.0]
  2065.  
  2066. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2067. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2068.  
  2069. ======================================================================
  2070. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 161, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2071. ----------------------------------------------------------------------
  2072. Traceback (most recent call last):
  2073. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2074. self.test(*self.arg)
  2075. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2076. gpuval = f(img, kern)
  2077. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2078. storage_map=self.fn.storage_map)
  2079. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2080. outputs = self.fn()
  2081. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2082. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2083. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85aa144ad0>, Scalar(float32)]
  2084. Inputs shapes: [(1, 1, 3, 2), (3, 1, 4, 8), (3, 1, 6, 9), 'No shapes', ()]
  2085. Inputs strides: [(0, 0, 2, 1), (64, 0, 16, 1), (54, 0, 9, 1), 'No strides', ()]
  2086. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9ef30>, 1.0]
  2087.  
  2088. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2089. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2090.  
  2091. ======================================================================
  2092. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 162, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2093. ----------------------------------------------------------------------
  2094. Traceback (most recent call last):
  2095. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2096. self.test(*self.arg)
  2097. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2098. gpuval = f(img, kern)
  2099. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2100. storage_map=self.fn.storage_map)
  2101. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2102. outputs = self.fn()
  2103. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2104. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2105. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85abccab50>, Scalar(float32)]
  2106. Inputs shapes: [(1, 2, 3, 2), (1, 1, 4, 8), (1, 2, 6, 9), 'No shapes', ()]
  2107. Inputs strides: [(0, 6, 2, 1), (0, 0, 16, 1), (0, 54, 9, 1), 'No strides', ()]
  2108. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e07f41e8>, 1.0]
  2109.  
  2110. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2111. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2112.  
  2113. ======================================================================
  2114. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 163, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2115. ----------------------------------------------------------------------
  2116. Traceback (most recent call last):
  2117. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2118. self.test(*self.arg)
  2119. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2120. gpuval = f(img, kern)
  2121. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2122. storage_map=self.fn.storage_map)
  2123. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2124. outputs = self.fn()
  2125. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2126. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2127. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85a972f9d0>, Scalar(float32)]
  2128. Inputs shapes: [(1, 2, 3, 2), (3, 1, 4, 8), (3, 2, 6, 9), 'No shapes', ()]
  2129. Inputs strides: [(0, 6, 2, 1), (64, 0, 16, 1), (108, 54, 9, 1), 'No strides', ()]
  2130. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572490>, 1.0]
  2131.  
  2132. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2133. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2134.  
  2135. ======================================================================
  2136. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 164, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2137. ----------------------------------------------------------------------
  2138. Traceback (most recent call last):
  2139. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2140. self.test(*self.arg)
  2141. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2142. gpuval = f(img, kern)
  2143. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2144. storage_map=self.fn.storage_map)
  2145. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2146. outputs = self.fn()
  2147. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2148. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2149. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c90392d0>, Scalar(float32)]
  2150. Inputs shapes: [(2, 1, 3, 2), (3, 2, 4, 8), (3, 1, 6, 9), 'No shapes', ()]
  2151. Inputs strides: [(6, 0, 2, 1), (128, 64, 16, 1), (54, 0, 9, 1), 'No strides', ()]
  2152. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572710>, 1.0]
  2153.  
  2154. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2155. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2156.  
  2157. ======================================================================
  2158. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 165, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2159. ----------------------------------------------------------------------
  2160. Traceback (most recent call last):
  2161. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2162. self.test(*self.arg)
  2163. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2164. gpuval = f(img, kern)
  2165. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2166. storage_map=self.fn.storage_map)
  2167. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2168. outputs = self.fn()
  2169. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2170. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2171. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ab6a5750>, Scalar(float32)]
  2172. Inputs shapes: [(2, 2, 3, 2), (1, 2, 4, 8), (1, 2, 6, 9), 'No shapes', ()]
  2173. Inputs strides: [(12, 6, 2, 1), (0, 64, 16, 1), (0, 54, 9, 1), 'No strides', ()]
  2174. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572990>, 1.0]
  2175.  
  2176. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2177. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2178.  
  2179. ======================================================================
  2180. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 166, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2181. ----------------------------------------------------------------------
  2182. Traceback (most recent call last):
  2183. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2184. self.test(*self.arg)
  2185. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2186. gpuval = f(img, kern)
  2187. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2188. storage_map=self.fn.storage_map)
  2189. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2190. outputs = self.fn()
  2191. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2192. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2193. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c74fe1d0>, Scalar(float32)]
  2194. Inputs shapes: [(2, 2, 3, 2), (2, 2, 4, 8), (2, 2, 6, 9), 'No shapes', ()]
  2195. Inputs strides: [(12, 6, 2, 1), (128, 64, 16, 1), (108, 54, 9, 1), 'No strides', ()]
  2196. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572c38>, 1.0]
  2197.  
  2198. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2199. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2200.  
  2201. ======================================================================
  2202. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 167, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2203. ----------------------------------------------------------------------
  2204. Traceback (most recent call last):
  2205. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2206. self.test(*self.arg)
  2207. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2208. gpuval = f(img, kern)
  2209. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2210. storage_map=self.fn.storage_map)
  2211. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2212. outputs = self.fn()
  2213. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2214. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2215. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d940d950>, Scalar(float32)]
  2216. Inputs shapes: [(2, 4, 3, 2), (3, 2, 4, 8), (3, 4, 6, 9), 'No shapes', ()]
  2217. Inputs strides: [(24, 6, 2, 1), (128, 64, 16, 1), (216, 54, 9, 1), 'No strides', ()]
  2218. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572eb8>, 1.0]
  2219.  
  2220. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2221. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2222.  
  2223. ======================================================================
  2224. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 168, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2225. ----------------------------------------------------------------------
  2226. Traceback (most recent call last):
  2227. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2228. self.test(*self.arg)
  2229. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2230. gpuval = f(img, kern)
  2231. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2232. storage_map=self.fn.storage_map)
  2233. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2234. outputs = self.fn()
  2235. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2236. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2237. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddb6f6d0>, Scalar(float32)]
  2238. Inputs shapes: [(2, 1, 2, 3), (1, 2, 4, 8), (1, 1, 5, 10), 'No shapes', ()]
  2239. Inputs strides: [(6, 0, 3, 1), (0, 64, 16, 1), (0, 0, 10, 1), 'No strides', ()]
  2240. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85df607120>, 1.0]
  2241.  
  2242. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2243. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2244.  
  2245. ======================================================================
  2246. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 169, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2247. ----------------------------------------------------------------------
  2248. Traceback (most recent call last):
  2249. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2250. self.test(*self.arg)
  2251. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2252. gpuval = f(img, kern)
  2253. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2254. storage_map=self.fn.storage_map)
  2255. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2256. outputs = self.fn()
  2257. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2258. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2259. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c748be50>, Scalar(float32)]
  2260. Inputs shapes: [(1, 1, 2, 3), (3, 1, 4, 8), (3, 1, 5, 10), 'No shapes', ()]
  2261. Inputs strides: [(0, 0, 3, 1), (64, 0, 16, 1), (50, 0, 10, 1), 'No strides', ()]
  2262. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5c3a0>, 1.0]
  2263.  
  2264. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2265. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2266.  
  2267. ======================================================================
  2268. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 170, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2269. ----------------------------------------------------------------------
  2270. Traceback (most recent call last):
  2271. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2272. self.test(*self.arg)
  2273. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2274. gpuval = f(img, kern)
  2275. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2276. storage_map=self.fn.storage_map)
  2277. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2278. outputs = self.fn()
  2279. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2280. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2281. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c748b910>, Scalar(float32)]
  2282. Inputs shapes: [(1, 2, 2, 3), (1, 1, 4, 8), (1, 2, 5, 10), 'No shapes', ()]
  2283. Inputs strides: [(0, 6, 3, 1), (0, 0, 16, 1), (0, 50, 10, 1), 'No strides', ()]
  2284. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5c620>, 1.0]
  2285.  
  2286. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2287. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2288.  
  2289. ======================================================================
  2290. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 171, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2291. ----------------------------------------------------------------------
  2292. Traceback (most recent call last):
  2293. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2294. self.test(*self.arg)
  2295. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2296. gpuval = f(img, kern)
  2297. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2298. storage_map=self.fn.storage_map)
  2299. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2300. outputs = self.fn()
  2301. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2302. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2303. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c5a38150>, Scalar(float32)]
  2304. Inputs shapes: [(1, 2, 2, 3), (3, 1, 4, 8), (3, 2, 5, 10), 'No shapes', ()]
  2305. Inputs strides: [(0, 6, 3, 1), (64, 0, 16, 1), (100, 50, 10, 1), 'No strides', ()]
  2306. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5c8c8>, 1.0]
  2307.  
  2308. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2309. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2310.  
  2311. ======================================================================
  2312. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 172, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2313. ----------------------------------------------------------------------
  2314. Traceback (most recent call last):
  2315. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2316. self.test(*self.arg)
  2317. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2318. gpuval = f(img, kern)
  2319. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2320. storage_map=self.fn.storage_map)
  2321. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2322. outputs = self.fn()
  2323. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2324. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2325. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c7465d10>, Scalar(float32)]
  2326. Inputs shapes: [(2, 1, 2, 3), (3, 2, 4, 8), (3, 1, 5, 10), 'No shapes', ()]
  2327. Inputs strides: [(6, 0, 3, 1), (128, 64, 16, 1), (50, 0, 10, 1), 'No strides', ()]
  2328. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5cb48>, 1.0]
  2329.  
  2330. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2331. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2332.  
  2333. ======================================================================
  2334. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 173, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2335. ----------------------------------------------------------------------
  2336. Traceback (most recent call last):
  2337. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2338. self.test(*self.arg)
  2339. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2340. gpuval = f(img, kern)
  2341. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2342. storage_map=self.fn.storage_map)
  2343. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2344. outputs = self.fn()
  2345. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2346. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2347. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c59ea090>, Scalar(float32)]
  2348. Inputs shapes: [(2, 2, 2, 3), (1, 2, 4, 8), (1, 2, 5, 10), 'No shapes', ()]
  2349. Inputs strides: [(12, 6, 3, 1), (0, 64, 16, 1), (0, 50, 10, 1), 'No strides', ()]
  2350. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5cdc8>, 1.0]
  2351.  
  2352. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2353. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2354.  
  2355. ======================================================================
  2356. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 174, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2357. ----------------------------------------------------------------------
  2358. Traceback (most recent call last):
  2359. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2360. self.test(*self.arg)
  2361. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2362. gpuval = f(img, kern)
  2363. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2364. storage_map=self.fn.storage_map)
  2365. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2366. outputs = self.fn()
  2367. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2368. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2369. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac5fe290>, Scalar(float32)]
  2370. Inputs shapes: [(2, 2, 2, 3), (2, 2, 4, 8), (2, 2, 5, 10), 'No shapes', ()]
  2371. Inputs strides: [(12, 6, 3, 1), (128, 64, 16, 1), (100, 50, 10, 1), 'No strides', ()]
  2372. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85af1230a8>, 1.0]
  2373.  
  2374. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2375. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2376.  
  2377. ======================================================================
  2378. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 175, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2379. ----------------------------------------------------------------------
  2380. Traceback (most recent call last):
  2381. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2382. self.test(*self.arg)
  2383. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2384. gpuval = f(img, kern)
  2385. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2386. storage_map=self.fn.storage_map)
  2387. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2388. outputs = self.fn()
  2389. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2390. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2391. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7344c90>, Scalar(float32)]
  2392. Inputs shapes: [(2, 4, 2, 3), (3, 2, 4, 8), (3, 4, 5, 10), 'No shapes', ()]
  2393. Inputs strides: [(24, 6, 3, 1), (128, 64, 16, 1), (200, 50, 10, 1), 'No strides', ()]
  2394. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c7479328>, 1.0]
  2395.  
  2396. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2397. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2398.  
  2399. ======================================================================
  2400. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 176, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2401. ----------------------------------------------------------------------
  2402. Traceback (most recent call last):
  2403. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2404. self.test(*self.arg)
  2405. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2406. gpuval = f(img, kern)
  2407. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2408. storage_map=self.fn.storage_map)
  2409. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2410. outputs = self.fn()
  2411. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2412. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2413. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, True, True, True)), <theano.gof.type.CDataType object at 0x7f85c7344c90>, Scalar(float32)]
  2414. Inputs shapes: [(2, 1, 1, 1), (1, 2, 1, 1), (1, 1, 1, 1), 'No shapes', ()]
  2415. Inputs strides: [(1, 0, 0, 0), (0, 4, 0, 0), (0, 0, 0, 0), 'No strides', ()]
  2416. Inputs values: [<CudaNdarray object at 0x7f85d3c1caf0>, <CudaNdarray object at 0x7f85a93dc9b0>, <CudaNdarray object at 0x7f85d3c1c630>, <PyCObject object at 0x7f85c74791c0>, 1.0]
  2417.  
  2418. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2419. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2420.  
  2421. ======================================================================
  2422. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 180, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2423. ----------------------------------------------------------------------
  2424. Traceback (most recent call last):
  2425. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2426. self.test(*self.arg)
  2427. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2428. gpuval = f(img, kern)
  2429. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2430. storage_map=self.fn.storage_map)
  2431. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2432. outputs = self.fn()
  2433. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2434. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2435. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, True, True, True)), <theano.gof.type.CDataType object at 0x7f85ddad1fd0>, Scalar(float32)]
  2436. Inputs shapes: [(2, 1, 1, 1), (3, 2, 1, 1), (3, 1, 1, 1), 'No shapes', ()]
  2437. Inputs strides: [(1, 0, 0, 0), (8, 4, 0, 0), (1, 0, 0, 0), 'No strides', ()]
  2438. Inputs values: [<CudaNdarray object at 0x7f85af3779b0>, 'not shown', <CudaNdarray object at 0x7f85af377170>, <PyCObject object at 0x7f85e6c5ce40>, 1.0]
  2439.  
  2440. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2441. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2442.  
  2443. ======================================================================
  2444. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 181, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2445. ----------------------------------------------------------------------
  2446. Traceback (most recent call last):
  2447. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2448. self.test(*self.arg)
  2449. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2450. gpuval = f(img, kern)
  2451. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2452. storage_map=self.fn.storage_map)
  2453. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2454. outputs = self.fn()
  2455. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2456. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2457. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, False, True, True)), <theano.gof.type.CDataType object at 0x7f85dee0d990>, Scalar(float32)]
  2458. Inputs shapes: [(2, 2, 1, 1), (1, 2, 1, 1), (1, 2, 1, 1), 'No shapes', ()]
  2459. Inputs strides: [(2, 1, 0, 0), (0, 4, 0, 0), (0, 1, 0, 0), 'No strides', ()]
  2460. Inputs values: [<CudaNdarray object at 0x7f85af377d30>, <CudaNdarray object at 0x7f85c748b4b0>, <CudaNdarray object at 0x7f85af377a70>, <PyCObject object at 0x7f85e6c5ceb8>, 1.0]
  2461.  
  2462. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2463. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2464.  
  2465. ======================================================================
  2466. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 182, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2467. ----------------------------------------------------------------------
  2468. Traceback (most recent call last):
  2469. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2470. self.test(*self.arg)
  2471. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2472. gpuval = f(img, kern)
  2473. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2474. storage_map=self.fn.storage_map)
  2475. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2476. outputs = self.fn()
  2477. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2478. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2479. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), <theano.gof.type.CDataType object at 0x7f85af40dad0>, Scalar(float32)]
  2480. Inputs shapes: [(2, 2, 1, 1), (2, 2, 1, 1), (2, 2, 1, 1), 'No shapes', ()]
  2481. Inputs strides: [(2, 1, 0, 0), (8, 4, 0, 0), (2, 1, 0, 0), 'No strides', ()]
  2482. Inputs values: [<CudaNdarray object at 0x7f85af40d8f0>, <CudaNdarray object at 0x7f85deba9230>, <CudaNdarray object at 0x7f85af40dcf0>, <PyCObject object at 0x7f85e6c5cda0>, 1.0]
  2483.  
  2484. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2485. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2486.  
  2487. ======================================================================
  2488. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 183, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2489. ----------------------------------------------------------------------
  2490. Traceback (most recent call last):
  2491. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2492. self.test(*self.arg)
  2493. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2494. gpuval = f(img, kern)
  2495. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2496. storage_map=self.fn.storage_map)
  2497. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2498. outputs = self.fn()
  2499. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2500. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2501. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), <theano.gof.type.CDataType object at 0x7f85abcca490>, Scalar(float32)]
  2502. Inputs shapes: [(2, 4, 1, 1), (3, 2, 1, 1), (3, 4, 1, 1), 'No shapes', ()]
  2503. Inputs strides: [(4, 1, 0, 0), (8, 4, 0, 0), (4, 1, 0, 0), 'No strides', ()]
  2504. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5ce40>, 1.0]
  2505.  
  2506. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2507. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2508.  
  2509. ======================================================================
  2510. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 184, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2511. ----------------------------------------------------------------------
  2512. Traceback (most recent call last):
  2513. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2514. self.test(*self.arg)
  2515. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2516. gpuval = f(img, kern)
  2517. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2518. storage_map=self.fn.storage_map)
  2519. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2520. outputs = self.fn()
  2521. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2522. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2523. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af377090>, Scalar(float32)]
  2524. Inputs shapes: [(2, 1, 2, 2), (1, 2, 2, 2), (1, 1, 3, 3), 'No shapes', ()]
  2525. Inputs strides: [(4, 0, 2, 1), (0, 16, 8, 2), (0, 0, 3, 1), 'No strides', ()]
  2526. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5c5a8>, 1.0]
  2527.  
  2528. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2529. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2530.  
  2531. ======================================================================
  2532. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 185, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2533. ----------------------------------------------------------------------
  2534. Traceback (most recent call last):
  2535. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2536. self.test(*self.arg)
  2537. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2538. gpuval = f(img, kern)
  2539. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2540. storage_map=self.fn.storage_map)
  2541. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2542. outputs = self.fn()
  2543. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2544. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2545. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dbddca50>, Scalar(float32)]
  2546. Inputs shapes: [(1, 1, 2, 2), (3, 1, 2, 2), (3, 1, 3, 3), 'No shapes', ()]
  2547. Inputs strides: [(0, 0, 2, 1), (16, 0, 8, 2), (9, 0, 3, 1), 'No strides', ()]
  2548. Inputs values: [<CudaNdarray object at 0x7f85dee24e30>, 'not shown', 'not shown', <PyCObject object at 0x7f85e6c5c0a8>, 1.0]
  2549.  
  2550. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2551. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2552.  
  2553. ======================================================================
  2554. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 186, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2555. ----------------------------------------------------------------------
  2556. Traceback (most recent call last):
  2557. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2558. self.test(*self.arg)
  2559. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2560. gpuval = f(img, kern)
  2561. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2562. storage_map=self.fn.storage_map)
  2563. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2564. outputs = self.fn()
  2565. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2566. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2567. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dcd7b3d0>, Scalar(float32)]
  2568. Inputs shapes: [(1, 2, 2, 2), (1, 1, 2, 2), (1, 2, 3, 3), 'No shapes', ()]
  2569. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 2), (0, 9, 3, 1), 'No strides', ()]
  2570. Inputs values: ['not shown', <CudaNdarray object at 0x7f85af28bd30>, 'not shown', <PyCObject object at 0x7f85e6572b70>, 1.0]
  2571.  
  2572. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2573. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2574.  
  2575. ======================================================================
  2576. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 187, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2577. ----------------------------------------------------------------------
  2578. Traceback (most recent call last):
  2579. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2580. self.test(*self.arg)
  2581. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2582. gpuval = f(img, kern)
  2583. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2584. storage_map=self.fn.storage_map)
  2585. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2586. outputs = self.fn()
  2587. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2588. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2589. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c74fe1d0>, Scalar(float32)]
  2590. Inputs shapes: [(1, 2, 2, 2), (3, 1, 2, 2), (3, 2, 3, 3), 'No shapes', ()]
  2591. Inputs strides: [(0, 4, 2, 1), (16, 0, 8, 2), (18, 9, 3, 1), 'No strides', ()]
  2592. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572698>, 1.0]
  2593.  
  2594. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2595. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2596.  
  2597. ======================================================================
  2598. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 188, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2599. ----------------------------------------------------------------------
  2600. Traceback (most recent call last):
  2601. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2602. self.test(*self.arg)
  2603. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2604. gpuval = f(img, kern)
  2605. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2606. storage_map=self.fn.storage_map)
  2607. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2608. outputs = self.fn()
  2609. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2610. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2611. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c9011990>, Scalar(float32)]
  2612. Inputs shapes: [(2, 1, 2, 2), (3, 2, 2, 2), (3, 1, 3, 3), 'No shapes', ()]
  2613. Inputs strides: [(4, 0, 2, 1), (32, 16, 8, 2), (9, 0, 3, 1), 'No strides', ()]
  2614. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6572198>, 1.0]
  2615.  
  2616. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2617. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2618.  
  2619. ======================================================================
  2620. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 189, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2621. ----------------------------------------------------------------------
  2622. Traceback (most recent call last):
  2623. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2624. self.test(*self.arg)
  2625. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2626. gpuval = f(img, kern)
  2627. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2628. storage_map=self.fn.storage_map)
  2629. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2630. outputs = self.fn()
  2631. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2632. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2633. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85deeaa050>, Scalar(float32)]
  2634. Inputs shapes: [(2, 2, 2, 2), (1, 2, 2, 2), (1, 2, 3, 3), 'No shapes', ()]
  2635. Inputs strides: [(8, 4, 2, 1), (0, 16, 8, 2), (0, 9, 3, 1), 'No strides', ()]
  2636. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9ec60>, 1.0]
  2637.  
  2638. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2639. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2640.  
  2641. ======================================================================
  2642. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 190, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2643. ----------------------------------------------------------------------
  2644. Traceback (most recent call last):
  2645. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2646. self.test(*self.arg)
  2647. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2648. gpuval = f(img, kern)
  2649. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2650. storage_map=self.fn.storage_map)
  2651. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2652. outputs = self.fn()
  2653. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2654. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2655. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af310150>, Scalar(float32)]
  2656. Inputs shapes: [(2, 2, 2, 2), (2, 2, 2, 2), (2, 2, 3, 3), 'No shapes', ()]
  2657. Inputs strides: [(8, 4, 2, 1), (32, 16, 8, 2), (18, 9, 3, 1), 'No strides', ()]
  2658. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9e788>, 1.0]
  2659.  
  2660. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2661. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2662.  
  2663. ======================================================================
  2664. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 191, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2665. ----------------------------------------------------------------------
  2666. Traceback (most recent call last):
  2667. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2668. self.test(*self.arg)
  2669. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2670. gpuval = f(img, kern)
  2671. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2672. storage_map=self.fn.storage_map)
  2673. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2674. outputs = self.fn()
  2675. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2676. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2677. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e06ec590>, Scalar(float32)]
  2678. Inputs shapes: [(2, 4, 2, 2), (3, 2, 2, 2), (3, 4, 3, 3), 'No shapes', ()]
  2679. Inputs strides: [(16, 4, 2, 1), (32, 16, 8, 2), (36, 9, 3, 1), 'No strides', ()]
  2680. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85deb9e288>, 1.0]
  2681.  
  2682. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2683. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2684.  
  2685. ======================================================================
  2686. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 192, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2687. ----------------------------------------------------------------------
  2688. Traceback (most recent call last):
  2689. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2690. self.test(*self.arg)
  2691. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2692. gpuval = f(img, kern)
  2693. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2694. storage_map=self.fn.storage_map)
  2695. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2696. outputs = self.fn()
  2697. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2698. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2699. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d9434ad0>, Scalar(float32)]
  2700. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 3), (1, 1, 4, 4), 'No shapes', ()]
  2701. Inputs strides: [(4, 0, 2, 1), (0, 36, 12, 2), (0, 0, 4, 1), 'No strides', ()]
  2702. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b558>, 1.0]
  2703.  
  2704. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2705. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2706.  
  2707. ======================================================================
  2708. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 193, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2709. ----------------------------------------------------------------------
  2710. Traceback (most recent call last):
  2711. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2712. self.test(*self.arg)
  2713. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2714. gpuval = f(img, kern)
  2715. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2716. storage_map=self.fn.storage_map)
  2717. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2718. outputs = self.fn()
  2719. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2720. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2721. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d37bfad0>, Scalar(float32)]
  2722. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 3), (3, 1, 4, 4), 'No shapes', ()]
  2723. Inputs strides: [(0, 0, 2, 1), (36, 0, 12, 2), (16, 0, 4, 1), 'No strides', ()]
  2724. Inputs values: [<CudaNdarray object at 0x7f85c737a470>, 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b940>, 1.0]
  2725.  
  2726. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2727. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2728.  
  2729. ======================================================================
  2730. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 194, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2731. ----------------------------------------------------------------------
  2732. Traceback (most recent call last):
  2733. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2734. self.test(*self.arg)
  2735. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2736. gpuval = f(img, kern)
  2737. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2738. storage_map=self.fn.storage_map)
  2739. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2740. outputs = self.fn()
  2741. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2742. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2743. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dedf6a90>, Scalar(float32)]
  2744. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 3), (1, 2, 4, 4), 'No shapes', ()]
  2745. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 2), (0, 16, 4, 1), 'No strides', ()]
  2746. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd3b6e8>, 1.0]
  2747.  
  2748. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2749. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2750.  
  2751. ======================================================================
  2752. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 195, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2753. ----------------------------------------------------------------------
  2754. Traceback (most recent call last):
  2755. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2756. self.test(*self.arg)
  2757. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2758. gpuval = f(img, kern)
  2759. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2760. storage_map=self.fn.storage_map)
  2761. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2762. outputs = self.fn()
  2763. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2764. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2765. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c74745d0>, Scalar(float32)]
  2766. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 3), (3, 2, 4, 4), 'No shapes', ()]
  2767. Inputs strides: [(0, 4, 2, 1), (36, 0, 12, 2), (32, 16, 4, 1), 'No strides', ()]
  2768. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5feb8>, 1.0]
  2769.  
  2770. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2771. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2772.  
  2773. ======================================================================
  2774. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 196, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2775. ----------------------------------------------------------------------
  2776. Traceback (most recent call last):
  2777. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2778. self.test(*self.arg)
  2779. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2780. gpuval = f(img, kern)
  2781. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2782. storage_map=self.fn.storage_map)
  2783. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2784. outputs = self.fn()
  2785. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2786. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2787. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e07b31d0>, Scalar(float32)]
  2788. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 3), (3, 1, 4, 4), 'No shapes', ()]
  2789. Inputs strides: [(4, 0, 2, 1), (72, 36, 12, 2), (16, 0, 4, 1), 'No strides', ()]
  2790. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fb20>, 1.0]
  2791.  
  2792. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2793. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2794.  
  2795. ======================================================================
  2796. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 197, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2797. ----------------------------------------------------------------------
  2798. Traceback (most recent call last):
  2799. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2800. self.test(*self.arg)
  2801. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2802. gpuval = f(img, kern)
  2803. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2804. storage_map=self.fn.storage_map)
  2805. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2806. outputs = self.fn()
  2807. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2808. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2809. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c901e210>, Scalar(float32)]
  2810. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 3), (1, 2, 4, 4), 'No shapes', ()]
  2811. Inputs strides: [(8, 4, 2, 1), (0, 36, 12, 2), (0, 16, 4, 1), 'No strides', ()]
  2812. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5f8a0>, 1.0]
  2813.  
  2814. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2815. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2816.  
  2817. ======================================================================
  2818. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 198, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2819. ----------------------------------------------------------------------
  2820. Traceback (most recent call last):
  2821. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2822. self.test(*self.arg)
  2823. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2824. gpuval = f(img, kern)
  2825. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2826. storage_map=self.fn.storage_map)
  2827. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2828. outputs = self.fn()
  2829. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2830. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2831. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e6c01c50>, Scalar(float32)]
  2832. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 3), (2, 2, 4, 4), 'No shapes', ()]
  2833. Inputs strides: [(8, 4, 2, 1), (72, 36, 12, 2), (32, 16, 4, 1), 'No strides', ()]
  2834. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85c5a5fb48>, 1.0]
  2835.  
  2836. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2837. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2838.  
  2839. ======================================================================
  2840. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 199, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2841. ----------------------------------------------------------------------
  2842. Traceback (most recent call last):
  2843. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2844. self.test(*self.arg)
  2845. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2846. gpuval = f(img, kern)
  2847. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2848. storage_map=self.fn.storage_map)
  2849. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2850. outputs = self.fn()
  2851. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2852. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2853. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d3c3bfd0>, Scalar(float32)]
  2854. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 3), (3, 4, 4, 4), 'No shapes', ()]
  2855. Inputs strides: [(16, 4, 2, 1), (72, 36, 12, 2), (64, 16, 4, 1), 'No strides', ()]
  2856. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a4e0>, 1.0]
  2857.  
  2858. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2859. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2860.  
  2861. ======================================================================
  2862. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 200, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2863. ----------------------------------------------------------------------
  2864. Traceback (most recent call last):
  2865. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2866. self.test(*self.arg)
  2867. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2868. gpuval = f(img, kern)
  2869. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2870. storage_map=self.fn.storage_map)
  2871. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2872. outputs = self.fn()
  2873. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2874. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2875. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcfcda10>, Scalar(float32)]
  2876. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 3), (1, 1, 5, 4), 'No shapes', ()]
  2877. Inputs strides: [(4, 0, 2, 1), (0, 48, 12, 2), (0, 0, 4, 1), 'No strides', ()]
  2878. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a3c8>, 1.0]
  2879.  
  2880. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2881. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2882.  
  2883. ======================================================================
  2884. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 201, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2885. ----------------------------------------------------------------------
  2886. Traceback (most recent call last):
  2887. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2888. self.test(*self.arg)
  2889. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2890. gpuval = f(img, kern)
  2891. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2892. storage_map=self.fn.storage_map)
  2893. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2894. outputs = self.fn()
  2895. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2896. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2897. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af3108d0>, Scalar(float32)]
  2898. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 3), (3, 1, 5, 4), 'No shapes', ()]
  2899. Inputs strides: [(0, 0, 2, 1), (48, 0, 12, 2), (20, 0, 4, 1), 'No strides', ()]
  2900. Inputs values: [<CudaNdarray object at 0x7f85d93f1830>, 'not shown', 'not shown', <PyCObject object at 0x7f85a973a210>, 1.0]
  2901.  
  2902. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2903. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2904.  
  2905. ======================================================================
  2906. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 202, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2907. ----------------------------------------------------------------------
  2908. Traceback (most recent call last):
  2909. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2910. self.test(*self.arg)
  2911. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2912. gpuval = f(img, kern)
  2913. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2914. storage_map=self.fn.storage_map)
  2915. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2916. outputs = self.fn()
  2917. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2918. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2919. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dd049810>, Scalar(float32)]
  2920. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 3), (1, 2, 5, 4), 'No shapes', ()]
  2921. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 2), (0, 20, 4, 1), 'No strides', ()]
  2922. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a9e0>, 1.0]
  2923.  
  2924. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2925. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2926.  
  2927. ======================================================================
  2928. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 203, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2929. ----------------------------------------------------------------------
  2930. Traceback (most recent call last):
  2931. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2932. self.test(*self.arg)
  2933. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2934. gpuval = f(img, kern)
  2935. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2936. storage_map=self.fn.storage_map)
  2937. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2938. outputs = self.fn()
  2939. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2940. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2941. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af399810>, Scalar(float32)]
  2942. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 3), (3, 2, 5, 4), 'No shapes', ()]
  2943. Inputs strides: [(0, 4, 2, 1), (48, 0, 12, 2), (40, 20, 4, 1), 'No strides', ()]
  2944. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a0d0>, 1.0]
  2945.  
  2946. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2947. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2948.  
  2949. ======================================================================
  2950. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 204, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2951. ----------------------------------------------------------------------
  2952. Traceback (most recent call last):
  2953. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2954. self.test(*self.arg)
  2955. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2956. gpuval = f(img, kern)
  2957. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2958. storage_map=self.fn.storage_map)
  2959. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2960. outputs = self.fn()
  2961. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2962. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2963. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c72f2610>, Scalar(float32)]
  2964. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 3), (3, 1, 5, 4), 'No shapes', ()]
  2965. Inputs strides: [(4, 0, 2, 1), (96, 48, 12, 2), (20, 0, 4, 1), 'No strides', ()]
  2966. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94334e0>, 1.0]
  2967.  
  2968. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2969. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2970.  
  2971. ======================================================================
  2972. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 205, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2973. ----------------------------------------------------------------------
  2974. Traceback (most recent call last):
  2975. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2976. self.test(*self.arg)
  2977. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  2978. gpuval = f(img, kern)
  2979. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  2980. storage_map=self.fn.storage_map)
  2981. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  2982. outputs = self.fn()
  2983. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  2984. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  2985. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85df3c81d0>, Scalar(float32)]
  2986. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 3), (1, 2, 5, 4), 'No shapes', ()]
  2987. Inputs strides: [(8, 4, 2, 1), (0, 48, 12, 2), (0, 20, 4, 1), 'No strides', ()]
  2988. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433648>, 1.0]
  2989.  
  2990. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  2991. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  2992.  
  2993. ======================================================================
  2994. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 206, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  2995. ----------------------------------------------------------------------
  2996. Traceback (most recent call last):
  2997. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  2998. self.test(*self.arg)
  2999. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3000. gpuval = f(img, kern)
  3001. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3002. storage_map=self.fn.storage_map)
  3003. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3004. outputs = self.fn()
  3005. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3006. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3007. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e06ece10>, Scalar(float32)]
  3008. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 3), (2, 2, 5, 4), 'No shapes', ()]
  3009. Inputs strides: [(8, 4, 2, 1), (96, 48, 12, 2), (40, 20, 4, 1), 'No strides', ()]
  3010. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94332d8>, 1.0]
  3011.  
  3012. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3013. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3014.  
  3015. ======================================================================
  3016. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 207, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3017. ----------------------------------------------------------------------
  3018. Traceback (most recent call last):
  3019. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3020. self.test(*self.arg)
  3021. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3022. gpuval = f(img, kern)
  3023. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3024. storage_map=self.fn.storage_map)
  3025. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3026. outputs = self.fn()
  3027. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3028. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3029. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d8917d90>, Scalar(float32)]
  3030. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 3), (3, 4, 5, 4), 'No shapes', ()]
  3031. Inputs strides: [(16, 4, 2, 1), (96, 48, 12, 2), (80, 20, 4, 1), 'No strides', ()]
  3032. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433f30>, 1.0]
  3033.  
  3034. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3035. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3036.  
  3037. ======================================================================
  3038. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 208, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3039. ----------------------------------------------------------------------
  3040. Traceback (most recent call last):
  3041. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3042. self.test(*self.arg)
  3043. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3044. gpuval = f(img, kern)
  3045. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3046. storage_map=self.fn.storage_map)
  3047. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3048. outputs = self.fn()
  3049. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3050. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3051. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c75ccf50>, Scalar(float32)]
  3052. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 4), (1, 1, 4, 5), 'No shapes', ()]
  3053. Inputs strides: [(4, 0, 2, 1), (0, 48, 16, 2), (0, 0, 5, 1), 'No strides', ()]
  3054. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1670>, 1.0]
  3055.  
  3056. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3057. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3058.  
  3059. ======================================================================
  3060. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 209, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3061. ----------------------------------------------------------------------
  3062. Traceback (most recent call last):
  3063. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3064. self.test(*self.arg)
  3065. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3066. gpuval = f(img, kern)
  3067. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3068. storage_map=self.fn.storage_map)
  3069. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3070. outputs = self.fn()
  3071. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3072. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3073. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d3c1c7d0>, Scalar(float32)]
  3074. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 4), (3, 1, 4, 5), 'No shapes', ()]
  3075. Inputs strides: [(0, 0, 2, 1), (48, 0, 16, 2), (20, 0, 5, 1), 'No strides', ()]
  3076. Inputs values: [<CudaNdarray object at 0x7f85af10d0b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf19b8>, 1.0]
  3077.  
  3078. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3079. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3080.  
  3081. ======================================================================
  3082. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 210, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3083. ----------------------------------------------------------------------
  3084. Traceback (most recent call last):
  3085. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3086. self.test(*self.arg)
  3087. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3088. gpuval = f(img, kern)
  3089. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3090. storage_map=self.fn.storage_map)
  3091. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3092. outputs = self.fn()
  3093. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3094. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3095. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d37bf410>, Scalar(float32)]
  3096. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 4), (1, 2, 4, 5), 'No shapes', ()]
  3097. Inputs strides: [(0, 4, 2, 1), (0, 0, 16, 2), (0, 20, 5, 1), 'No strides', ()]
  3098. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1af8>, 1.0]
  3099.  
  3100. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3101. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3102.  
  3103. ======================================================================
  3104. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 211, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3105. ----------------------------------------------------------------------
  3106. Traceback (most recent call last):
  3107. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3108. self.test(*self.arg)
  3109. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3110. gpuval = f(img, kern)
  3111. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3112. storage_map=self.fn.storage_map)
  3113. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3114. outputs = self.fn()
  3115. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3116. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3117. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85df37e3d0>, Scalar(float32)]
  3118. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 4), (3, 2, 4, 5), 'No shapes', ()]
  3119. Inputs strides: [(0, 4, 2, 1), (48, 0, 16, 2), (40, 20, 5, 1), 'No strides', ()]
  3120. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1cd8>, 1.0]
  3121.  
  3122. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3123. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3124.  
  3125. ======================================================================
  3126. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 212, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3127. ----------------------------------------------------------------------
  3128. Traceback (most recent call last):
  3129. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3130. self.test(*self.arg)
  3131. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3132. gpuval = f(img, kern)
  3133. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3134. storage_map=self.fn.storage_map)
  3135. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3136. outputs = self.fn()
  3137. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3138. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3139. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85a94217d0>, Scalar(float32)]
  3140. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 4), (3, 1, 4, 5), 'No shapes', ()]
  3141. Inputs strides: [(4, 0, 2, 1), (96, 48, 16, 2), (20, 0, 5, 1), 'No strides', ()]
  3142. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614760>, 1.0]
  3143.  
  3144. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3145. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3146.  
  3147. ======================================================================
  3148. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 213, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3149. ----------------------------------------------------------------------
  3150. Traceback (most recent call last):
  3151. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3152. self.test(*self.arg)
  3153. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3154. gpuval = f(img, kern)
  3155. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3156. storage_map=self.fn.storage_map)
  3157. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3158. outputs = self.fn()
  3159. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3160. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3161. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e072add0>, Scalar(float32)]
  3162. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 4), (1, 2, 4, 5), 'No shapes', ()]
  3163. Inputs strides: [(8, 4, 2, 1), (0, 48, 16, 2), (0, 20, 5, 1), 'No strides', ()]
  3164. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614cd8>, 1.0]
  3165.  
  3166. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3167. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3168.  
  3169. ======================================================================
  3170. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 214, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3171. ----------------------------------------------------------------------
  3172. Traceback (most recent call last):
  3173. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3174. self.test(*self.arg)
  3175. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3176. gpuval = f(img, kern)
  3177. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3178. storage_map=self.fn.storage_map)
  3179. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3180. outputs = self.fn()
  3181. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3182. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3183. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dbbb24d0>, Scalar(float32)]
  3184. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 4), (2, 2, 4, 5), 'No shapes', ()]
  3185. Inputs strides: [(8, 4, 2, 1), (96, 48, 16, 2), (40, 20, 5, 1), 'No strides', ()]
  3186. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614da0>, 1.0]
  3187.  
  3188. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3189. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3190.  
  3191. ======================================================================
  3192. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 215, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3193. ----------------------------------------------------------------------
  3194. Traceback (most recent call last):
  3195. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3196. self.test(*self.arg)
  3197. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3198. gpuval = f(img, kern)
  3199. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3200. storage_map=self.fn.storage_map)
  3201. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3202. outputs = self.fn()
  3203. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3204. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3205. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dee4f0d0>, Scalar(float32)]
  3206. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 4), (3, 4, 4, 5), 'No shapes', ()]
  3207. Inputs strides: [(16, 4, 2, 1), (96, 48, 16, 2), (80, 20, 5, 1), 'No strides', ()]
  3208. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a96143c8>, 1.0]
  3209.  
  3210. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3211. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3212.  
  3213. ======================================================================
  3214. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 216, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3215. ----------------------------------------------------------------------
  3216. Traceback (most recent call last):
  3217. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3218. self.test(*self.arg)
  3219. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3220. gpuval = f(img, kern)
  3221. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3222. storage_map=self.fn.storage_map)
  3223. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3224. outputs = self.fn()
  3225. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3226. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3227. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac74b590>, Scalar(float32)]
  3228. Inputs shapes: [(2, 1, 3, 2), (1, 2, 4, 4), (1, 1, 6, 5), 'No shapes', ()]
  3229. Inputs strides: [(6, 0, 2, 1), (0, 64, 16, 2), (0, 0, 5, 1), 'No strides', ()]
  3230. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4b98>, 1.0]
  3231.  
  3232. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3233. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3234.  
  3235. ======================================================================
  3236. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 217, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3237. ----------------------------------------------------------------------
  3238. Traceback (most recent call last):
  3239. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3240. self.test(*self.arg)
  3241. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3242. gpuval = f(img, kern)
  3243. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3244. storage_map=self.fn.storage_map)
  3245. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3246. outputs = self.fn()
  3247. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3248. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3249. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af2e4c90>, Scalar(float32)]
  3250. Inputs shapes: [(1, 1, 3, 2), (3, 1, 4, 4), (3, 1, 6, 5), 'No shapes', ()]
  3251. Inputs strides: [(0, 0, 2, 1), (64, 0, 16, 2), (30, 0, 5, 1), 'No strides', ()]
  3252. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4918>, 1.0]
  3253.  
  3254. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3255. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3256.  
  3257. ======================================================================
  3258. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 218, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3259. ----------------------------------------------------------------------
  3260. Traceback (most recent call last):
  3261. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3262. self.test(*self.arg)
  3263. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3264. gpuval = f(img, kern)
  3265. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3266. storage_map=self.fn.storage_map)
  3267. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3268. outputs = self.fn()
  3269. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3270. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3271. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85a91bde50>, Scalar(float32)]
  3272. Inputs shapes: [(1, 2, 3, 2), (1, 1, 4, 4), (1, 2, 6, 5), 'No shapes', ()]
  3273. Inputs strides: [(0, 6, 2, 1), (0, 0, 16, 2), (0, 30, 5, 1), 'No strides', ()]
  3274. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4670>, 1.0]
  3275.  
  3276. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3277. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3278.  
  3279. ======================================================================
  3280. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 219, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3281. ----------------------------------------------------------------------
  3282. Traceback (most recent call last):
  3283. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3284. self.test(*self.arg)
  3285. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3286. gpuval = f(img, kern)
  3287. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3288. storage_map=self.fn.storage_map)
  3289. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3290. outputs = self.fn()
  3291. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3292. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3293. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ab6999d0>, Scalar(float32)]
  3294. Inputs shapes: [(1, 2, 3, 2), (3, 1, 4, 4), (3, 2, 6, 5), 'No shapes', ()]
  3295. Inputs strides: [(0, 6, 2, 1), (64, 0, 16, 2), (60, 30, 5, 1), 'No strides', ()]
  3296. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee12620>, 1.0]
  3297.  
  3298. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3299. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3300.  
  3301. ======================================================================
  3302. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 220, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3303. ----------------------------------------------------------------------
  3304. Traceback (most recent call last):
  3305. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3306. self.test(*self.arg)
  3307. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3308. gpuval = f(img, kern)
  3309. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3310. storage_map=self.fn.storage_map)
  3311. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3312. outputs = self.fn()
  3313. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3314. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3315. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c7361e50>, Scalar(float32)]
  3316. Inputs shapes: [(2, 1, 3, 2), (3, 2, 4, 4), (3, 1, 6, 5), 'No shapes', ()]
  3317. Inputs strides: [(6, 0, 2, 1), (128, 64, 16, 2), (30, 0, 5, 1), 'No strides', ()]
  3318. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee124b8>, 1.0]
  3319.  
  3320. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3321. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3322.  
  3323. ======================================================================
  3324. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 221, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3325. ----------------------------------------------------------------------
  3326. Traceback (most recent call last):
  3327. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3328. self.test(*self.arg)
  3329. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3330. gpuval = f(img, kern)
  3331. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3332. storage_map=self.fn.storage_map)
  3333. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3334. outputs = self.fn()
  3335. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3336. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3337. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af292d50>, Scalar(float32)]
  3338. Inputs shapes: [(2, 2, 3, 2), (1, 2, 4, 4), (1, 2, 6, 5), 'No shapes', ()]
  3339. Inputs strides: [(12, 6, 2, 1), (0, 64, 16, 2), (0, 30, 5, 1), 'No strides', ()]
  3340. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b01e8>, 1.0]
  3341.  
  3342. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3343. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3344.  
  3345. ======================================================================
  3346. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 222, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3347. ----------------------------------------------------------------------
  3348. Traceback (most recent call last):
  3349. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3350. self.test(*self.arg)
  3351. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3352. gpuval = f(img, kern)
  3353. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3354. storage_map=self.fn.storage_map)
  3355. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3356. outputs = self.fn()
  3357. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3358. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3359. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c5a20350>, Scalar(float32)]
  3360. Inputs shapes: [(2, 2, 3, 2), (2, 2, 4, 4), (2, 2, 6, 5), 'No shapes', ()]
  3361. Inputs strides: [(12, 6, 2, 1), (128, 64, 16, 2), (60, 30, 5, 1), 'No strides', ()]
  3362. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0648>, 1.0]
  3363.  
  3364. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3365. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3366.  
  3367. ======================================================================
  3368. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 223, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3369. ----------------------------------------------------------------------
  3370. Traceback (most recent call last):
  3371. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3372. self.test(*self.arg)
  3373. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3374. gpuval = f(img, kern)
  3375. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3376. storage_map=self.fn.storage_map)
  3377. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3378. outputs = self.fn()
  3379. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3380. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3381. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac5ab790>, Scalar(float32)]
  3382. Inputs shapes: [(2, 4, 3, 2), (3, 2, 4, 4), (3, 4, 6, 5), 'No shapes', ()]
  3383. Inputs strides: [(24, 6, 2, 1), (128, 64, 16, 2), (120, 30, 5, 1), 'No strides', ()]
  3384. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b0968>, 1.0]
  3385.  
  3386. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3387. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3388.  
  3389. ======================================================================
  3390. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 224, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3391. ----------------------------------------------------------------------
  3392. Traceback (most recent call last):
  3393. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3394. self.test(*self.arg)
  3395. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3396. gpuval = f(img, kern)
  3397. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3398. storage_map=self.fn.storage_map)
  3399. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3400. outputs = self.fn()
  3401. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3402. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3403. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddb4bc10>, Scalar(float32)]
  3404. Inputs shapes: [(2, 1, 2, 3), (1, 2, 4, 4), (1, 1, 5, 6), 'No shapes', ()]
  3405. Inputs strides: [(6, 0, 3, 1), (0, 64, 16, 2), (0, 0, 6, 1), 'No strides', ()]
  3406. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac1b00f8>, 1.0]
  3407.  
  3408. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3409. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3410.  
  3411. ======================================================================
  3412. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 225, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3413. ----------------------------------------------------------------------
  3414. Traceback (most recent call last):
  3415. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3416. self.test(*self.arg)
  3417. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3418. gpuval = f(img, kern)
  3419. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3420. storage_map=self.fn.storage_map)
  3421. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3422. outputs = self.fn()
  3423. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3424. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3425. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c75cc850>, Scalar(float32)]
  3426. Inputs shapes: [(1, 1, 2, 3), (3, 1, 4, 4), (3, 1, 5, 6), 'No shapes', ()]
  3427. Inputs strides: [(0, 0, 3, 1), (64, 0, 16, 2), (30, 0, 6, 1), 'No strides', ()]
  3428. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd223a0>, 1.0]
  3429.  
  3430. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3431. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3432.  
  3433. ======================================================================
  3434. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 226, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3435. ----------------------------------------------------------------------
  3436. Traceback (most recent call last):
  3437. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3438. self.test(*self.arg)
  3439. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3440. gpuval = f(img, kern)
  3441. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3442. storage_map=self.fn.storage_map)
  3443. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3444. outputs = self.fn()
  3445. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3446. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3447. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e0786d90>, Scalar(float32)]
  3448. Inputs shapes: [(1, 2, 2, 3), (1, 1, 4, 4), (1, 2, 5, 6), 'No shapes', ()]
  3449. Inputs strides: [(0, 6, 3, 1), (0, 0, 16, 2), (0, 30, 6, 1), 'No strides', ()]
  3450. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dc2bf058>, 1.0]
  3451.  
  3452. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3453. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3454.  
  3455. ======================================================================
  3456. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 227, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3457. ----------------------------------------------------------------------
  3458. Traceback (most recent call last):
  3459. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3460. self.test(*self.arg)
  3461. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3462. gpuval = f(img, kern)
  3463. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3464. storage_map=self.fn.storage_map)
  3465. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3466. outputs = self.fn()
  3467. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3468. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3469. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac74b250>, Scalar(float32)]
  3470. Inputs shapes: [(1, 2, 2, 3), (3, 1, 4, 4), (3, 2, 5, 6), 'No shapes', ()]
  3471. Inputs strides: [(0, 6, 3, 1), (64, 0, 16, 2), (60, 30, 6, 1), 'No strides', ()]
  3472. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd04300>, 1.0]
  3473.  
  3474. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3475. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3476.  
  3477. ======================================================================
  3478. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 228, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3479. ----------------------------------------------------------------------
  3480. Traceback (most recent call last):
  3481. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3482. self.test(*self.arg)
  3483. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3484. gpuval = f(img, kern)
  3485. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3486. storage_map=self.fn.storage_map)
  3487. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3488. outputs = self.fn()
  3489. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3490. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3491. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac59d7d0>, Scalar(float32)]
  3492. Inputs shapes: [(2, 1, 2, 3), (3, 2, 4, 4), (3, 1, 5, 6), 'No shapes', ()]
  3493. Inputs strides: [(6, 0, 3, 1), (128, 64, 16, 2), (30, 0, 6, 1), 'No strides', ()]
  3494. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd04580>, 1.0]
  3495.  
  3496. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3497. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3498.  
  3499. ======================================================================
  3500. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 229, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3501. ----------------------------------------------------------------------
  3502. Traceback (most recent call last):
  3503. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3504. self.test(*self.arg)
  3505. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3506. gpuval = f(img, kern)
  3507. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3508. storage_map=self.fn.storage_map)
  3509. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3510. outputs = self.fn()
  3511. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3512. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3513. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c75114d0>, Scalar(float32)]
  3514. Inputs shapes: [(2, 2, 2, 3), (1, 2, 4, 4), (1, 2, 5, 6), 'No shapes', ()]
  3515. Inputs strides: [(12, 6, 3, 1), (0, 64, 16, 2), (0, 30, 6, 1), 'No strides', ()]
  3516. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd04800>, 1.0]
  3517.  
  3518. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3519. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3520.  
  3521. ======================================================================
  3522. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 230, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3523. ----------------------------------------------------------------------
  3524. Traceback (most recent call last):
  3525. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3526. self.test(*self.arg)
  3527. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3528. gpuval = f(img, kern)
  3529. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3530. storage_map=self.fn.storage_map)
  3531. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3532. outputs = self.fn()
  3533. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3534. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3535. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af22cf10>, Scalar(float32)]
  3536. Inputs shapes: [(2, 2, 2, 3), (2, 2, 4, 4), (2, 2, 5, 6), 'No shapes', ()]
  3537. Inputs strides: [(12, 6, 3, 1), (128, 64, 16, 2), (60, 30, 6, 1), 'No strides', ()]
  3538. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd04aa8>, 1.0]
  3539.  
  3540. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3541. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3542.  
  3543. ======================================================================
  3544. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 231, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85f61fda10>, None)
  3545. ----------------------------------------------------------------------
  3546. Traceback (most recent call last):
  3547. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3548. self.test(*self.arg)
  3549. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3550. gpuval = f(img, kern)
  3551. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3552. storage_map=self.fn.storage_map)
  3553. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3554. outputs = self.fn()
  3555. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3556. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3557. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85aa602510>, Scalar(float32)]
  3558. Inputs shapes: [(2, 4, 2, 3), (3, 2, 4, 4), (3, 4, 5, 6), 'No shapes', ()]
  3559. Inputs strides: [(24, 6, 3, 1), (128, 64, 16, 2), (120, 30, 6, 1), 'No strides', ()]
  3560. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd04d28>, 1.0]
  3561.  
  3562. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3563. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3564.  
  3565. ======================================================================
  3566. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 64, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3567. ----------------------------------------------------------------------
  3568. Traceback (most recent call last):
  3569. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3570. self.test(*self.arg)
  3571. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3572. gpuval = f(img, kern)
  3573. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3574. storage_map=self.fn.storage_map)
  3575. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3576. outputs = self.fn()
  3577. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3578. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3579. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, False, True)), CudaNdarrayType(float32, (True, True, False, True)), <theano.gof.type.CDataType object at 0x7f85af2a04d0>, Scalar(float32)]
  3580. Inputs shapes: [(2, 1, 1, 1), (1, 2, 2, 1), (1, 1, 2, 1), 'No shapes', ()]
  3581. Inputs strides: [(1, 0, 0, 0), (0, 4, 2, 0), (0, 0, 1, 0), 'No strides', ()]
  3582. Inputs values: [<CudaNdarray object at 0x7f85af45e930>, <CudaNdarray object at 0x7f85aa114d30>, <CudaNdarray object at 0x7f85d3c0f7b0>, <PyCObject object at 0x7f85a9614ad0>, 1.0]
  3583.  
  3584. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3585. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3586.  
  3587. ======================================================================
  3588. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 2, 2), (1, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 65, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3589. ----------------------------------------------------------------------
  3590. Traceback (most recent call last):
  3591. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3592. self.test(*self.arg)
  3593. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3594. gpuval = f(img, kern)
  3595. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3596. storage_map=self.fn.storage_map)
  3597. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3598. outputs = self.fn()
  3599. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3600. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3601. Inputs types: [CudaNdarrayType(float32, (True, True, True, True)), CudaNdarrayType(float32, (False, True, False, True)), CudaNdarrayType(float32, (False, True, False, True)), <theano.gof.type.CDataType object at 0x7f85c5a27ed0>, Scalar(float32)]
  3602. Inputs shapes: [(1, 1, 1, 1), (3, 1, 2, 1), (3, 1, 2, 1), 'No shapes', ()]
  3603. Inputs strides: [(0, 0, 0, 0), (4, 0, 2, 0), (2, 0, 1, 0), 'No strides', ()]
  3604. Inputs values: [<CudaNdarray object at 0x7f85dc2913f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85a96148f0>, 1.0]
  3605.  
  3606. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3607. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3608.  
  3609. ======================================================================
  3610. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 2, 2), (2, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 66, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3611. ----------------------------------------------------------------------
  3612. Traceback (most recent call last):
  3613. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3614. self.test(*self.arg)
  3615. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3616. gpuval = f(img, kern)
  3617. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3618. storage_map=self.fn.storage_map)
  3619. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3620. outputs = self.fn()
  3621. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3622. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3623. Inputs types: [CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, True, False, True)), CudaNdarrayType(float32, (True, False, False, True)), <theano.gof.type.CDataType object at 0x7f85ac70aa10>, Scalar(float32)]
  3624. Inputs shapes: [(1, 2, 1, 1), (1, 1, 2, 1), (1, 2, 2, 1), 'No shapes', ()]
  3625. Inputs strides: [(0, 1, 0, 0), (0, 0, 2, 0), (0, 2, 1, 0), 'No strides', ()]
  3626. Inputs values: [<CudaNdarray object at 0x7f85d8955770>, <CudaNdarray object at 0x7f85dc2913f0>, <CudaNdarray object at 0x7f85d8955730>, <PyCObject object at 0x7f85a9614878>, 1.0]
  3627.  
  3628. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3629. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3630.  
  3631. ======================================================================
  3632. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 2, 2), (2, 1, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 67, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3633. ----------------------------------------------------------------------
  3634. Traceback (most recent call last):
  3635. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3636. self.test(*self.arg)
  3637. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3638. gpuval = f(img, kern)
  3639. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3640. storage_map=self.fn.storage_map)
  3641. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3642. outputs = self.fn()
  3643. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3644. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3645. Inputs types: [CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (False, True, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85e06b5690>, Scalar(float32)]
  3646. Inputs shapes: [(1, 2, 1, 1), (3, 1, 2, 1), (3, 2, 2, 1), 'No shapes', ()]
  3647. Inputs strides: [(0, 1, 0, 0), (4, 0, 2, 0), (4, 2, 1, 0), 'No strides', ()]
  3648. Inputs values: [<CudaNdarray object at 0x7f85dedf6370>, 'not shown', 'not shown', <PyCObject object at 0x7f85a94330f8>, 1.0]
  3649.  
  3650. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3651. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3652.  
  3653. ======================================================================
  3654. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 68, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3655. ----------------------------------------------------------------------
  3656. Traceback (most recent call last):
  3657. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3658. self.test(*self.arg)
  3659. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3660. gpuval = f(img, kern)
  3661. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3662. storage_map=self.fn.storage_map)
  3663. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3664. outputs = self.fn()
  3665. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3666. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3667. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, True, False, True)), <theano.gof.type.CDataType object at 0x7f85af2a05d0>, Scalar(float32)]
  3668. Inputs shapes: [(2, 1, 1, 1), (3, 2, 2, 1), (3, 1, 2, 1), 'No shapes', ()]
  3669. Inputs strides: [(1, 0, 0, 0), (8, 4, 2, 0), (2, 0, 1, 0), 'No strides', ()]
  3670. Inputs values: [<CudaNdarray object at 0x7f85dedf61f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85a94333a0>, 1.0]
  3671.  
  3672. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3673. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3674.  
  3675. ======================================================================
  3676. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 69, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3677. ----------------------------------------------------------------------
  3678. Traceback (most recent call last):
  3679. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3680. self.test(*self.arg)
  3681. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3682. gpuval = f(img, kern)
  3683. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3684. storage_map=self.fn.storage_map)
  3685. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3686. outputs = self.fn()
  3687. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3688. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3689. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, False, True)), CudaNdarrayType(float32, (True, False, False, True)), <theano.gof.type.CDataType object at 0x7f85a9ef4810>, Scalar(float32)]
  3690. Inputs shapes: [(2, 2, 1, 1), (1, 2, 2, 1), (1, 2, 2, 1), 'No shapes', ()]
  3691. Inputs strides: [(2, 1, 0, 0), (0, 4, 2, 0), (0, 2, 1, 0), 'No strides', ()]
  3692. Inputs values: [<CudaNdarray object at 0x7f85dedf6ab0>, <CudaNdarray object at 0x7f85e0713930>, <CudaNdarray object at 0x7f85dedf6770>, <PyCObject object at 0x7f85a94337b0>, 1.0]
  3693.  
  3694. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3695. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3696.  
  3697. ======================================================================
  3698. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 70, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3699. ----------------------------------------------------------------------
  3700. Traceback (most recent call last):
  3701. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3702. self.test(*self.arg)
  3703. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3704. gpuval = f(img, kern)
  3705. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3706. storage_map=self.fn.storage_map)
  3707. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3708. outputs = self.fn()
  3709. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3710. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3711. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85dd01bbd0>, Scalar(float32)]
  3712. Inputs shapes: [(2, 2, 1, 1), (2, 2, 2, 1), (2, 2, 2, 1), 'No shapes', ()]
  3713. Inputs strides: [(2, 1, 0, 0), (8, 4, 2, 0), (4, 2, 1, 0), 'No strides', ()]
  3714. Inputs values: [<CudaNdarray object at 0x7f85c748bcf0>, 'not shown', 'not shown', <PyCObject object at 0x7f85a9433940>, 1.0]
  3715.  
  3716. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3717. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3718.  
  3719. ======================================================================
  3720. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 71, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3721. ----------------------------------------------------------------------
  3722. Traceback (most recent call last):
  3723. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3724. self.test(*self.arg)
  3725. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3726. gpuval = f(img, kern)
  3727. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3728. storage_map=self.fn.storage_map)
  3729. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3730. outputs = self.fn()
  3731. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3732. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3733. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, False, True)), CudaNdarrayType(float32, (False, False, False, True)), <theano.gof.type.CDataType object at 0x7f85ac59d890>, Scalar(float32)]
  3734. Inputs shapes: [(2, 4, 1, 1), (3, 2, 2, 1), (3, 4, 2, 1), 'No shapes', ()]
  3735. Inputs strides: [(4, 1, 0, 0), (8, 4, 2, 0), (8, 2, 1, 0), 'No strides', ()]
  3736. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433698>, 1.0]
  3737.  
  3738. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3739. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3740.  
  3741. ======================================================================
  3742. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 72, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3743. ----------------------------------------------------------------------
  3744. Traceback (most recent call last):
  3745. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3746. self.test(*self.arg)
  3747. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3748. gpuval = f(img, kern)
  3749. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3750. storage_map=self.fn.storage_map)
  3751. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3752. outputs = self.fn()
  3753. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3754. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3755. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85df0f30d0>, Scalar(float32)]
  3756. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 2), (1, 1, 5, 3), 'No shapes', ()]
  3757. Inputs strides: [(4, 0, 2, 1), (0, 16, 4, 2), (0, 0, 3, 1), 'No strides', ()]
  3758. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d7c88>, 1.0]
  3759.  
  3760. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3761. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3762.  
  3763. ======================================================================
  3764. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 73, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3765. ----------------------------------------------------------------------
  3766. Traceback (most recent call last):
  3767. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3768. self.test(*self.arg)
  3769. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3770. gpuval = f(img, kern)
  3771. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3772. storage_map=self.fn.storage_map)
  3773. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3774. outputs = self.fn()
  3775. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3776. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3777. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abd28c90>, Scalar(float32)]
  3778. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 2), (3, 1, 5, 3), 'No shapes', ()]
  3779. Inputs strides: [(0, 0, 2, 1), (16, 0, 4, 2), (15, 0, 3, 1), 'No strides', ()]
  3780. Inputs values: [<CudaNdarray object at 0x7f85e72823b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85d96d77d8>, 1.0]
  3781.  
  3782. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3783. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3784.  
  3785. ======================================================================
  3786. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 74, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3787. ----------------------------------------------------------------------
  3788. Traceback (most recent call last):
  3789. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3790. self.test(*self.arg)
  3791. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3792. gpuval = f(img, kern)
  3793. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3794. storage_map=self.fn.storage_map)
  3795. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3796. outputs = self.fn()
  3797. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3798. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3799. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c7422990>, Scalar(float32)]
  3800. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 2), (1, 2, 5, 3), 'No shapes', ()]
  3801. Inputs strides: [(0, 4, 2, 1), (0, 0, 4, 2), (0, 15, 3, 1), 'No strides', ()]
  3802. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d7378>, 1.0]
  3803.  
  3804. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3805. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3806.  
  3807. ======================================================================
  3808. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 75, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3809. ----------------------------------------------------------------------
  3810. Traceback (most recent call last):
  3811. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3812. self.test(*self.arg)
  3813. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3814. gpuval = f(img, kern)
  3815. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3816. storage_map=self.fn.storage_map)
  3817. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3818. outputs = self.fn()
  3819. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3820. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3821. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85abcca4d0>, Scalar(float32)]
  3822. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 2), (3, 2, 5, 3), 'No shapes', ()]
  3823. Inputs strides: [(0, 4, 2, 1), (16, 0, 4, 2), (30, 15, 3, 1), 'No strides', ()]
  3824. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e07cd120>, 1.0]
  3825.  
  3826. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3827. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3828.  
  3829. ======================================================================
  3830. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 76, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3831. ----------------------------------------------------------------------
  3832. Traceback (most recent call last):
  3833. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3834. self.test(*self.arg)
  3835. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3836. gpuval = f(img, kern)
  3837. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3838. storage_map=self.fn.storage_map)
  3839. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3840. outputs = self.fn()
  3841. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3842. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3843. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d946b110>, Scalar(float32)]
  3844. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 2), (3, 1, 5, 3), 'No shapes', ()]
  3845. Inputs strides: [(4, 0, 2, 1), (32, 16, 4, 2), (15, 0, 3, 1), 'No strides', ()]
  3846. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e03a0>, 1.0]
  3847.  
  3848. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3849. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3850.  
  3851. ======================================================================
  3852. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 77, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3853. ----------------------------------------------------------------------
  3854. Traceback (most recent call last):
  3855. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3856. self.test(*self.arg)
  3857. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3858. gpuval = f(img, kern)
  3859. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3860. storage_map=self.fn.storage_map)
  3861. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3862. outputs = self.fn()
  3863. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3864. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3865. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d941d310>, Scalar(float32)]
  3866. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 2), (1, 2, 5, 3), 'No shapes', ()]
  3867. Inputs strides: [(8, 4, 2, 1), (0, 16, 4, 2), (0, 15, 3, 1), 'No strides', ()]
  3868. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0620>, 1.0]
  3869.  
  3870. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3871. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3872.  
  3873. ======================================================================
  3874. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 78, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3875. ----------------------------------------------------------------------
  3876. Traceback (most recent call last):
  3877. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3878. self.test(*self.arg)
  3879. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3880. gpuval = f(img, kern)
  3881. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3882. storage_map=self.fn.storage_map)
  3883. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3884. outputs = self.fn()
  3885. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3886. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3887. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73fbed0>, Scalar(float32)]
  3888. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 2), (2, 2, 5, 3), 'No shapes', ()]
  3889. Inputs strides: [(8, 4, 2, 1), (32, 16, 4, 2), (30, 15, 3, 1), 'No strides', ()]
  3890. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e08c8>, 1.0]
  3891.  
  3892. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3893. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3894.  
  3895. ======================================================================
  3896. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 79, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3897. ----------------------------------------------------------------------
  3898. Traceback (most recent call last):
  3899. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3900. self.test(*self.arg)
  3901. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3902. gpuval = f(img, kern)
  3903. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3904. storage_map=self.fn.storage_map)
  3905. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3906. outputs = self.fn()
  3907. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3908. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3909. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dee8d650>, Scalar(float32)]
  3910. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 2), (3, 4, 5, 3), 'No shapes', ()]
  3911. Inputs strides: [(16, 4, 2, 1), (32, 16, 4, 2), (60, 15, 3, 1), 'No strides', ()]
  3912. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0b48>, 1.0]
  3913.  
  3914. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3915. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3916.  
  3917. ======================================================================
  3918. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 80, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3919. ----------------------------------------------------------------------
  3920. Traceback (most recent call last):
  3921. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3922. self.test(*self.arg)
  3923. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3924. gpuval = f(img, kern)
  3925. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3926. storage_map=self.fn.storage_map)
  3927. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3928. outputs = self.fn()
  3929. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3930. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3931. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddb1a5d0>, Scalar(float32)]
  3932. Inputs shapes: [(2, 1, 2, 2), (1, 2, 6, 3), (1, 1, 7, 4), 'No shapes', ()]
  3933. Inputs strides: [(4, 0, 2, 1), (0, 36, 6, 2), (0, 0, 4, 1), 'No strides', ()]
  3934. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0d78>, 1.0]
  3935.  
  3936. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3937. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3938.  
  3939. ======================================================================
  3940. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 81, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3941. ----------------------------------------------------------------------
  3942. Traceback (most recent call last):
  3943. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3944. self.test(*self.arg)
  3945. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3946. gpuval = f(img, kern)
  3947. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3948. storage_map=self.fn.storage_map)
  3949. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3950. outputs = self.fn()
  3951. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3952. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3953. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d9443350>, Scalar(float32)]
  3954. Inputs shapes: [(1, 1, 2, 2), (3, 1, 6, 3), (3, 1, 7, 4), 'No shapes', ()]
  3955. Inputs strides: [(0, 0, 2, 1), (36, 0, 6, 2), (28, 0, 4, 1), 'No strides', ()]
  3956. Inputs values: [<CudaNdarray object at 0x7f85dcfcdeb0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dee3b030>, 1.0]
  3957.  
  3958. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3959. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3960.  
  3961. ======================================================================
  3962. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 82, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3963. ----------------------------------------------------------------------
  3964. Traceback (most recent call last):
  3965. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3966. self.test(*self.arg)
  3967. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3968. gpuval = f(img, kern)
  3969. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3970. storage_map=self.fn.storage_map)
  3971. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3972. outputs = self.fn()
  3973. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3974. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3975. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e0743210>, Scalar(float32)]
  3976. Inputs shapes: [(1, 2, 2, 2), (1, 1, 6, 3), (1, 2, 7, 4), 'No shapes', ()]
  3977. Inputs strides: [(0, 4, 2, 1), (0, 0, 6, 2), (0, 28, 4, 1), 'No strides', ()]
  3978. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc02b0>, 1.0]
  3979.  
  3980. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  3981. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  3982.  
  3983. ======================================================================
  3984. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 83, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  3985. ----------------------------------------------------------------------
  3986. Traceback (most recent call last):
  3987. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  3988. self.test(*self.arg)
  3989. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  3990. gpuval = f(img, kern)
  3991. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  3992. storage_map=self.fn.storage_map)
  3993. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  3994. outputs = self.fn()
  3995. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  3996. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  3997. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7498690>, Scalar(float32)]
  3998. Inputs shapes: [(1, 2, 2, 2), (3, 1, 6, 3), (3, 2, 7, 4), 'No shapes', ()]
  3999. Inputs strides: [(0, 4, 2, 1), (36, 0, 6, 2), (56, 28, 4, 1), 'No strides', ()]
  4000. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0558>, 1.0]
  4001.  
  4002. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4003. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4004.  
  4005. ======================================================================
  4006. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 84, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4007. ----------------------------------------------------------------------
  4008. Traceback (most recent call last):
  4009. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4010. self.test(*self.arg)
  4011. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4012. gpuval = f(img, kern)
  4013. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4014. storage_map=self.fn.storage_map)
  4015. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4016. outputs = self.fn()
  4017. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4018. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4019. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d96d2190>, Scalar(float32)]
  4020. Inputs shapes: [(2, 1, 2, 2), (3, 2, 6, 3), (3, 1, 7, 4), 'No shapes', ()]
  4021. Inputs strides: [(4, 0, 2, 1), (72, 36, 6, 2), (28, 0, 4, 1), 'No strides', ()]
  4022. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc07d8>, 1.0]
  4023.  
  4024. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4025. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4026.  
  4027. ======================================================================
  4028. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 85, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4029. ----------------------------------------------------------------------
  4030. Traceback (most recent call last):
  4031. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4032. self.test(*self.arg)
  4033. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4034. gpuval = f(img, kern)
  4035. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4036. storage_map=self.fn.storage_map)
  4037. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4038. outputs = self.fn()
  4039. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4040. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4041. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d8959cd0>, Scalar(float32)]
  4042. Inputs shapes: [(2, 2, 2, 2), (1, 2, 6, 3), (1, 2, 7, 4), 'No shapes', ()]
  4043. Inputs strides: [(8, 4, 2, 1), (0, 36, 6, 2), (0, 28, 4, 1), 'No strides', ()]
  4044. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0a58>, 1.0]
  4045.  
  4046. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4047. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4048.  
  4049. ======================================================================
  4050. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 86, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4051. ----------------------------------------------------------------------
  4052. Traceback (most recent call last):
  4053. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4054. self.test(*self.arg)
  4055. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4056. gpuval = f(img, kern)
  4057. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4058. storage_map=self.fn.storage_map)
  4059. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4060. outputs = self.fn()
  4061. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4062. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4063. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85a91bdb50>, Scalar(float32)]
  4064. Inputs shapes: [(2, 2, 2, 2), (2, 2, 6, 3), (2, 2, 7, 4), 'No shapes', ()]
  4065. Inputs strides: [(8, 4, 2, 1), (72, 36, 6, 2), (56, 28, 4, 1), 'No strides', ()]
  4066. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0d00>, 1.0]
  4067.  
  4068. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4069. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4070.  
  4071. ======================================================================
  4072. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 87, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4073. ----------------------------------------------------------------------
  4074. Traceback (most recent call last):
  4075. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4076. self.test(*self.arg)
  4077. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4078. gpuval = f(img, kern)
  4079. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4080. storage_map=self.fn.storage_map)
  4081. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4082. outputs = self.fn()
  4083. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4084. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4085. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73274d0>, Scalar(float32)]
  4086. Inputs shapes: [(2, 4, 2, 2), (3, 2, 6, 3), (3, 4, 7, 4), 'No shapes', ()]
  4087. Inputs strides: [(16, 4, 2, 1), (72, 36, 6, 2), (112, 28, 4, 1), 'No strides', ()]
  4088. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0f80>, 1.0]
  4089.  
  4090. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4091. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4092.  
  4093. ======================================================================
  4094. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 88, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4095. ----------------------------------------------------------------------
  4096. Traceback (most recent call last):
  4097. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4098. self.test(*self.arg)
  4099. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4100. gpuval = f(img, kern)
  4101. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4102. storage_map=self.fn.storage_map)
  4103. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4104. outputs = self.fn()
  4105. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4106. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4107. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85a93dc250>, Scalar(float32)]
  4108. Inputs shapes: [(2, 1, 2, 2), (1, 2, 8, 3), (1, 1, 9, 4), 'No shapes', ()]
  4109. Inputs strides: [(4, 0, 2, 1), (0, 48, 6, 2), (0, 0, 4, 1), 'No strides', ()]
  4110. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e65551e8>, 1.0]
  4111.  
  4112. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4113. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4114.  
  4115. ======================================================================
  4116. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 89, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4117. ----------------------------------------------------------------------
  4118. Traceback (most recent call last):
  4119. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4120. self.test(*self.arg)
  4121. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4122. gpuval = f(img, kern)
  4123. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4124. storage_map=self.fn.storage_map)
  4125. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4126. outputs = self.fn()
  4127. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4128. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4129. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c73b2990>, Scalar(float32)]
  4130. Inputs shapes: [(1, 1, 2, 2), (3, 1, 8, 3), (3, 1, 9, 4), 'No shapes', ()]
  4131. Inputs strides: [(0, 0, 2, 1), (48, 0, 6, 2), (36, 0, 4, 1), 'No strides', ()]
  4132. Inputs values: [<CudaNdarray object at 0x7f85dee246f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85dee86468>, 1.0]
  4133.  
  4134. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4135. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4136.  
  4137. ======================================================================
  4138. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 90, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4139. ----------------------------------------------------------------------
  4140. Traceback (most recent call last):
  4141. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4142. self.test(*self.arg)
  4143. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4144. gpuval = f(img, kern)
  4145. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4146. storage_map=self.fn.storage_map)
  4147. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4148. outputs = self.fn()
  4149. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4150. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4151. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af400710>, Scalar(float32)]
  4152. Inputs shapes: [(1, 2, 2, 2), (1, 1, 8, 3), (1, 2, 9, 4), 'No shapes', ()]
  4153. Inputs strides: [(0, 4, 2, 1), (0, 0, 6, 2), (0, 36, 4, 1), 'No strides', ()]
  4154. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee866e8>, 1.0]
  4155.  
  4156. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4157. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4158.  
  4159. ======================================================================
  4160. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 91, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4161. ----------------------------------------------------------------------
  4162. Traceback (most recent call last):
  4163. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4164. self.test(*self.arg)
  4165. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4166. gpuval = f(img, kern)
  4167. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4168. storage_map=self.fn.storage_map)
  4169. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4170. outputs = self.fn()
  4171. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4172. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4173. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85aa829150>, Scalar(float32)]
  4174. Inputs shapes: [(1, 2, 2, 2), (3, 1, 8, 3), (3, 2, 9, 4), 'No shapes', ()]
  4175. Inputs strides: [(0, 4, 2, 1), (48, 0, 6, 2), (72, 36, 4, 1), 'No strides', ()]
  4176. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86990>, 1.0]
  4177.  
  4178. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4179. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4180.  
  4181. ======================================================================
  4182. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 92, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4183. ----------------------------------------------------------------------
  4184. Traceback (most recent call last):
  4185. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4186. self.test(*self.arg)
  4187. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4188. gpuval = f(img, kern)
  4189. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4190. storage_map=self.fn.storage_map)
  4191. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4192. outputs = self.fn()
  4193. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4194. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4195. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af22c9d0>, Scalar(float32)]
  4196. Inputs shapes: [(2, 1, 2, 2), (3, 2, 8, 3), (3, 1, 9, 4), 'No shapes', ()]
  4197. Inputs strides: [(4, 0, 2, 1), (96, 48, 6, 2), (36, 0, 4, 1), 'No strides', ()]
  4198. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86c10>, 1.0]
  4199.  
  4200. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4201. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4202.  
  4203. ======================================================================
  4204. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 93, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4205. ----------------------------------------------------------------------
  4206. Traceback (most recent call last):
  4207. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4208. self.test(*self.arg)
  4209. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4210. gpuval = f(img, kern)
  4211. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4212. storage_map=self.fn.storage_map)
  4213. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4214. outputs = self.fn()
  4215. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4216. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4217. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e07b3890>, Scalar(float32)]
  4218. Inputs shapes: [(2, 2, 2, 2), (1, 2, 8, 3), (1, 2, 9, 4), 'No shapes', ()]
  4219. Inputs strides: [(8, 4, 2, 1), (0, 48, 6, 2), (0, 36, 4, 1), 'No strides', ()]
  4220. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86e90>, 1.0]
  4221.  
  4222. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4223. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4224.  
  4225. ======================================================================
  4226. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 94, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4227. ----------------------------------------------------------------------
  4228. Traceback (most recent call last):
  4229. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4230. self.test(*self.arg)
  4231. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4232. gpuval = f(img, kern)
  4233. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4234. storage_map=self.fn.storage_map)
  4235. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4236. outputs = self.fn()
  4237. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4238. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4239. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d8997a90>, Scalar(float32)]
  4240. Inputs shapes: [(2, 2, 2, 2), (2, 2, 8, 3), (2, 2, 9, 4), 'No shapes', ()]
  4241. Inputs strides: [(8, 4, 2, 1), (96, 48, 6, 2), (72, 36, 4, 1), 'No strides', ()]
  4242. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac194170>, 1.0]
  4243.  
  4244. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4245. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4246.  
  4247. ======================================================================
  4248. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 95, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4249. ----------------------------------------------------------------------
  4250. Traceback (most recent call last):
  4251. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4252. self.test(*self.arg)
  4253. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4254. gpuval = f(img, kern)
  4255. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4256. storage_map=self.fn.storage_map)
  4257. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4258. outputs = self.fn()
  4259. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4260. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4261. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af2bfe50>, Scalar(float32)]
  4262. Inputs shapes: [(2, 4, 2, 2), (3, 2, 8, 3), (3, 4, 9, 4), 'No shapes', ()]
  4263. Inputs strides: [(16, 4, 2, 1), (96, 48, 6, 2), (144, 36, 4, 1), 'No strides', ()]
  4264. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe3f0>, 1.0]
  4265.  
  4266. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4267. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4268.  
  4269. ======================================================================
  4270. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 96, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4271. ----------------------------------------------------------------------
  4272. Traceback (most recent call last):
  4273. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4274. self.test(*self.arg)
  4275. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4276. gpuval = f(img, kern)
  4277. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4278. storage_map=self.fn.storage_map)
  4279. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4280. outputs = self.fn()
  4281. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4282. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4283. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac1575d0>, Scalar(float32)]
  4284. Inputs shapes: [(2, 1, 2, 2), (1, 2, 6, 4), (1, 1, 7, 5), 'No shapes', ()]
  4285. Inputs strides: [(4, 0, 2, 1), (0, 48, 8, 2), (0, 0, 5, 1), 'No strides', ()]
  4286. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe620>, 1.0]
  4287.  
  4288. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4289. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4290.  
  4291. ======================================================================
  4292. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 97, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4293. ----------------------------------------------------------------------
  4294. Traceback (most recent call last):
  4295. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4296. self.test(*self.arg)
  4297. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4298. gpuval = f(img, kern)
  4299. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4300. storage_map=self.fn.storage_map)
  4301. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4302. outputs = self.fn()
  4303. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4304. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4305. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dc286310>, Scalar(float32)]
  4306. Inputs shapes: [(1, 1, 2, 2), (3, 1, 6, 4), (3, 1, 7, 5), 'No shapes', ()]
  4307. Inputs strides: [(0, 0, 2, 1), (48, 0, 8, 2), (35, 0, 5, 1), 'No strides', ()]
  4308. Inputs values: [<CudaNdarray object at 0x7f85af338bf0>, 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe8a0>, 1.0]
  4309.  
  4310. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4311. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4312.  
  4313. ======================================================================
  4314. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 98, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4315. ----------------------------------------------------------------------
  4316. Traceback (most recent call last):
  4317. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4318. self.test(*self.arg)
  4319. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4320. gpuval = f(img, kern)
  4321. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4322. storage_map=self.fn.storage_map)
  4323. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4324. outputs = self.fn()
  4325. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4326. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4327. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85aa5e9d10>, Scalar(float32)]
  4328. Inputs shapes: [(1, 2, 2, 2), (1, 1, 6, 4), (1, 2, 7, 5), 'No shapes', ()]
  4329. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 2), (0, 35, 5, 1), 'No strides', ()]
  4330. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6feb20>, 1.0]
  4331.  
  4332. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4333. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4334.  
  4335. ======================================================================
  4336. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 99, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4337. ----------------------------------------------------------------------
  4338. Traceback (most recent call last):
  4339. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4340. self.test(*self.arg)
  4341. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4342. gpuval = f(img, kern)
  4343. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4344. storage_map=self.fn.storage_map)
  4345. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4346. outputs = self.fn()
  4347. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4348. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4349. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e07b3110>, Scalar(float32)]
  4350. Inputs shapes: [(1, 2, 2, 2), (3, 1, 6, 4), (3, 2, 7, 5), 'No shapes', ()]
  4351. Inputs strides: [(0, 4, 2, 1), (48, 0, 8, 2), (70, 35, 5, 1), 'No strides', ()]
  4352. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fedc8>, 1.0]
  4353.  
  4354. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4355. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4356.  
  4357. ======================================================================
  4358. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 100, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4359. ----------------------------------------------------------------------
  4360. Traceback (most recent call last):
  4361. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4362. self.test(*self.arg)
  4363. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4364. gpuval = f(img, kern)
  4365. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4366. storage_map=self.fn.storage_map)
  4367. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4368. outputs = self.fn()
  4369. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4370. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4371. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c73398d0>, Scalar(float32)]
  4372. Inputs shapes: [(2, 1, 2, 2), (3, 2, 6, 4), (3, 1, 7, 5), 'No shapes', ()]
  4373. Inputs strides: [(4, 0, 2, 1), (96, 48, 8, 2), (35, 0, 5, 1), 'No strides', ()]
  4374. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85aa5d8080>, 1.0]
  4375.  
  4376. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4377. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4378.  
  4379. ======================================================================
  4380. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 101, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4381. ----------------------------------------------------------------------
  4382. Traceback (most recent call last):
  4383. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4384. self.test(*self.arg)
  4385. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4386. gpuval = f(img, kern)
  4387. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4388. storage_map=self.fn.storage_map)
  4389. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4390. outputs = self.fn()
  4391. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4392. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4393. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c7511cd0>, Scalar(float32)]
  4394. Inputs shapes: [(2, 2, 2, 2), (1, 2, 6, 4), (1, 2, 7, 5), 'No shapes', ()]
  4395. Inputs strides: [(8, 4, 2, 1), (0, 48, 8, 2), (0, 35, 5, 1), 'No strides', ()]
  4396. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac171300>, 1.0]
  4397.  
  4398. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4399. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4400.  
  4401. ======================================================================
  4402. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 102, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4403. ----------------------------------------------------------------------
  4404. Traceback (most recent call last):
  4405. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4406. self.test(*self.arg)
  4407. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4408. gpuval = f(img, kern)
  4409. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4410. storage_map=self.fn.storage_map)
  4411. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4412. outputs = self.fn()
  4413. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4414. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4415. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85abcdc0d0>, Scalar(float32)]
  4416. Inputs shapes: [(2, 2, 2, 2), (2, 2, 6, 4), (2, 2, 7, 5), 'No shapes', ()]
  4417. Inputs strides: [(8, 4, 2, 1), (96, 48, 8, 2), (70, 35, 5, 1), 'No strides', ()]
  4418. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe620>, 1.0]
  4419.  
  4420. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4421. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4422.  
  4423. ======================================================================
  4424. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 103, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4425. ----------------------------------------------------------------------
  4426. Traceback (most recent call last):
  4427. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4428. self.test(*self.arg)
  4429. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4430. gpuval = f(img, kern)
  4431. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4432. storage_map=self.fn.storage_map)
  4433. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4434. outputs = self.fn()
  4435. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4436. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4437. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af338550>, Scalar(float32)]
  4438. Inputs shapes: [(2, 4, 2, 2), (3, 2, 6, 4), (3, 4, 7, 5), 'No shapes', ()]
  4439. Inputs strides: [(16, 4, 2, 1), (96, 48, 8, 2), (140, 35, 5, 1), 'No strides', ()]
  4440. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe0f8>, 1.0]
  4441.  
  4442. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4443. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4444.  
  4445. ======================================================================
  4446. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 104, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4447. ----------------------------------------------------------------------
  4448. Traceback (most recent call last):
  4449. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4450. self.test(*self.arg)
  4451. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4452. gpuval = f(img, kern)
  4453. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4454. storage_map=self.fn.storage_map)
  4455. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4456. outputs = self.fn()
  4457. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4458. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4459. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85aa5e9a50>, Scalar(float32)]
  4460. Inputs shapes: [(2, 1, 3, 2), (1, 2, 8, 4), (1, 1, 10, 5), 'No shapes', ()]
  4461. Inputs strides: [(6, 0, 2, 1), (0, 64, 8, 2), (0, 0, 5, 1), 'No strides', ()]
  4462. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86b70>, 1.0]
  4463.  
  4464. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4465. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4466.  
  4467. ======================================================================
  4468. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 105, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4469. ----------------------------------------------------------------------
  4470. Traceback (most recent call last):
  4471. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4472. self.test(*self.arg)
  4473. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4474. gpuval = f(img, kern)
  4475. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4476. storage_map=self.fn.storage_map)
  4477. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4478. outputs = self.fn()
  4479. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4480. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4481. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcd710d0>, Scalar(float32)]
  4482. Inputs shapes: [(1, 1, 3, 2), (3, 1, 8, 4), (3, 1, 10, 5), 'No shapes', ()]
  4483. Inputs strides: [(0, 0, 2, 1), (64, 0, 8, 2), (50, 0, 5, 1), 'No strides', ()]
  4484. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86670>, 1.0]
  4485.  
  4486. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4487. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4488.  
  4489. ======================================================================
  4490. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 106, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4491. ----------------------------------------------------------------------
  4492. Traceback (most recent call last):
  4493. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4494. self.test(*self.arg)
  4495. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4496. gpuval = f(img, kern)
  4497. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4498. storage_map=self.fn.storage_map)
  4499. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4500. outputs = self.fn()
  4501. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4502. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4503. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d8901410>, Scalar(float32)]
  4504. Inputs shapes: [(1, 2, 3, 2), (1, 1, 8, 4), (1, 2, 10, 5), 'No shapes', ()]
  4505. Inputs strides: [(0, 6, 2, 1), (0, 0, 8, 2), (0, 50, 5, 1), 'No strides', ()]
  4506. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86170>, 1.0]
  4507.  
  4508. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4509. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4510.  
  4511. ======================================================================
  4512. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 107, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4513. ----------------------------------------------------------------------
  4514. Traceback (most recent call last):
  4515. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4516. self.test(*self.arg)
  4517. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4518. gpuval = f(img, kern)
  4519. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4520. storage_map=self.fn.storage_map)
  4521. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4522. outputs = self.fn()
  4523. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4524. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4525. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d9697210>, Scalar(float32)]
  4526. Inputs shapes: [(1, 2, 3, 2), (3, 1, 8, 4), (3, 2, 10, 5), 'No shapes', ()]
  4527. Inputs strides: [(0, 6, 2, 1), (64, 0, 8, 2), (100, 50, 5, 1), 'No strides', ()]
  4528. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0c60>, 1.0]
  4529.  
  4530. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4531. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4532.  
  4533. ======================================================================
  4534. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 108, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4535. ----------------------------------------------------------------------
  4536. Traceback (most recent call last):
  4537. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4538. self.test(*self.arg)
  4539. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4540. gpuval = f(img, kern)
  4541. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4542. storage_map=self.fn.storage_map)
  4543. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4544. outputs = self.fn()
  4545. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4546. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4547. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ab6a5d10>, Scalar(float32)]
  4548. Inputs shapes: [(2, 1, 3, 2), (3, 2, 8, 4), (3, 1, 10, 5), 'No shapes', ()]
  4549. Inputs strides: [(6, 0, 2, 1), (128, 64, 8, 2), (50, 0, 5, 1), 'No strides', ()]
  4550. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0760>, 1.0]
  4551.  
  4552. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4553. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4554.  
  4555. ======================================================================
  4556. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 109, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4557. ----------------------------------------------------------------------
  4558. Traceback (most recent call last):
  4559. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4560. self.test(*self.arg)
  4561. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4562. gpuval = f(img, kern)
  4563. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4564. storage_map=self.fn.storage_map)
  4565. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4566. outputs = self.fn()
  4567. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4568. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4569. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af111550>, Scalar(float32)]
  4570. Inputs shapes: [(2, 2, 3, 2), (1, 2, 8, 4), (1, 2, 10, 5), 'No shapes', ()]
  4571. Inputs strides: [(12, 6, 2, 1), (0, 64, 8, 2), (0, 50, 5, 1), 'No strides', ()]
  4572. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0260>, 1.0]
  4573.  
  4574. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4575. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4576.  
  4577. ======================================================================
  4578. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 110, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4579. ----------------------------------------------------------------------
  4580. Traceback (most recent call last):
  4581. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4582. self.test(*self.arg)
  4583. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4584. gpuval = f(img, kern)
  4585. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4586. storage_map=self.fn.storage_map)
  4587. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4588. outputs = self.fn()
  4589. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4590. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4591. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dbdfa410>, Scalar(float32)]
  4592. Inputs shapes: [(2, 2, 3, 2), (2, 2, 8, 4), (2, 2, 10, 5), 'No shapes', ()]
  4593. Inputs strides: [(12, 6, 2, 1), (128, 64, 8, 2), (100, 50, 5, 1), 'No strides', ()]
  4594. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0d78>, 1.0]
  4595.  
  4596. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4597. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4598.  
  4599. ======================================================================
  4600. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 111, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4601. ----------------------------------------------------------------------
  4602. Traceback (most recent call last):
  4603. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4604. self.test(*self.arg)
  4605. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4606. gpuval = f(img, kern)
  4607. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4608. storage_map=self.fn.storage_map)
  4609. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4610. outputs = self.fn()
  4611. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4612. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4613. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73dd9d0>, Scalar(float32)]
  4614. Inputs shapes: [(2, 4, 3, 2), (3, 2, 8, 4), (3, 4, 10, 5), 'No shapes', ()]
  4615. Inputs strides: [(24, 6, 2, 1), (128, 64, 8, 2), (200, 50, 5, 1), 'No strides', ()]
  4616. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0850>, 1.0]
  4617.  
  4618. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4619. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4620.  
  4621. ======================================================================
  4622. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 112, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4623. ----------------------------------------------------------------------
  4624. Traceback (most recent call last):
  4625. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4626. self.test(*self.arg)
  4627. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4628. gpuval = f(img, kern)
  4629. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4630. storage_map=self.fn.storage_map)
  4631. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4632. outputs = self.fn()
  4633. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4634. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4635. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac1630d0>, Scalar(float32)]
  4636. Inputs shapes: [(2, 1, 2, 3), (1, 2, 8, 4), (1, 1, 9, 6), 'No shapes', ()]
  4637. Inputs strides: [(6, 0, 3, 1), (0, 64, 8, 2), (0, 0, 6, 1), 'No strides', ()]
  4638. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0300>, 1.0]
  4639.  
  4640. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4641. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4642.  
  4643. ======================================================================
  4644. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 113, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4645. ----------------------------------------------------------------------
  4646. Traceback (most recent call last):
  4647. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4648. self.test(*self.arg)
  4649. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4650. gpuval = f(img, kern)
  4651. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4652. storage_map=self.fn.storage_map)
  4653. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4654. outputs = self.fn()
  4655. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4656. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4657. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e0728f50>, Scalar(float32)]
  4658. Inputs shapes: [(1, 1, 2, 3), (3, 1, 8, 4), (3, 1, 9, 6), 'No shapes', ()]
  4659. Inputs strides: [(0, 0, 3, 1), (64, 0, 8, 2), (54, 0, 6, 1), 'No strides', ()]
  4660. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dcd22fd0>, 1.0]
  4661.  
  4662. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4663. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4664.  
  4665. ======================================================================
  4666. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 114, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4667. ----------------------------------------------------------------------
  4668. Traceback (most recent call last):
  4669. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4670. self.test(*self.arg)
  4671. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4672. gpuval = f(img, kern)
  4673. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4674. storage_map=self.fn.storage_map)
  4675. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4676. outputs = self.fn()
  4677. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4678. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4679. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ac1638d0>, Scalar(float32)]
  4680. Inputs shapes: [(1, 2, 2, 3), (1, 1, 8, 4), (1, 2, 9, 6), 'No shapes', ()]
  4681. Inputs strides: [(0, 6, 3, 1), (0, 0, 8, 2), (0, 54, 6, 1), 'No strides', ()]
  4682. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d76e8>, 1.0]
  4683.  
  4684. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4685. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4686.  
  4687. ======================================================================
  4688. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 115, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4689. ----------------------------------------------------------------------
  4690. Traceback (most recent call last):
  4691. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4692. self.test(*self.arg)
  4693. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4694. gpuval = f(img, kern)
  4695. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4696. storage_map=self.fn.storage_map)
  4697. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4698. outputs = self.fn()
  4699. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4700. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4701. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ab69be10>, Scalar(float32)]
  4702. Inputs shapes: [(1, 2, 2, 3), (3, 1, 8, 4), (3, 2, 9, 6), 'No shapes', ()]
  4703. Inputs strides: [(0, 6, 3, 1), (64, 0, 8, 2), (108, 54, 6, 1), 'No strides', ()]
  4704. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d79e0>, 1.0]
  4705.  
  4706. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4707. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4708.  
  4709. ======================================================================
  4710. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 116, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4711. ----------------------------------------------------------------------
  4712. Traceback (most recent call last):
  4713. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4714. self.test(*self.arg)
  4715. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4716. gpuval = f(img, kern)
  4717. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4718. storage_map=self.fn.storage_map)
  4719. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4720. outputs = self.fn()
  4721. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4722. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4723. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c743a6d0>, Scalar(float32)]
  4724. Inputs shapes: [(2, 1, 2, 3), (3, 2, 8, 4), (3, 1, 9, 6), 'No shapes', ()]
  4725. Inputs strides: [(6, 0, 3, 1), (128, 64, 8, 2), (54, 0, 6, 1), 'No strides', ()]
  4726. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d7670>, 1.0]
  4727.  
  4728. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4729. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4730.  
  4731. ======================================================================
  4732. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 117, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4733. ----------------------------------------------------------------------
  4734. Traceback (most recent call last):
  4735. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4736. self.test(*self.arg)
  4737. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4738. gpuval = f(img, kern)
  4739. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4740. storage_map=self.fn.storage_map)
  4741. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4742. outputs = self.fn()
  4743. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4744. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4745. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dc2a9550>, Scalar(float32)]
  4746. Inputs shapes: [(2, 2, 2, 3), (1, 2, 8, 4), (1, 2, 9, 6), 'No shapes', ()]
  4747. Inputs strides: [(12, 6, 3, 1), (0, 64, 8, 2), (0, 54, 6, 1), 'No strides', ()]
  4748. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96d7170>, 1.0]
  4749.  
  4750. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4751. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4752.  
  4753. ======================================================================
  4754. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 118, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4755. ----------------------------------------------------------------------
  4756. Traceback (most recent call last):
  4757. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4758. self.test(*self.arg)
  4759. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4760. gpuval = f(img, kern)
  4761. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4762. storage_map=self.fn.storage_map)
  4763. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4764. outputs = self.fn()
  4765. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4766. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4767. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85deea8ed0>, Scalar(float32)]
  4768. Inputs shapes: [(2, 2, 2, 3), (2, 2, 8, 4), (2, 2, 9, 6), 'No shapes', ()]
  4769. Inputs strides: [(12, 6, 3, 1), (128, 64, 8, 2), (108, 54, 6, 1), 'No strides', ()]
  4770. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85df329850>, 1.0]
  4771.  
  4772. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4773. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4774.  
  4775. ======================================================================
  4776. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (1, 2), (1, 1), -1, 0, True, None, 119, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4777. ----------------------------------------------------------------------
  4778. Traceback (most recent call last):
  4779. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4780. self.test(*self.arg)
  4781. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4782. gpuval = f(img, kern)
  4783. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4784. storage_map=self.fn.storage_map)
  4785. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4786. outputs = self.fn()
  4787. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4788. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4789. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c5a4fd10>, Scalar(float32)]
  4790. Inputs shapes: [(2, 4, 2, 3), (3, 2, 8, 4), (3, 4, 9, 6), 'No shapes', ()]
  4791. Inputs strides: [(24, 6, 3, 1), (128, 64, 8, 2), (216, 54, 6, 1), 'No strides', ()]
  4792. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85df329580>, 1.0]
  4793.  
  4794. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4795. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4796.  
  4797. ======================================================================
  4798. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 120, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4799. ----------------------------------------------------------------------
  4800. Traceback (most recent call last):
  4801. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4802. self.test(*self.arg)
  4803. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4804. gpuval = f(img, kern)
  4805. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4806. storage_map=self.fn.storage_map)
  4807. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4808. outputs = self.fn()
  4809. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4810. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4811. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, True, False)), CudaNdarrayType(float32, (True, True, True, False)), <theano.gof.type.CDataType object at 0x7f85deea4d10>, Scalar(float32)]
  4812. Inputs shapes: [(2, 1, 1, 1), (1, 2, 1, 2), (1, 1, 1, 2), 'No shapes', ()]
  4813. Inputs strides: [(1, 0, 0, 0), (0, 4, 0, 1), (0, 0, 0, 1), 'No strides', ()]
  4814. Inputs values: [<CudaNdarray object at 0x7f85dc2c04b0>, <CudaNdarray object at 0x7f85dbb86f30>, <CudaNdarray object at 0x7f85e07d6230>, <PyCObject object at 0x7f85df329530>, 1.0]
  4815.  
  4816. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4817. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4818.  
  4819. ======================================================================
  4820. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 124, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4821. ----------------------------------------------------------------------
  4822. Traceback (most recent call last):
  4823. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4824. self.test(*self.arg)
  4825. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4826. gpuval = f(img, kern)
  4827. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4828. storage_map=self.fn.storage_map)
  4829. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4830. outputs = self.fn()
  4831. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4832. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4833. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, True, True, False)), <theano.gof.type.CDataType object at 0x7f85abcca950>, Scalar(float32)]
  4834. Inputs shapes: [(2, 1, 1, 1), (3, 2, 1, 2), (3, 1, 1, 2), 'No shapes', ()]
  4835. Inputs strides: [(1, 0, 0, 0), (8, 4, 0, 1), (2, 0, 0, 1), 'No strides', ()]
  4836. Inputs values: [<CudaNdarray object at 0x7f85ddb65b70>, 'not shown', 'not shown', <PyCObject object at 0x7f85df3295f8>, 1.0]
  4837.  
  4838. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4839. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4840.  
  4841. ======================================================================
  4842. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 125, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4843. ----------------------------------------------------------------------
  4844. Traceback (most recent call last):
  4845. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4846. self.test(*self.arg)
  4847. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4848. gpuval = f(img, kern)
  4849. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4850. storage_map=self.fn.storage_map)
  4851. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4852. outputs = self.fn()
  4853. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4854. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4855. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, True, False)), CudaNdarrayType(float32, (True, False, True, False)), <theano.gof.type.CDataType object at 0x7f85abd54190>, Scalar(float32)]
  4856. Inputs shapes: [(2, 2, 1, 1), (1, 2, 1, 2), (1, 2, 1, 2), 'No shapes', ()]
  4857. Inputs strides: [(2, 1, 0, 0), (0, 4, 0, 1), (0, 2, 0, 1), 'No strides', ()]
  4858. Inputs values: [<CudaNdarray object at 0x7f85e6c3fa70>, <CudaNdarray object at 0x7f85ac5ab370>, <CudaNdarray object at 0x7f85e6c3fb70>, <PyCObject object at 0x7f85df329198>, 1.0]
  4859.  
  4860. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4861. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4862.  
  4863. ======================================================================
  4864. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 126, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4865. ----------------------------------------------------------------------
  4866. Traceback (most recent call last):
  4867. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4868. self.test(*self.arg)
  4869. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4870. gpuval = f(img, kern)
  4871. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4872. storage_map=self.fn.storage_map)
  4873. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4874. outputs = self.fn()
  4875. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4876. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4877. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, False, True, False)), <theano.gof.type.CDataType object at 0x7f85dee97810>, Scalar(float32)]
  4878. Inputs shapes: [(2, 2, 1, 1), (2, 2, 1, 2), (2, 2, 1, 2), 'No shapes', ()]
  4879. Inputs strides: [(2, 1, 0, 0), (8, 4, 0, 1), (4, 2, 0, 1), 'No strides', ()]
  4880. Inputs values: [<CudaNdarray object at 0x7f85d940d9b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85df329080>, 1.0]
  4881.  
  4882. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4883. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4884.  
  4885. ======================================================================
  4886. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 127, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4887. ----------------------------------------------------------------------
  4888. Traceback (most recent call last):
  4889. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4890. self.test(*self.arg)
  4891. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4892. gpuval = f(img, kern)
  4893. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4894. storage_map=self.fn.storage_map)
  4895. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4896. outputs = self.fn()
  4897. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4898. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4899. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, False)), CudaNdarrayType(float32, (False, False, True, False)), <theano.gof.type.CDataType object at 0x7f85ac14c210>, Scalar(float32)]
  4900. Inputs shapes: [(2, 4, 1, 1), (3, 2, 1, 2), (3, 4, 1, 2), 'No shapes', ()]
  4901. Inputs strides: [(4, 1, 0, 0), (8, 4, 0, 1), (8, 2, 0, 1), 'No strides', ()]
  4902. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a940>, 1.0]
  4903.  
  4904. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4905. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4906.  
  4907. ======================================================================
  4908. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 128, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4909. ----------------------------------------------------------------------
  4910. Traceback (most recent call last):
  4911. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4912. self.test(*self.arg)
  4913. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4914. gpuval = f(img, kern)
  4915. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4916. storage_map=self.fn.storage_map)
  4917. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4918. outputs = self.fn()
  4919. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4920. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4921. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c749fb10>, Scalar(float32)]
  4922. Inputs shapes: [(2, 1, 2, 2), (1, 2, 2, 4), (1, 1, 3, 5), 'No shapes', ()]
  4923. Inputs strides: [(4, 0, 2, 1), (0, 16, 8, 1), (0, 0, 5, 1), 'No strides', ()]
  4924. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a800>, 1.0]
  4925.  
  4926. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4927. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4928.  
  4929. ======================================================================
  4930. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 129, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4931. ----------------------------------------------------------------------
  4932. Traceback (most recent call last):
  4933. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4934. self.test(*self.arg)
  4935. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4936. gpuval = f(img, kern)
  4937. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4938. storage_map=self.fn.storage_map)
  4939. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4940. outputs = self.fn()
  4941. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4942. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4943. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85df34bb50>, Scalar(float32)]
  4944. Inputs shapes: [(1, 1, 2, 2), (3, 1, 2, 4), (3, 1, 3, 5), 'No shapes', ()]
  4945. Inputs strides: [(0, 0, 2, 1), (16, 0, 8, 1), (15, 0, 5, 1), 'No strides', ()]
  4946. Inputs values: [<CudaNdarray object at 0x7f85ddb2b530>, 'not shown', 'not shown', <PyCObject object at 0x7f85a973a580>, 1.0]
  4947.  
  4948. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4949. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4950.  
  4951. ======================================================================
  4952. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 130, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4953. ----------------------------------------------------------------------
  4954. Traceback (most recent call last):
  4955. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4956. self.test(*self.arg)
  4957. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4958. gpuval = f(img, kern)
  4959. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4960. storage_map=self.fn.storage_map)
  4961. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4962. outputs = self.fn()
  4963. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4964. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4965. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d93f1910>, Scalar(float32)]
  4966. Inputs shapes: [(1, 2, 2, 2), (1, 1, 2, 4), (1, 2, 3, 5), 'No shapes', ()]
  4967. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 1), (0, 15, 5, 1), 'No strides', ()]
  4968. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973aa58>, 1.0]
  4969.  
  4970. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4971. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4972.  
  4973. ======================================================================
  4974. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 131, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4975. ----------------------------------------------------------------------
  4976. Traceback (most recent call last):
  4977. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  4978. self.test(*self.arg)
  4979. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  4980. gpuval = f(img, kern)
  4981. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  4982. storage_map=self.fn.storage_map)
  4983. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  4984. outputs = self.fn()
  4985. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  4986. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  4987. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e7de6dd0>, Scalar(float32)]
  4988. Inputs shapes: [(1, 2, 2, 2), (3, 1, 2, 4), (3, 2, 3, 5), 'No shapes', ()]
  4989. Inputs strides: [(0, 4, 2, 1), (16, 0, 8, 1), (30, 15, 5, 1), 'No strides', ()]
  4990. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1850>, 1.0]
  4991.  
  4992. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  4993. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  4994.  
  4995. ======================================================================
  4996. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 132, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  4997. ----------------------------------------------------------------------
  4998. Traceback (most recent call last):
  4999. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5000. self.test(*self.arg)
  5001. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5002. gpuval = f(img, kern)
  5003. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5004. storage_map=self.fn.storage_map)
  5005. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5006. outputs = self.fn()
  5007. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5008. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5009. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dbb87850>, Scalar(float32)]
  5010. Inputs shapes: [(2, 1, 2, 2), (3, 2, 2, 4), (3, 1, 3, 5), 'No shapes', ()]
  5011. Inputs strides: [(4, 0, 2, 1), (32, 16, 8, 1), (15, 0, 5, 1), 'No strides', ()]
  5012. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1a08>, 1.0]
  5013.  
  5014. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5015. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5016.  
  5017. ======================================================================
  5018. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 133, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5019. ----------------------------------------------------------------------
  5020. Traceback (most recent call last):
  5021. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5022. self.test(*self.arg)
  5023. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5024. gpuval = f(img, kern)
  5025. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5026. storage_map=self.fn.storage_map)
  5027. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5028. outputs = self.fn()
  5029. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5030. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5031. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d941aed0>, Scalar(float32)]
  5032. Inputs shapes: [(2, 2, 2, 2), (1, 2, 2, 4), (1, 2, 3, 5), 'No shapes', ()]
  5033. Inputs strides: [(8, 4, 2, 1), (0, 16, 8, 1), (0, 15, 5, 1), 'No strides', ()]
  5034. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf10a8>, 1.0]
  5035.  
  5036. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5037. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5038.  
  5039. ======================================================================
  5040. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 134, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5041. ----------------------------------------------------------------------
  5042. Traceback (most recent call last):
  5043. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5044. self.test(*self.arg)
  5045. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5046. gpuval = f(img, kern)
  5047. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5048. storage_map=self.fn.storage_map)
  5049. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5050. outputs = self.fn()
  5051. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5052. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5053. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73ca090>, Scalar(float32)]
  5054. Inputs shapes: [(2, 2, 2, 2), (2, 2, 2, 4), (2, 2, 3, 5), 'No shapes', ()]
  5055. Inputs strides: [(8, 4, 2, 1), (32, 16, 8, 1), (30, 15, 5, 1), 'No strides', ()]
  5056. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4328>, 1.0]
  5057.  
  5058. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5059. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5060.  
  5061. ======================================================================
  5062. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 135, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5063. ----------------------------------------------------------------------
  5064. Traceback (most recent call last):
  5065. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5066. self.test(*self.arg)
  5067. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5068. gpuval = f(img, kern)
  5069. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5070. storage_map=self.fn.storage_map)
  5071. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5072. outputs = self.fn()
  5073. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5074. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5075. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85aa5ce4d0>, Scalar(float32)]
  5076. Inputs shapes: [(2, 4, 2, 2), (3, 2, 2, 4), (3, 4, 3, 5), 'No shapes', ()]
  5077. Inputs strides: [(16, 4, 2, 1), (32, 16, 8, 1), (60, 15, 5, 1), 'No strides', ()]
  5078. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4440>, 1.0]
  5079.  
  5080. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5081. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5082.  
  5083. ======================================================================
  5084. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 136, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5085. ----------------------------------------------------------------------
  5086. Traceback (most recent call last):
  5087. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5088. self.test(*self.arg)
  5089. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5090. gpuval = f(img, kern)
  5091. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5092. storage_map=self.fn.storage_map)
  5093. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5094. outputs = self.fn()
  5095. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5096. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5097. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac368e50>, Scalar(float32)]
  5098. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 6), (1, 1, 4, 7), 'No shapes', ()]
  5099. Inputs strides: [(4, 0, 2, 1), (0, 36, 12, 1), (0, 0, 7, 1), 'No strides', ()]
  5100. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4620>, 1.0]
  5101.  
  5102. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5103. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5104.  
  5105. ======================================================================
  5106. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 137, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5107. ----------------------------------------------------------------------
  5108. Traceback (most recent call last):
  5109. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5110. self.test(*self.arg)
  5111. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5112. gpuval = f(img, kern)
  5113. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5114. storage_map=self.fn.storage_map)
  5115. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5116. outputs = self.fn()
  5117. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5118. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5119. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af334290>, Scalar(float32)]
  5120. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 6), (3, 1, 4, 7), 'No shapes', ()]
  5121. Inputs strides: [(0, 0, 2, 1), (36, 0, 12, 1), (28, 0, 7, 1), 'No strides', ()]
  5122. Inputs values: [<CudaNdarray object at 0x7f85d9676d70>, 'not shown', 'not shown', <PyCObject object at 0x7f85dee121c0>, 1.0]
  5123.  
  5124. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5125. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5126.  
  5127. ======================================================================
  5128. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 138, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5129. ----------------------------------------------------------------------
  5130. Traceback (most recent call last):
  5131. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5132. self.test(*self.arg)
  5133. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5134. gpuval = f(img, kern)
  5135. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5136. storage_map=self.fn.storage_map)
  5137. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5138. outputs = self.fn()
  5139. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5140. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5141. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d940d690>, Scalar(float32)]
  5142. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 6), (1, 2, 4, 7), 'No shapes', ()]
  5143. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 1), (0, 28, 7, 1), 'No strides', ()]
  5144. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee12440>, 1.0]
  5145.  
  5146. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5147. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5148.  
  5149. ======================================================================
  5150. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 139, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5151. ----------------------------------------------------------------------
  5152. Traceback (most recent call last):
  5153. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5154. self.test(*self.arg)
  5155. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5156. gpuval = f(img, kern)
  5157. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5158. storage_map=self.fn.storage_map)
  5159. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5160. outputs = self.fn()
  5161. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5162. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5163. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85a9ef4dd0>, Scalar(float32)]
  5164. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 6), (3, 2, 4, 7), 'No shapes', ()]
  5165. Inputs strides: [(0, 4, 2, 1), (36, 0, 12, 1), (56, 28, 7, 1), 'No strides', ()]
  5166. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614170>, 1.0]
  5167.  
  5168. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5169. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5170.  
  5171. ======================================================================
  5172. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 140, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5173. ----------------------------------------------------------------------
  5174. Traceback (most recent call last):
  5175. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5176. self.test(*self.arg)
  5177. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5178. gpuval = f(img, kern)
  5179. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5180. storage_map=self.fn.storage_map)
  5181. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5182. outputs = self.fn()
  5183. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5184. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5185. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e0a3b690>, Scalar(float32)]
  5186. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 6), (3, 1, 4, 7), 'No shapes', ()]
  5187. Inputs strides: [(4, 0, 2, 1), (72, 36, 12, 1), (28, 0, 7, 1), 'No strides', ()]
  5188. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614530>, 1.0]
  5189.  
  5190. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5191. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5192.  
  5193. ======================================================================
  5194. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 141, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5195. ----------------------------------------------------------------------
  5196. Traceback (most recent call last):
  5197. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5198. self.test(*self.arg)
  5199. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5200. gpuval = f(img, kern)
  5201. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5202. storage_map=self.fn.storage_map)
  5203. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5204. outputs = self.fn()
  5205. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5206. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5207. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c735f050>, Scalar(float32)]
  5208. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 6), (1, 2, 4, 7), 'No shapes', ()]
  5209. Inputs strides: [(8, 4, 2, 1), (0, 36, 12, 1), (0, 28, 7, 1), 'No strides', ()]
  5210. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614210>, 1.0]
  5211.  
  5212. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5213. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5214.  
  5215. ======================================================================
  5216. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 142, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5217. ----------------------------------------------------------------------
  5218. Traceback (most recent call last):
  5219. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5220. self.test(*self.arg)
  5221. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5222. gpuval = f(img, kern)
  5223. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5224. storage_map=self.fn.storage_map)
  5225. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5226. outputs = self.fn()
  5227. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5228. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5229. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dede7250>, Scalar(float32)]
  5230. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 6), (2, 2, 4, 7), 'No shapes', ()]
  5231. Inputs strides: [(8, 4, 2, 1), (72, 36, 12, 1), (56, 28, 7, 1), 'No strides', ()]
  5232. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614300>, 1.0]
  5233.  
  5234. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5235. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5236.  
  5237. ======================================================================
  5238. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 143, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5239. ----------------------------------------------------------------------
  5240. Traceback (most recent call last):
  5241. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5242. self.test(*self.arg)
  5243. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5244. gpuval = f(img, kern)
  5245. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5246. storage_map=self.fn.storage_map)
  5247. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5248. outputs = self.fn()
  5249. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5250. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5251. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac7111d0>, Scalar(float32)]
  5252. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 6), (3, 4, 4, 7), 'No shapes', ()]
  5253. Inputs strides: [(16, 4, 2, 1), (72, 36, 12, 1), (112, 28, 7, 1), 'No strides', ()]
  5254. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433080>, 1.0]
  5255.  
  5256. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5257. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5258.  
  5259. ======================================================================
  5260. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 144, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5261. ----------------------------------------------------------------------
  5262. Traceback (most recent call last):
  5263. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5264. self.test(*self.arg)
  5265. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5266. gpuval = f(img, kern)
  5267. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5268. storage_map=self.fn.storage_map)
  5269. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5270. outputs = self.fn()
  5271. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5272. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5273. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85a93dcd10>, Scalar(float32)]
  5274. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 6), (1, 1, 5, 7), 'No shapes', ()]
  5275. Inputs strides: [(4, 0, 2, 1), (0, 48, 12, 1), (0, 0, 7, 1), 'No strides', ()]
  5276. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94332b0>, 1.0]
  5277.  
  5278. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5279. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5280.  
  5281. ======================================================================
  5282. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 145, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5283. ----------------------------------------------------------------------
  5284. Traceback (most recent call last):
  5285. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5286. self.test(*self.arg)
  5287. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5288. gpuval = f(img, kern)
  5289. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5290. storage_map=self.fn.storage_map)
  5291. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5292. outputs = self.fn()
  5293. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5294. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5295. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c72f2bd0>, Scalar(float32)]
  5296. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 6), (3, 1, 5, 7), 'No shapes', ()]
  5297. Inputs strides: [(0, 0, 2, 1), (48, 0, 12, 1), (35, 0, 7, 1), 'No strides', ()]
  5298. Inputs values: [<CudaNdarray object at 0x7f85e6c34af0>, 'not shown', 'not shown', <PyCObject object at 0x7f85a9433558>, 1.0]
  5299.  
  5300. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5301. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5302.  
  5303. ======================================================================
  5304. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 146, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5305. ----------------------------------------------------------------------
  5306. Traceback (most recent call last):
  5307. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5308. self.test(*self.arg)
  5309. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5310. gpuval = f(img, kern)
  5311. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5312. storage_map=self.fn.storage_map)
  5313. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5314. outputs = self.fn()
  5315. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5316. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5317. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85df0f3d90>, Scalar(float32)]
  5318. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 6), (1, 2, 5, 7), 'No shapes', ()]
  5319. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 1), (0, 35, 7, 1), 'No strides', ()]
  5320. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94334e0>, 1.0]
  5321.  
  5322. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5323. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5324.  
  5325. ======================================================================
  5326. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 147, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5327. ----------------------------------------------------------------------
  5328. Traceback (most recent call last):
  5329. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5330. self.test(*self.arg)
  5331. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5332. gpuval = f(img, kern)
  5333. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5334. storage_map=self.fn.storage_map)
  5335. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5336. outputs = self.fn()
  5337. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5338. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5339. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af2d55d0>, Scalar(float32)]
  5340. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 6), (3, 2, 5, 7), 'No shapes', ()]
  5341. Inputs strides: [(0, 4, 2, 1), (48, 0, 12, 1), (70, 35, 7, 1), 'No strides', ()]
  5342. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94335d0>, 1.0]
  5343.  
  5344. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5345. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5346.  
  5347. ======================================================================
  5348. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 148, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5349. ----------------------------------------------------------------------
  5350. Traceback (most recent call last):
  5351. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5352. self.test(*self.arg)
  5353. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5354. gpuval = f(img, kern)
  5355. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5356. storage_map=self.fn.storage_map)
  5357. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5358. outputs = self.fn()
  5359. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5360. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5361. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dd01b610>, Scalar(float32)]
  5362. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 6), (3, 1, 5, 7), 'No shapes', ()]
  5363. Inputs strides: [(4, 0, 2, 1), (96, 48, 12, 1), (35, 0, 7, 1), 'No strides', ()]
  5364. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e7ce2968>, 1.0]
  5365.  
  5366. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5367. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5368.  
  5369. ======================================================================
  5370. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 149, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5371. ----------------------------------------------------------------------
  5372. Traceback (most recent call last):
  5373. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5374. self.test(*self.arg)
  5375. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5376. gpuval = f(img, kern)
  5377. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5378. storage_map=self.fn.storage_map)
  5379. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5380. outputs = self.fn()
  5381. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5382. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5383. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af40d290>, Scalar(float32)]
  5384. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 6), (1, 2, 5, 7), 'No shapes', ()]
  5385. Inputs strides: [(8, 4, 2, 1), (0, 48, 12, 1), (0, 35, 7, 1), 'No strides', ()]
  5386. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85e6c481c0>, 1.0]
  5387.  
  5388. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5389. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5390.  
  5391. ======================================================================
  5392. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 150, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5393. ----------------------------------------------------------------------
  5394. Traceback (most recent call last):
  5395. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5396. self.test(*self.arg)
  5397. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5398. gpuval = f(img, kern)
  5399. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5400. storage_map=self.fn.storage_map)
  5401. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5402. outputs = self.fn()
  5403. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5404. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5405. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85df5ecc90>, Scalar(float32)]
  5406. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 6), (2, 2, 5, 7), 'No shapes', ()]
  5407. Inputs strides: [(8, 4, 2, 1), (96, 48, 12, 1), (70, 35, 7, 1), 'No strides', ()]
  5408. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ab668468>, 1.0]
  5409.  
  5410. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5411. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5412.  
  5413. ======================================================================
  5414. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 151, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5415. ----------------------------------------------------------------------
  5416. Traceback (most recent call last):
  5417. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5418. self.test(*self.arg)
  5419. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5420. gpuval = f(img, kern)
  5421. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5422. storage_map=self.fn.storage_map)
  5423. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5424. outputs = self.fn()
  5425. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5426. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5427. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85af2d5a90>, Scalar(float32)]
  5428. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 6), (3, 4, 5, 7), 'No shapes', ()]
  5429. Inputs strides: [(16, 4, 2, 1), (96, 48, 12, 1), (140, 35, 7, 1), 'No strides', ()]
  5430. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ab6686e8>, 1.0]
  5431.  
  5432. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5433. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5434.  
  5435. ======================================================================
  5436. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 152, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5437. ----------------------------------------------------------------------
  5438. Traceback (most recent call last):
  5439. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5440. self.test(*self.arg)
  5441. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5442. gpuval = f(img, kern)
  5443. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5444. storage_map=self.fn.storage_map)
  5445. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5446. outputs = self.fn()
  5447. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5448. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5449. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abfdb8d0>, Scalar(float32)]
  5450. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 8), (1, 1, 4, 9), 'No shapes', ()]
  5451. Inputs strides: [(4, 0, 2, 1), (0, 48, 16, 1), (0, 0, 9, 1), 'No strides', ()]
  5452. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ab668918>, 1.0]
  5453.  
  5454. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5455. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5456.  
  5457. ======================================================================
  5458. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 153, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5459. ----------------------------------------------------------------------
  5460. Traceback (most recent call last):
  5461. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5462. self.test(*self.arg)
  5463. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5464. gpuval = f(img, kern)
  5465. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5466. storage_map=self.fn.storage_map)
  5467. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5468. outputs = self.fn()
  5469. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5470. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5471. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dbb87c10>, Scalar(float32)]
  5472. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 8), (3, 1, 4, 9), 'No shapes', ()]
  5473. Inputs strides: [(0, 0, 2, 1), (48, 0, 16, 1), (36, 0, 9, 1), 'No strides', ()]
  5474. Inputs values: [<CudaNdarray object at 0x7f85ab6a24f0>, 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fed00>, 1.0]
  5475.  
  5476. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5477. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5478.  
  5479. ======================================================================
  5480. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 154, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5481. ----------------------------------------------------------------------
  5482. Traceback (most recent call last):
  5483. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5484. self.test(*self.arg)
  5485. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5486. gpuval = f(img, kern)
  5487. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5488. storage_map=self.fn.storage_map)
  5489. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5490. outputs = self.fn()
  5491. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5492. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5493. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dc2a9950>, Scalar(float32)]
  5494. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 8), (1, 2, 4, 9), 'No shapes', ()]
  5495. Inputs strides: [(0, 4, 2, 1), (0, 0, 16, 1), (0, 36, 9, 1), 'No strides', ()]
  5496. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe800>, 1.0]
  5497.  
  5498. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5499. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5500.  
  5501. ======================================================================
  5502. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 155, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5503. ----------------------------------------------------------------------
  5504. Traceback (most recent call last):
  5505. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5506. self.test(*self.arg)
  5507. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5508. gpuval = f(img, kern)
  5509. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5510. storage_map=self.fn.storage_map)
  5511. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5512. outputs = self.fn()
  5513. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5514. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5515. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ab69b2d0>, Scalar(float32)]
  5516. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 8), (3, 2, 4, 9), 'No shapes', ()]
  5517. Inputs strides: [(0, 4, 2, 1), (48, 0, 16, 1), (72, 36, 9, 1), 'No strides', ()]
  5518. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe288>, 1.0]
  5519.  
  5520. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5521. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5522.  
  5523. ======================================================================
  5524. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 156, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5525. ----------------------------------------------------------------------
  5526. Traceback (most recent call last):
  5527. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5528. self.test(*self.arg)
  5529. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5530. gpuval = f(img, kern)
  5531. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5532. storage_map=self.fn.storage_map)
  5533. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5534. outputs = self.fn()
  5535. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5536. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5537. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddb45250>, Scalar(float32)]
  5538. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 8), (3, 1, 4, 9), 'No shapes', ()]
  5539. Inputs strides: [(4, 0, 2, 1), (96, 48, 16, 1), (36, 0, 9, 1), 'No strides', ()]
  5540. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86080>, 1.0]
  5541.  
  5542. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5543. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5544.  
  5545. ======================================================================
  5546. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 157, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5547. ----------------------------------------------------------------------
  5548. Traceback (most recent call last):
  5549. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5550. self.test(*self.arg)
  5551. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5552. gpuval = f(img, kern)
  5553. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5554. storage_map=self.fn.storage_map)
  5555. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5556. outputs = self.fn()
  5557. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5558. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5559. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af244450>, Scalar(float32)]
  5560. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 8), (1, 2, 4, 9), 'No shapes', ()]
  5561. Inputs strides: [(8, 4, 2, 1), (0, 48, 16, 1), (0, 36, 9, 1), 'No strides', ()]
  5562. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86af8>, 1.0]
  5563.  
  5564. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5565. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5566.  
  5567. ======================================================================
  5568. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 158, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5569. ----------------------------------------------------------------------
  5570. Traceback (most recent call last):
  5571. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5572. self.test(*self.arg)
  5573. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5574. gpuval = f(img, kern)
  5575. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5576. storage_map=self.fn.storage_map)
  5577. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5578. outputs = self.fn()
  5579. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5580. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5581. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85abe3a110>, Scalar(float32)]
  5582. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 8), (2, 2, 4, 9), 'No shapes', ()]
  5583. Inputs strides: [(8, 4, 2, 1), (96, 48, 16, 1), (72, 36, 9, 1), 'No strides', ()]
  5584. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86d28>, 1.0]
  5585.  
  5586. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5587. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5588.  
  5589. ======================================================================
  5590. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 159, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5591. ----------------------------------------------------------------------
  5592. Traceback (most recent call last):
  5593. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5594. self.test(*self.arg)
  5595. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5596. gpuval = f(img, kern)
  5597. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5598. storage_map=self.fn.storage_map)
  5599. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5600. outputs = self.fn()
  5601. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5602. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5603. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7424cd0>, Scalar(float32)]
  5604. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 8), (3, 4, 4, 9), 'No shapes', ()]
  5605. Inputs strides: [(16, 4, 2, 1), (96, 48, 16, 1), (144, 36, 9, 1), 'No strides', ()]
  5606. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86990>, 1.0]
  5607.  
  5608. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5609. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5610.  
  5611. ======================================================================
  5612. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 160, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5613. ----------------------------------------------------------------------
  5614. Traceback (most recent call last):
  5615. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5616. self.test(*self.arg)
  5617. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5618. gpuval = f(img, kern)
  5619. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5620. storage_map=self.fn.storage_map)
  5621. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5622. outputs = self.fn()
  5623. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5624. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5625. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dee36e10>, Scalar(float32)]
  5626. Inputs shapes: [(2, 1, 3, 2), (1, 2, 4, 8), (1, 1, 6, 9), 'No shapes', ()]
  5627. Inputs strides: [(6, 0, 2, 1), (0, 64, 16, 1), (0, 0, 9, 1), 'No strides', ()]
  5628. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86288>, 1.0]
  5629.  
  5630. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5631. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5632.  
  5633. ======================================================================
  5634. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 161, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5635. ----------------------------------------------------------------------
  5636. Traceback (most recent call last):
  5637. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5638. self.test(*self.arg)
  5639. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5640. gpuval = f(img, kern)
  5641. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5642. storage_map=self.fn.storage_map)
  5643. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5644. outputs = self.fn()
  5645. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5646. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5647. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d967c610>, Scalar(float32)]
  5648. Inputs shapes: [(1, 1, 3, 2), (3, 1, 4, 8), (3, 1, 6, 9), 'No shapes', ()]
  5649. Inputs strides: [(0, 0, 2, 1), (64, 0, 16, 1), (54, 0, 9, 1), 'No strides', ()]
  5650. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0648>, 1.0]
  5651.  
  5652. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5653. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5654.  
  5655. ======================================================================
  5656. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 162, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5657. ----------------------------------------------------------------------
  5658. Traceback (most recent call last):
  5659. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5660. self.test(*self.arg)
  5661. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5662. gpuval = f(img, kern)
  5663. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5664. storage_map=self.fn.storage_map)
  5665. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5666. outputs = self.fn()
  5667. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5668. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5669. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ddadfbd0>, Scalar(float32)]
  5670. Inputs shapes: [(1, 2, 3, 2), (1, 1, 4, 8), (1, 2, 6, 9), 'No shapes', ()]
  5671. Inputs strides: [(0, 6, 2, 1), (0, 0, 16, 1), (0, 54, 9, 1), 'No strides', ()]
  5672. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0e90>, 1.0]
  5673.  
  5674. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5675. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5676.  
  5677. ======================================================================
  5678. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 163, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5679. ----------------------------------------------------------------------
  5680. Traceback (most recent call last):
  5681. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5682. self.test(*self.arg)
  5683. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5684. gpuval = f(img, kern)
  5685. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5686. storage_map=self.fn.storage_map)
  5687. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5688. outputs = self.fn()
  5689. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5690. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5691. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d88e2f90>, Scalar(float32)]
  5692. Inputs shapes: [(1, 2, 3, 2), (3, 1, 4, 8), (3, 2, 6, 9), 'No shapes', ()]
  5693. Inputs strides: [(0, 6, 2, 1), (64, 0, 16, 1), (108, 54, 9, 1), 'No strides', ()]
  5694. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc09b8>, 1.0]
  5695.  
  5696. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5697. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5698.  
  5699. ======================================================================
  5700. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 164, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5701. ----------------------------------------------------------------------
  5702. Traceback (most recent call last):
  5703. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5704. self.test(*self.arg)
  5705. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5706. gpuval = f(img, kern)
  5707. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5708. storage_map=self.fn.storage_map)
  5709. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5710. outputs = self.fn()
  5711. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5712. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5713. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dcd3ea90>, Scalar(float32)]
  5714. Inputs shapes: [(2, 1, 3, 2), (3, 2, 4, 8), (3, 1, 6, 9), 'No shapes', ()]
  5715. Inputs strides: [(6, 0, 2, 1), (128, 64, 16, 1), (54, 0, 9, 1), 'No strides', ()]
  5716. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0418>, 1.0]
  5717.  
  5718. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5719. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5720.  
  5721. ======================================================================
  5722. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 165, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5723. ----------------------------------------------------------------------
  5724. Traceback (most recent call last):
  5725. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5726. self.test(*self.arg)
  5727. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5728. gpuval = f(img, kern)
  5729. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5730. storage_map=self.fn.storage_map)
  5731. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5732. outputs = self.fn()
  5733. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5734. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5735. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85a99b4d90>, Scalar(float32)]
  5736. Inputs shapes: [(2, 2, 3, 2), (1, 2, 4, 8), (1, 2, 6, 9), 'No shapes', ()]
  5737. Inputs strides: [(12, 6, 2, 1), (0, 64, 16, 1), (0, 54, 9, 1), 'No strides', ()]
  5738. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0120>, 1.0]
  5739.  
  5740. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5741. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5742.  
  5743. ======================================================================
  5744. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 166, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5745. ----------------------------------------------------------------------
  5746. Traceback (most recent call last):
  5747. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5748. self.test(*self.arg)
  5749. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5750. gpuval = f(img, kern)
  5751. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5752. storage_map=self.fn.storage_map)
  5753. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5754. outputs = self.fn()
  5755. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5756. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5757. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d9469f90>, Scalar(float32)]
  5758. Inputs shapes: [(2, 2, 3, 2), (2, 2, 4, 8), (2, 2, 6, 9), 'No shapes', ()]
  5759. Inputs strides: [(12, 6, 2, 1), (128, 64, 16, 1), (108, 54, 9, 1), 'No strides', ()]
  5760. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0c38>, 1.0]
  5761.  
  5762. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5763. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5764.  
  5765. ======================================================================
  5766. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 167, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5767. ----------------------------------------------------------------------
  5768. Traceback (most recent call last):
  5769. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5770. self.test(*self.arg)
  5771. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5772. gpuval = f(img, kern)
  5773. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5774. storage_map=self.fn.storage_map)
  5775. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5776. outputs = self.fn()
  5777. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5778. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5779. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d946b910>, Scalar(float32)]
  5780. Inputs shapes: [(2, 4, 3, 2), (3, 2, 4, 8), (3, 4, 6, 9), 'No shapes', ()]
  5781. Inputs strides: [(24, 6, 2, 1), (128, 64, 16, 1), (216, 54, 9, 1), 'No strides', ()]
  5782. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0f30>, 1.0]
  5783.  
  5784. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5785. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5786.  
  5787. ======================================================================
  5788. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 168, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5789. ----------------------------------------------------------------------
  5790. Traceback (most recent call last):
  5791. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5792. self.test(*self.arg)
  5793. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5794. gpuval = f(img, kern)
  5795. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5796. storage_map=self.fn.storage_map)
  5797. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5798. outputs = self.fn()
  5799. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5800. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5801. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e659f490>, Scalar(float32)]
  5802. Inputs shapes: [(2, 1, 2, 3), (1, 2, 4, 8), (1, 1, 5, 10), 'No shapes', ()]
  5803. Inputs strides: [(6, 0, 3, 1), (0, 64, 16, 1), (0, 0, 10, 1), 'No strides', ()]
  5804. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0a08>, 1.0]
  5805.  
  5806. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5807. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5808.  
  5809. ======================================================================
  5810. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 169, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5811. ----------------------------------------------------------------------
  5812. Traceback (most recent call last):
  5813. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5814. self.test(*self.arg)
  5815. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5816. gpuval = f(img, kern)
  5817. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5818. storage_map=self.fn.storage_map)
  5819. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5820. outputs = self.fn()
  5821. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5822. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5823. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c7485c90>, Scalar(float32)]
  5824. Inputs shapes: [(1, 1, 2, 3), (3, 1, 4, 8), (3, 1, 5, 10), 'No shapes', ()]
  5825. Inputs strides: [(0, 0, 3, 1), (64, 0, 16, 1), (50, 0, 10, 1), 'No strides', ()]
  5826. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e05a8>, 1.0]
  5827.  
  5828. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5829. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5830.  
  5831. ======================================================================
  5832. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 170, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5833. ----------------------------------------------------------------------
  5834. Traceback (most recent call last):
  5835. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5836. self.test(*self.arg)
  5837. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5838. gpuval = f(img, kern)
  5839. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5840. storage_map=self.fn.storage_map)
  5841. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5842. outputs = self.fn()
  5843. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5844. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5845. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dc2910d0>, Scalar(float32)]
  5846. Inputs shapes: [(1, 2, 2, 3), (1, 1, 4, 8), (1, 2, 5, 10), 'No shapes', ()]
  5847. Inputs strides: [(0, 6, 3, 1), (0, 0, 16, 1), (0, 50, 10, 1), 'No strides', ()]
  5848. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e00f8>, 1.0]
  5849.  
  5850. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5851. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5852.  
  5853. ======================================================================
  5854. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 171, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5855. ----------------------------------------------------------------------
  5856. Traceback (most recent call last):
  5857. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5858. self.test(*self.arg)
  5859. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5860. gpuval = f(img, kern)
  5861. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5862. storage_map=self.fn.storage_map)
  5863. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5864. outputs = self.fn()
  5865. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5866. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5867. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d94604d0>, Scalar(float32)]
  5868. Inputs shapes: [(1, 2, 2, 3), (3, 1, 4, 8), (3, 2, 5, 10), 'No shapes', ()]
  5869. Inputs strides: [(0, 6, 3, 1), (64, 0, 16, 1), (100, 50, 10, 1), 'No strides', ()]
  5870. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433a30>, 1.0]
  5871.  
  5872. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5873. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5874.  
  5875. ======================================================================
  5876. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 172, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5877. ----------------------------------------------------------------------
  5878. Traceback (most recent call last):
  5879. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5880. self.test(*self.arg)
  5881. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5882. gpuval = f(img, kern)
  5883. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5884. storage_map=self.fn.storage_map)
  5885. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5886. outputs = self.fn()
  5887. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5888. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5889. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c749fd90>, Scalar(float32)]
  5890. Inputs shapes: [(2, 1, 2, 3), (3, 2, 4, 8), (3, 1, 5, 10), 'No shapes', ()]
  5891. Inputs strides: [(6, 0, 3, 1), (128, 64, 16, 1), (50, 0, 10, 1), 'No strides', ()]
  5892. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433cb0>, 1.0]
  5893.  
  5894. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5895. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5896.  
  5897. ======================================================================
  5898. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 173, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5899. ----------------------------------------------------------------------
  5900. Traceback (most recent call last):
  5901. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5902. self.test(*self.arg)
  5903. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5904. gpuval = f(img, kern)
  5905. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5906. storage_map=self.fn.storage_map)
  5907. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5908. outputs = self.fn()
  5909. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5910. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5911. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85abd24c10>, Scalar(float32)]
  5912. Inputs shapes: [(2, 2, 2, 3), (1, 2, 4, 8), (1, 2, 5, 10), 'No shapes', ()]
  5913. Inputs strides: [(12, 6, 3, 1), (0, 64, 16, 1), (0, 50, 10, 1), 'No strides', ()]
  5914. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433508>, 1.0]
  5915.  
  5916. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5917. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5918.  
  5919. ======================================================================
  5920. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 174, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5921. ----------------------------------------------------------------------
  5922. Traceback (most recent call last):
  5923. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5924. self.test(*self.arg)
  5925. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5926. gpuval = f(img, kern)
  5927. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5928. storage_map=self.fn.storage_map)
  5929. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5930. outputs = self.fn()
  5931. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5932. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5933. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d8984a10>, Scalar(float32)]
  5934. Inputs shapes: [(2, 2, 2, 3), (2, 2, 4, 8), (2, 2, 5, 10), 'No shapes', ()]
  5935. Inputs strides: [(12, 6, 3, 1), (128, 64, 16, 1), (100, 50, 10, 1), 'No strides', ()]
  5936. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433558>, 1.0]
  5937.  
  5938. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5939. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5940.  
  5941. ======================================================================
  5942. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (2, 1), (1, 1), -1, 0, True, None, 175, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5943. ----------------------------------------------------------------------
  5944. Traceback (most recent call last):
  5945. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5946. self.test(*self.arg)
  5947. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5948. gpuval = f(img, kern)
  5949. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5950. storage_map=self.fn.storage_map)
  5951. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5952. outputs = self.fn()
  5953. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5954. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5955. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7485ed0>, Scalar(float32)]
  5956. Inputs shapes: [(2, 4, 2, 3), (3, 2, 4, 8), (3, 4, 5, 10), 'No shapes', ()]
  5957. Inputs strides: [(24, 6, 3, 1), (128, 64, 16, 1), (200, 50, 10, 1), 'No strides', ()]
  5958. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94332d8>, 1.0]
  5959.  
  5960. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5961. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5962.  
  5963. ======================================================================
  5964. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 176, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5965. ----------------------------------------------------------------------
  5966. Traceback (most recent call last):
  5967. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5968. self.test(*self.arg)
  5969. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5970. gpuval = f(img, kern)
  5971. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5972. storage_map=self.fn.storage_map)
  5973. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5974. outputs = self.fn()
  5975. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5976. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5977. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, True, True, True)), <theano.gof.type.CDataType object at 0x7f85af399c10>, Scalar(float32)]
  5978. Inputs shapes: [(2, 1, 1, 1), (1, 2, 1, 1), (1, 1, 1, 1), 'No shapes', ()]
  5979. Inputs strides: [(1, 0, 0, 0), (0, 4, 0, 0), (0, 0, 0, 0), 'No strides', ()]
  5980. Inputs values: [<CudaNdarray object at 0x7f85ac6f6e70>, <CudaNdarray object at 0x7f85df38a5f0>, <CudaNdarray object at 0x7f85ac6f6370>, <PyCObject object at 0x7f85a9433850>, 1.0]
  5981.  
  5982. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  5983. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  5984.  
  5985. ======================================================================
  5986. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (1, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 180, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  5987. ----------------------------------------------------------------------
  5988. Traceback (most recent call last):
  5989. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  5990. self.test(*self.arg)
  5991. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  5992. gpuval = f(img, kern)
  5993. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  5994. storage_map=self.fn.storage_map)
  5995. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  5996. outputs = self.fn()
  5997. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  5998. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  5999. Inputs types: [CudaNdarrayType(float32, (False, True, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, True, True, True)), <theano.gof.type.CDataType object at 0x7f85d968d690>, Scalar(float32)]
  6000. Inputs shapes: [(2, 1, 1, 1), (3, 2, 1, 1), (3, 1, 1, 1), 'No shapes', ()]
  6001. Inputs strides: [(1, 0, 0, 0), (8, 4, 0, 0), (1, 0, 0, 0), 'No strides', ()]
  6002. Inputs values: [<CudaNdarray object at 0x7f85df0e04f0>, 'not shown', <CudaNdarray object at 0x7f85dcff4770>, <PyCObject object at 0x7f85a9433788>, 1.0]
  6003.  
  6004. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6005. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6006.  
  6007. ======================================================================
  6008. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 181, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6009. ----------------------------------------------------------------------
  6010. Traceback (most recent call last):
  6011. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6012. self.test(*self.arg)
  6013. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6014. gpuval = f(img, kern)
  6015. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6016. storage_map=self.fn.storage_map)
  6017. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6018. outputs = self.fn()
  6019. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6020. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6021. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (True, False, True, True)), CudaNdarrayType(float32, (True, False, True, True)), <theano.gof.type.CDataType object at 0x7f85dee42550>, Scalar(float32)]
  6022. Inputs shapes: [(2, 2, 1, 1), (1, 2, 1, 1), (1, 2, 1, 1), 'No shapes', ()]
  6023. Inputs strides: [(2, 1, 0, 0), (0, 4, 0, 0), (0, 1, 0, 0), 'No strides', ()]
  6024. Inputs values: [<CudaNdarray object at 0x7f85dcff4770>, <CudaNdarray object at 0x7f85ac6f6230>, <CudaNdarray object at 0x7f85c7339530>, <PyCObject object at 0x7f85a9433aa8>, 1.0]
  6025.  
  6026. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6027. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6028.  
  6029. ======================================================================
  6030. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 2, 2), (2, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 182, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6031. ----------------------------------------------------------------------
  6032. Traceback (most recent call last):
  6033. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6034. self.test(*self.arg)
  6035. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6036. gpuval = f(img, kern)
  6037. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6038. storage_map=self.fn.storage_map)
  6039. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6040. outputs = self.fn()
  6041. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6042. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6043. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), <theano.gof.type.CDataType object at 0x7f85c75b5290>, Scalar(float32)]
  6044. Inputs shapes: [(2, 2, 1, 1), (2, 2, 1, 1), (2, 2, 1, 1), 'No shapes', ()]
  6045. Inputs strides: [(2, 1, 0, 0), (8, 4, 0, 0), (2, 1, 0, 0), 'No strides', ()]
  6046. Inputs values: [<CudaNdarray object at 0x7f85e079c830>, <CudaNdarray object at 0x7f85c901e8b0>, <CudaNdarray object at 0x7f85e079ceb0>, <PyCObject object at 0x7f85a9433da0>, 1.0]
  6047.  
  6048. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6049. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6050.  
  6051. ======================================================================
  6052. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 2, 2), (4, 2, 1, 1), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 183, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6053. ----------------------------------------------------------------------
  6054. Traceback (most recent call last):
  6055. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6056. self.test(*self.arg)
  6057. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6058. gpuval = f(img, kern)
  6059. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6060. storage_map=self.fn.storage_map)
  6061. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6062. outputs = self.fn()
  6063. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6064. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6065. Inputs types: [CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), CudaNdarrayType(float32, (False, False, True, True)), <theano.gof.type.CDataType object at 0x7f85a94211d0>, Scalar(float32)]
  6066. Inputs shapes: [(2, 4, 1, 1), (3, 2, 1, 1), (3, 4, 1, 1), 'No shapes', ()]
  6067. Inputs strides: [(4, 1, 0, 0), (8, 4, 0, 0), (4, 1, 0, 0), 'No strides', ()]
  6068. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433788>, 1.0]
  6069.  
  6070. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6071. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6072.  
  6073. ======================================================================
  6074. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 184, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6075. ----------------------------------------------------------------------
  6076. Traceback (most recent call last):
  6077. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6078. self.test(*self.arg)
  6079. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6080. gpuval = f(img, kern)
  6081. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6082. storage_map=self.fn.storage_map)
  6083. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6084. outputs = self.fn()
  6085. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6086. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6087. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abe52590>, Scalar(float32)]
  6088. Inputs shapes: [(2, 1, 2, 2), (1, 2, 2, 2), (1, 1, 3, 3), 'No shapes', ()]
  6089. Inputs strides: [(4, 0, 2, 1), (0, 16, 8, 2), (0, 0, 3, 1), 'No strides', ()]
  6090. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433f58>, 1.0]
  6091.  
  6092. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6093. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6094.  
  6095. ======================================================================
  6096. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 185, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6097. ----------------------------------------------------------------------
  6098. Traceback (most recent call last):
  6099. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6100. self.test(*self.arg)
  6101. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6102. gpuval = f(img, kern)
  6103. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6104. storage_map=self.fn.storage_map)
  6105. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6106. outputs = self.fn()
  6107. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6108. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6109. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c5a4f5d0>, Scalar(float32)]
  6110. Inputs shapes: [(1, 1, 2, 2), (3, 1, 2, 2), (3, 1, 3, 3), 'No shapes', ()]
  6111. Inputs strides: [(0, 0, 2, 1), (16, 0, 8, 2), (9, 0, 3, 1), 'No strides', ()]
  6112. Inputs values: [<CudaNdarray object at 0x7f85d8946d70>, 'not shown', 'not shown', <PyCObject object at 0x7f85a9614300>, 1.0]
  6113.  
  6114. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6115. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6116.  
  6117. ======================================================================
  6118. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 186, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6119. ----------------------------------------------------------------------
  6120. Traceback (most recent call last):
  6121. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6122. self.test(*self.arg)
  6123. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6124. gpuval = f(img, kern)
  6125. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6126. storage_map=self.fn.storage_map)
  6127. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6128. outputs = self.fn()
  6129. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6130. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6131. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c901e4d0>, Scalar(float32)]
  6132. Inputs shapes: [(1, 2, 2, 2), (1, 1, 2, 2), (1, 2, 3, 3), 'No shapes', ()]
  6133. Inputs strides: [(0, 4, 2, 1), (0, 0, 8, 2), (0, 9, 3, 1), 'No strides', ()]
  6134. Inputs values: ['not shown', <CudaNdarray object at 0x7f85c7571570>, 'not shown', <PyCObject object at 0x7f85a9614a08>, 1.0]
  6135.  
  6136. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6137. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6138.  
  6139. ======================================================================
  6140. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 4, 4), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 187, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6141. ----------------------------------------------------------------------
  6142. Traceback (most recent call last):
  6143. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6144. self.test(*self.arg)
  6145. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6146. gpuval = f(img, kern)
  6147. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6148. storage_map=self.fn.storage_map)
  6149. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6150. outputs = self.fn()
  6151. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6152. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6153. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e07c5d90>, Scalar(float32)]
  6154. Inputs shapes: [(1, 2, 2, 2), (3, 1, 2, 2), (3, 2, 3, 3), 'No shapes', ()]
  6155. Inputs strides: [(0, 4, 2, 1), (16, 0, 8, 2), (18, 9, 3, 1), 'No strides', ()]
  6156. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614e40>, 1.0]
  6157.  
  6158. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6159. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6160.  
  6161. ======================================================================
  6162. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 188, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6163. ----------------------------------------------------------------------
  6164. Traceback (most recent call last):
  6165. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6166. self.test(*self.arg)
  6167. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6168. gpuval = f(img, kern)
  6169. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6170. storage_map=self.fn.storage_map)
  6171. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6172. outputs = self.fn()
  6173. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6174. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6175. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c72edcd0>, Scalar(float32)]
  6176. Inputs shapes: [(2, 1, 2, 2), (3, 2, 2, 2), (3, 1, 3, 3), 'No shapes', ()]
  6177. Inputs strides: [(4, 0, 2, 1), (32, 16, 8, 2), (9, 0, 3, 1), 'No strides', ()]
  6178. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614be8>, 1.0]
  6179.  
  6180. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6181. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6182.  
  6183. ======================================================================
  6184. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 189, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6185. ----------------------------------------------------------------------
  6186. Traceback (most recent call last):
  6187. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6188. self.test(*self.arg)
  6189. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6190. gpuval = f(img, kern)
  6191. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6192. storage_map=self.fn.storage_map)
  6193. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6194. outputs = self.fn()
  6195. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6196. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6197. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85e06c1110>, Scalar(float32)]
  6198. Inputs shapes: [(2, 2, 2, 2), (1, 2, 2, 2), (1, 2, 3, 3), 'No shapes', ()]
  6199. Inputs strides: [(8, 4, 2, 1), (0, 16, 8, 2), (0, 9, 3, 1), 'No strides', ()]
  6200. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614e90>, 1.0]
  6201.  
  6202. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6203. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6204.  
  6205. ======================================================================
  6206. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 4, 4), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 190, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6207. ----------------------------------------------------------------------
  6208. Traceback (most recent call last):
  6209. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6210. self.test(*self.arg)
  6211. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6212. gpuval = f(img, kern)
  6213. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6214. storage_map=self.fn.storage_map)
  6215. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6216. outputs = self.fn()
  6217. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6218. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6219. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85a9615990>, Scalar(float32)]
  6220. Inputs shapes: [(2, 2, 2, 2), (2, 2, 2, 2), (2, 2, 3, 3), 'No shapes', ()]
  6221. Inputs strides: [(8, 4, 2, 1), (32, 16, 8, 2), (18, 9, 3, 1), 'No strides', ()]
  6222. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614c60>, 1.0]
  6223.  
  6224. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6225. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6226.  
  6227. ======================================================================
  6228. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 4, 4), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 191, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6229. ----------------------------------------------------------------------
  6230. Traceback (most recent call last):
  6231. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6232. self.test(*self.arg)
  6233. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6234. gpuval = f(img, kern)
  6235. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6236. storage_map=self.fn.storage_map)
  6237. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6238. outputs = self.fn()
  6239. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6240. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6241. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac15eed0>, Scalar(float32)]
  6242. Inputs shapes: [(2, 4, 2, 2), (3, 2, 2, 2), (3, 4, 3, 3), 'No shapes', ()]
  6243. Inputs strides: [(16, 4, 2, 1), (32, 16, 8, 2), (36, 9, 3, 1), 'No strides', ()]
  6244. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee128a0>, 1.0]
  6245.  
  6246. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6247. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6248.  
  6249. ======================================================================
  6250. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 192, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6251. ----------------------------------------------------------------------
  6252. Traceback (most recent call last):
  6253. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6254. self.test(*self.arg)
  6255. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6256. gpuval = f(img, kern)
  6257. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6258. storage_map=self.fn.storage_map)
  6259. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6260. outputs = self.fn()
  6261. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6262. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6263. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85aa13f1d0>, Scalar(float32)]
  6264. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 3), (1, 1, 4, 4), 'No shapes', ()]
  6265. Inputs strides: [(4, 0, 2, 1), (0, 36, 12, 2), (0, 0, 4, 1), 'No strides', ()]
  6266. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee12c38>, 1.0]
  6267.  
  6268. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6269. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6270.  
  6271. ======================================================================
  6272. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 193, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6273. ----------------------------------------------------------------------
  6274. Traceback (most recent call last):
  6275. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6276. self.test(*self.arg)
  6277. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6278. gpuval = f(img, kern)
  6279. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6280. storage_map=self.fn.storage_map)
  6281. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6282. outputs = self.fn()
  6283. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6284. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6285. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85d93f19d0>, Scalar(float32)]
  6286. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 3), (3, 1, 4, 4), 'No shapes', ()]
  6287. Inputs strides: [(0, 0, 2, 1), (36, 0, 12, 2), (16, 0, 4, 1), 'No strides', ()]
  6288. Inputs values: [<CudaNdarray object at 0x7f85af2ec630>, 'not shown', 'not shown', <PyCObject object at 0x7f85ddae48c8>, 1.0]
  6289.  
  6290. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6291. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6292.  
  6293. ======================================================================
  6294. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 194, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6295. ----------------------------------------------------------------------
  6296. Traceback (most recent call last):
  6297. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6298. self.test(*self.arg)
  6299. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6300. gpuval = f(img, kern)
  6301. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6302. storage_map=self.fn.storage_map)
  6303. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6304. outputs = self.fn()
  6305. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6306. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6307. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85a93dc650>, Scalar(float32)]
  6308. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 3), (1, 2, 4, 4), 'No shapes', ()]
  6309. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 2), (0, 16, 4, 1), 'No strides', ()]
  6310. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4710>, 1.0]
  6311.  
  6312. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6313. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6314.  
  6315. ======================================================================
  6316. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 195, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6317. ----------------------------------------------------------------------
  6318. Traceback (most recent call last):
  6319. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6320. self.test(*self.arg)
  6321. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6322. gpuval = f(img, kern)
  6323. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6324. storage_map=self.fn.storage_map)
  6325. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6326. outputs = self.fn()
  6327. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6328. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6329. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac13e050>, Scalar(float32)]
  6330. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 3), (3, 2, 4, 4), 'No shapes', ()]
  6331. Inputs strides: [(0, 4, 2, 1), (36, 0, 12, 2), (32, 16, 4, 1), 'No strides', ()]
  6332. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae42d8>, 1.0]
  6333.  
  6334. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6335. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6336.  
  6337. ======================================================================
  6338. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 196, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6339. ----------------------------------------------------------------------
  6340. Traceback (most recent call last):
  6341. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6342. self.test(*self.arg)
  6343. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6344. gpuval = f(img, kern)
  6345. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6346. storage_map=self.fn.storage_map)
  6347. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6348. outputs = self.fn()
  6349. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6350. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6351. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c72f2bd0>, Scalar(float32)]
  6352. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 3), (3, 1, 4, 4), 'No shapes', ()]
  6353. Inputs strides: [(4, 0, 2, 1), (72, 36, 12, 2), (16, 0, 4, 1), 'No strides', ()]
  6354. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ddae45a8>, 1.0]
  6355.  
  6356. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6357. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6358.  
  6359. ======================================================================
  6360. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 197, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6361. ----------------------------------------------------------------------
  6362. Traceback (most recent call last):
  6363. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6364. self.test(*self.arg)
  6365. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6366. gpuval = f(img, kern)
  6367. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6368. storage_map=self.fn.storage_map)
  6369. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6370. outputs = self.fn()
  6371. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6372. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6373. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c72eda10>, Scalar(float32)]
  6374. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 3), (1, 2, 4, 4), 'No shapes', ()]
  6375. Inputs strides: [(8, 4, 2, 1), (0, 36, 12, 2), (0, 16, 4, 1), 'No strides', ()]
  6376. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1a30>, 1.0]
  6377.  
  6378. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6379. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6380.  
  6381. ======================================================================
  6382. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 198, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6383. ----------------------------------------------------------------------
  6384. Traceback (most recent call last):
  6385. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6386. self.test(*self.arg)
  6387. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6388. gpuval = f(img, kern)
  6389. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6390. storage_map=self.fn.storage_map)
  6391. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6392. outputs = self.fn()
  6393. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6394. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6395. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85df34b250>, Scalar(float32)]
  6396. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 3), (2, 2, 4, 4), 'No shapes', ()]
  6397. Inputs strides: [(8, 4, 2, 1), (72, 36, 12, 2), (32, 16, 4, 1), 'No strides', ()]
  6398. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1f08>, 1.0]
  6399.  
  6400. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6401. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6402.  
  6403. ======================================================================
  6404. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 199, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6405. ----------------------------------------------------------------------
  6406. Traceback (most recent call last):
  6407. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6408. self.test(*self.arg)
  6409. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6410. gpuval = f(img, kern)
  6411. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6412. storage_map=self.fn.storage_map)
  6413. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6414. outputs = self.fn()
  6415. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6416. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6417. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85a9ef4490>, Scalar(float32)]
  6418. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 3), (3, 4, 4, 4), 'No shapes', ()]
  6419. Inputs strides: [(16, 4, 2, 1), (72, 36, 12, 2), (64, 16, 4, 1), 'No strides', ()]
  6420. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1a80>, 1.0]
  6421.  
  6422. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6423. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6424.  
  6425. ======================================================================
  6426. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 200, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6427. ----------------------------------------------------------------------
  6428. Traceback (most recent call last):
  6429. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6430. self.test(*self.arg)
  6431. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6432. gpuval = f(img, kern)
  6433. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6434. storage_map=self.fn.storage_map)
  6435. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6436. outputs = self.fn()
  6437. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6438. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6439. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af40d1d0>, Scalar(float32)]
  6440. Inputs shapes: [(2, 1, 2, 2), (1, 2, 4, 3), (1, 1, 5, 4), 'No shapes', ()]
  6441. Inputs strides: [(4, 0, 2, 1), (0, 48, 12, 2), (0, 0, 4, 1), 'No strides', ()]
  6442. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1350>, 1.0]
  6443.  
  6444. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6445. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6446.  
  6447. ======================================================================
  6448. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 201, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6449. ----------------------------------------------------------------------
  6450. Traceback (most recent call last):
  6451. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6452. self.test(*self.arg)
  6453. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6454. gpuval = f(img, kern)
  6455. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6456. storage_map=self.fn.storage_map)
  6457. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6458. outputs = self.fn()
  6459. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6460. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6461. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85e6c01c50>, Scalar(float32)]
  6462. Inputs shapes: [(1, 1, 2, 2), (3, 1, 4, 3), (3, 1, 5, 4), 'No shapes', ()]
  6463. Inputs strides: [(0, 0, 2, 1), (48, 0, 12, 2), (20, 0, 4, 1), 'No strides', ()]
  6464. Inputs values: [<CudaNdarray object at 0x7f85ac74b1b0>, 'not shown', 'not shown', <PyCObject object at 0x7f85ddae4710>, 1.0]
  6465.  
  6466. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6467. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6468.  
  6469. ======================================================================
  6470. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 202, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6471. ----------------------------------------------------------------------
  6472. Traceback (most recent call last):
  6473. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6474. self.test(*self.arg)
  6475. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6476. gpuval = f(img, kern)
  6477. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6478. storage_map=self.fn.storage_map)
  6479. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6480. outputs = self.fn()
  6481. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6482. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6483. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85af334310>, Scalar(float32)]
  6484. Inputs shapes: [(1, 2, 2, 2), (1, 1, 4, 3), (1, 2, 5, 4), 'No shapes', ()]
  6485. Inputs strides: [(0, 4, 2, 1), (0, 0, 12, 2), (0, 20, 4, 1), 'No strides', ()]
  6486. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee122d8>, 1.0]
  6487.  
  6488. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6489. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6490.  
  6491. ======================================================================
  6492. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 6), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 203, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6493. ----------------------------------------------------------------------
  6494. Traceback (most recent call last):
  6495. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6496. self.test(*self.arg)
  6497. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6498. gpuval = f(img, kern)
  6499. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6500. storage_map=self.fn.storage_map)
  6501. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6502. outputs = self.fn()
  6503. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6504. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6505. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73caf50>, Scalar(float32)]
  6506. Inputs shapes: [(1, 2, 2, 2), (3, 1, 4, 3), (3, 2, 5, 4), 'No shapes', ()]
  6507. Inputs strides: [(0, 4, 2, 1), (48, 0, 12, 2), (40, 20, 4, 1), 'No strides', ()]
  6508. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614a58>, 1.0]
  6509.  
  6510. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6511. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6512.  
  6513. ======================================================================
  6514. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 204, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6515. ----------------------------------------------------------------------
  6516. Traceback (most recent call last):
  6517. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6518. self.test(*self.arg)
  6519. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6520. gpuval = f(img, kern)
  6521. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6522. storage_map=self.fn.storage_map)
  6523. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6524. outputs = self.fn()
  6525. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6526. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6527. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abceef90>, Scalar(float32)]
  6528. Inputs shapes: [(2, 1, 2, 2), (3, 2, 4, 3), (3, 1, 5, 4), 'No shapes', ()]
  6529. Inputs strides: [(4, 0, 2, 1), (96, 48, 12, 2), (20, 0, 4, 1), 'No strides', ()]
  6530. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9614620>, 1.0]
  6531.  
  6532. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6533. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6534.  
  6535. ======================================================================
  6536. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 205, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6537. ----------------------------------------------------------------------
  6538. Traceback (most recent call last):
  6539. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6540. self.test(*self.arg)
  6541. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6542. gpuval = f(img, kern)
  6543. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6544. storage_map=self.fn.storage_map)
  6545. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6546. outputs = self.fn()
  6547. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6548. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6549. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85dbb4fbd0>, Scalar(float32)]
  6550. Inputs shapes: [(2, 2, 2, 2), (1, 2, 4, 3), (1, 2, 5, 4), 'No shapes', ()]
  6551. Inputs strides: [(8, 4, 2, 1), (0, 48, 12, 2), (0, 20, 4, 1), 'No strides', ()]
  6552. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433198>, 1.0]
  6553.  
  6554. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6555. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6556.  
  6557. ======================================================================
  6558. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 6), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 206, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6559. ----------------------------------------------------------------------
  6560. Traceback (most recent call last):
  6561. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6562. self.test(*self.arg)
  6563. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6564. gpuval = f(img, kern)
  6565. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6566. storage_map=self.fn.storage_map)
  6567. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6568. outputs = self.fn()
  6569. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6570. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6571. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85ac5ab150>, Scalar(float32)]
  6572. Inputs shapes: [(2, 2, 2, 2), (2, 2, 4, 3), (2, 2, 5, 4), 'No shapes', ()]
  6573. Inputs strides: [(8, 4, 2, 1), (96, 48, 12, 2), (40, 20, 4, 1), 'No strides', ()]
  6574. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433148>, 1.0]
  6575.  
  6576. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6577. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6578.  
  6579. ======================================================================
  6580. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 6), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 207, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6581. ----------------------------------------------------------------------
  6582. Traceback (most recent call last):
  6583. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6584. self.test(*self.arg)
  6585. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6586. gpuval = f(img, kern)
  6587. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6588. storage_map=self.fn.storage_map)
  6589. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6590. outputs = self.fn()
  6591. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6592. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6593. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7344b90>, Scalar(float32)]
  6594. Inputs shapes: [(2, 4, 2, 2), (3, 2, 4, 3), (3, 4, 5, 4), 'No shapes', ()]
  6595. Inputs strides: [(16, 4, 2, 1), (96, 48, 12, 2), (80, 20, 4, 1), 'No strides', ()]
  6596. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a94338f0>, 1.0]
  6597.  
  6598. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6599. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6600.  
  6601. ======================================================================
  6602. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 208, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6603. ----------------------------------------------------------------------
  6604. Traceback (most recent call last):
  6605. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6606. self.test(*self.arg)
  6607. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6608. gpuval = f(img, kern)
  6609. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6610. storage_map=self.fn.storage_map)
  6611. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6612. outputs = self.fn()
  6613. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6614. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6615. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af377710>, Scalar(float32)]
  6616. Inputs shapes: [(2, 1, 2, 2), (1, 2, 3, 4), (1, 1, 4, 5), 'No shapes', ()]
  6617. Inputs strides: [(4, 0, 2, 1), (0, 48, 16, 2), (0, 0, 5, 1), 'No strides', ()]
  6618. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a9433c38>, 1.0]
  6619.  
  6620. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6621. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6622.  
  6623. ======================================================================
  6624. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (1, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 209, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6625. ----------------------------------------------------------------------
  6626. Traceback (most recent call last):
  6627. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6628. self.test(*self.arg)
  6629. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6630. gpuval = f(img, kern)
  6631. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6632. storage_map=self.fn.storage_map)
  6633. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6634. outputs = self.fn()
  6635. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6636. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6637. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dbdfa890>, Scalar(float32)]
  6638. Inputs shapes: [(1, 1, 2, 2), (3, 1, 3, 4), (3, 1, 4, 5), 'No shapes', ()]
  6639. Inputs strides: [(0, 0, 2, 1), (48, 0, 16, 2), (20, 0, 5, 1), 'No strides', ()]
  6640. Inputs values: [<CudaNdarray object at 0x7f85e6c01b70>, 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0530>, 1.0]
  6641.  
  6642. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6643. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6644.  
  6645. ======================================================================
  6646. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 210, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6647. ----------------------------------------------------------------------
  6648. Traceback (most recent call last):
  6649. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6650. self.test(*self.arg)
  6651. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6652. gpuval = f(img, kern)
  6653. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6654. storage_map=self.fn.storage_map)
  6655. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6656. outputs = self.fn()
  6657. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6658. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6659. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d88dc790>, Scalar(float32)]
  6660. Inputs shapes: [(1, 2, 2, 2), (1, 1, 3, 4), (1, 2, 4, 5), 'No shapes', ()]
  6661. Inputs strides: [(0, 4, 2, 1), (0, 0, 16, 2), (0, 20, 5, 1), 'No strides', ()]
  6662. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0f80>, 1.0]
  6663.  
  6664. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6665. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6666.  
  6667. ======================================================================
  6668. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 6, 8), (2, 1, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 211, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6669. ----------------------------------------------------------------------
  6670. Traceback (most recent call last):
  6671. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6672. self.test(*self.arg)
  6673. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6674. gpuval = f(img, kern)
  6675. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6676. storage_map=self.fn.storage_map)
  6677. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6678. outputs = self.fn()
  6679. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6680. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6681. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e6c3f410>, Scalar(float32)]
  6682. Inputs shapes: [(1, 2, 2, 2), (3, 1, 3, 4), (3, 2, 4, 5), 'No shapes', ()]
  6683. Inputs strides: [(0, 4, 2, 1), (48, 0, 16, 2), (40, 20, 5, 1), 'No strides', ()]
  6684. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e03a0>, 1.0]
  6685.  
  6686. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6687. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6688.  
  6689. ======================================================================
  6690. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (1, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 212, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6691. ----------------------------------------------------------------------
  6692. Traceback (most recent call last):
  6693. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6694. self.test(*self.arg)
  6695. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6696. gpuval = f(img, kern)
  6697. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6698. storage_map=self.fn.storage_map)
  6699. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6700. outputs = self.fn()
  6701. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6702. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6703. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ac6f6c90>, Scalar(float32)]
  6704. Inputs shapes: [(2, 1, 2, 2), (3, 2, 3, 4), (3, 1, 4, 5), 'No shapes', ()]
  6705. Inputs strides: [(4, 0, 2, 1), (96, 48, 16, 2), (20, 0, 5, 1), 'No strides', ()]
  6706. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85d96e0d00>, 1.0]
  6707.  
  6708. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6709. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6710.  
  6711. ======================================================================
  6712. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 213, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6713. ----------------------------------------------------------------------
  6714. Traceback (most recent call last):
  6715. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6716. self.test(*self.arg)
  6717. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6718. gpuval = f(img, kern)
  6719. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6720. storage_map=self.fn.storage_map)
  6721. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6722. outputs = self.fn()
  6723. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6724. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6725. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85d943c650>, Scalar(float32)]
  6726. Inputs shapes: [(2, 2, 2, 2), (1, 2, 3, 4), (1, 2, 4, 5), 'No shapes', ()]
  6727. Inputs strides: [(8, 4, 2, 1), (0, 48, 16, 2), (0, 20, 5, 1), 'No strides', ()]
  6728. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0a08>, 1.0]
  6729.  
  6730. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6731. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6732.  
  6733. ======================================================================
  6734. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 6, 8), (2, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 214, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6735. ----------------------------------------------------------------------
  6736. Traceback (most recent call last):
  6737. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6738. self.test(*self.arg)
  6739. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6740. gpuval = f(img, kern)
  6741. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6742. storage_map=self.fn.storage_map)
  6743. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6744. outputs = self.fn()
  6745. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6746. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6747. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85df33c550>, Scalar(float32)]
  6748. Inputs shapes: [(2, 2, 2, 2), (2, 2, 3, 4), (2, 2, 4, 5), 'No shapes', ()]
  6749. Inputs strides: [(8, 4, 2, 1), (96, 48, 16, 2), (40, 20, 5, 1), 'No strides', ()]
  6750. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0670>, 1.0]
  6751.  
  6752. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6753. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6754.  
  6755. ======================================================================
  6756. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 6, 8), (4, 2, 2, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 215, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6757. ----------------------------------------------------------------------
  6758. Traceback (most recent call last):
  6759. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6760. self.test(*self.arg)
  6761. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6762. gpuval = f(img, kern)
  6763. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6764. storage_map=self.fn.storage_map)
  6765. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6766. outputs = self.fn()
  6767. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6768. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6769. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d946b290>, Scalar(float32)]
  6770. Inputs shapes: [(2, 4, 2, 2), (3, 2, 3, 4), (3, 4, 4, 5), 'No shapes', ()]
  6771. Inputs strides: [(16, 4, 2, 1), (96, 48, 16, 2), (80, 20, 5, 1), 'No strides', ()]
  6772. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0238>, 1.0]
  6773.  
  6774. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6775. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6776.  
  6777. ======================================================================
  6778. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 216, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6779. ----------------------------------------------------------------------
  6780. Traceback (most recent call last):
  6781. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6782. self.test(*self.arg)
  6783. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6784. gpuval = f(img, kern)
  6785. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6786. storage_map=self.fn.storage_map)
  6787. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6788. outputs = self.fn()
  6789. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6790. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6791. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85ddae9f50>, Scalar(float32)]
  6792. Inputs shapes: [(2, 1, 3, 2), (1, 2, 4, 4), (1, 1, 6, 5), 'No shapes', ()]
  6793. Inputs strides: [(6, 0, 2, 1), (0, 64, 16, 2), (0, 0, 5, 1), 'No strides', ()]
  6794. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdc0bc0>, 1.0]
  6795.  
  6796. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6797. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6798.  
  6799. ======================================================================
  6800. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 217, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6801. ----------------------------------------------------------------------
  6802. Traceback (most recent call last):
  6803. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6804. self.test(*self.arg)
  6805. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6806. gpuval = f(img, kern)
  6807. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6808. storage_map=self.fn.storage_map)
  6809. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6810. outputs = self.fn()
  6811. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6812. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6813. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dc270f90>, Scalar(float32)]
  6814. Inputs shapes: [(1, 1, 3, 2), (3, 1, 4, 4), (3, 1, 6, 5), 'No shapes', ()]
  6815. Inputs strides: [(0, 0, 2, 1), (64, 0, 16, 2), (30, 0, 5, 1), 'No strides', ()]
  6816. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86710>, 1.0]
  6817.  
  6818. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6819. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6820.  
  6821. ======================================================================
  6822. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 218, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6823. ----------------------------------------------------------------------
  6824. Traceback (most recent call last):
  6825. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6826. self.test(*self.arg)
  6827. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6828. gpuval = f(img, kern)
  6829. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6830. storage_map=self.fn.storage_map)
  6831. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6832. outputs = self.fn()
  6833. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6834. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6835. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85c72ff710>, Scalar(float32)]
  6836. Inputs shapes: [(1, 2, 3, 2), (1, 1, 4, 4), (1, 2, 6, 5), 'No shapes', ()]
  6837. Inputs strides: [(0, 6, 2, 1), (0, 0, 16, 2), (0, 30, 5, 1), 'No strides', ()]
  6838. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee86670>, 1.0]
  6839.  
  6840. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6841. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6842.  
  6843. ======================================================================
  6844. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 219, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6845. ----------------------------------------------------------------------
  6846. Traceback (most recent call last):
  6847. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6848. self.test(*self.arg)
  6849. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6850. gpuval = f(img, kern)
  6851. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6852. storage_map=self.fn.storage_map)
  6853. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6854. outputs = self.fn()
  6855. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6856. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6857. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e07e1750>, Scalar(float32)]
  6858. Inputs shapes: [(1, 2, 3, 2), (3, 1, 4, 4), (3, 2, 6, 5), 'No shapes', ()]
  6859. Inputs strides: [(0, 6, 2, 1), (64, 0, 16, 2), (60, 30, 5, 1), 'No strides', ()]
  6860. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dee865f8>, 1.0]
  6861.  
  6862. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6863. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6864.  
  6865. ======================================================================
  6866. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 220, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6867. ----------------------------------------------------------------------
  6868. Traceback (most recent call last):
  6869. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6870. self.test(*self.arg)
  6871. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6872. gpuval = f(img, kern)
  6873. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6874. storage_map=self.fn.storage_map)
  6875. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6876. outputs = self.fn()
  6877. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6878. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6879. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85abd24a90>, Scalar(float32)]
  6880. Inputs shapes: [(2, 1, 3, 2), (3, 2, 4, 4), (3, 1, 6, 5), 'No shapes', ()]
  6881. Inputs strides: [(6, 0, 2, 1), (128, 64, 16, 2), (30, 0, 5, 1), 'No strides', ()]
  6882. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe328>, 1.0]
  6883.  
  6884. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6885. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6886.  
  6887. ======================================================================
  6888. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 221, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6889. ----------------------------------------------------------------------
  6890. Traceback (most recent call last):
  6891. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6892. self.test(*self.arg)
  6893. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6894. gpuval = f(img, kern)
  6895. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6896. storage_map=self.fn.storage_map)
  6897. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6898. outputs = self.fn()
  6899. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6900. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6901. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ac6f3210>, Scalar(float32)]
  6902. Inputs shapes: [(2, 2, 3, 2), (1, 2, 4, 4), (1, 2, 6, 5), 'No shapes', ()]
  6903. Inputs strides: [(12, 6, 2, 1), (0, 64, 16, 2), (0, 30, 5, 1), 'No strides', ()]
  6904. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6feb48>, 1.0]
  6905.  
  6906. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6907. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6908.  
  6909. ======================================================================
  6910. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 222, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6911. ----------------------------------------------------------------------
  6912. Traceback (most recent call last):
  6913. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6914. self.test(*self.arg)
  6915. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6916. gpuval = f(img, kern)
  6917. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6918. storage_map=self.fn.storage_map)
  6919. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6920. outputs = self.fn()
  6921. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6922. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6923. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85e72822d0>, Scalar(float32)]
  6924. Inputs shapes: [(2, 2, 3, 2), (2, 2, 4, 4), (2, 2, 6, 5), 'No shapes', ()]
  6925. Inputs strides: [(12, 6, 2, 1), (128, 64, 16, 2), (60, 30, 5, 1), 'No strides', ()]
  6926. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fe940>, 1.0]
  6927.  
  6928. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6929. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6930.  
  6931. ======================================================================
  6932. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 3, 2), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 223, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6933. ----------------------------------------------------------------------
  6934. Traceback (most recent call last):
  6935. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6936. self.test(*self.arg)
  6937. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6938. gpuval = f(img, kern)
  6939. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6940. storage_map=self.fn.storage_map)
  6941. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6942. outputs = self.fn()
  6943. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6944. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6945. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d88e2590>, Scalar(float32)]
  6946. Inputs shapes: [(2, 4, 3, 2), (3, 2, 4, 4), (3, 4, 6, 5), 'No shapes', ()]
  6947. Inputs strides: [(24, 6, 2, 1), (128, 64, 16, 2), (120, 30, 5, 1), 'No strides', ()]
  6948. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fedf0>, 1.0]
  6949.  
  6950. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6951. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6952.  
  6953. ======================================================================
  6954. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 224, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6955. ----------------------------------------------------------------------
  6956. Traceback (most recent call last):
  6957. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6958. self.test(*self.arg)
  6959. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6960. gpuval = f(img, kern)
  6961. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6962. storage_map=self.fn.storage_map)
  6963. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6964. outputs = self.fn()
  6965. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6966. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6967. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), <theano.gof.type.CDataType object at 0x7f85dedf6f90>, Scalar(float32)]
  6968. Inputs shapes: [(2, 1, 2, 3), (1, 2, 4, 4), (1, 1, 5, 6), 'No shapes', ()]
  6969. Inputs strides: [(6, 0, 3, 1), (0, 64, 16, 2), (0, 0, 6, 1), 'No strides', ()]
  6970. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85ac6fef80>, 1.0]
  6971.  
  6972. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6973. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6974.  
  6975. ======================================================================
  6976. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (1, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 225, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6977. ----------------------------------------------------------------------
  6978. Traceback (most recent call last):
  6979. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  6980. self.test(*self.arg)
  6981. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  6982. gpuval = f(img, kern)
  6983. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  6984. storage_map=self.fn.storage_map)
  6985. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  6986. outputs = self.fn()
  6987. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  6988. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  6989. Inputs types: [CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85c73b4750>, Scalar(float32)]
  6990. Inputs shapes: [(1, 1, 2, 3), (3, 1, 4, 4), (3, 1, 5, 6), 'No shapes', ()]
  6991. Inputs strides: [(0, 0, 3, 1), (64, 0, 16, 2), (30, 0, 6, 1), 'No strides', ()]
  6992. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1df0>, 1.0]
  6993.  
  6994. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  6995. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  6996.  
  6997. ======================================================================
  6998. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 226, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  6999. ----------------------------------------------------------------------
  7000. Traceback (most recent call last):
  7001. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7002. self.test(*self.arg)
  7003. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7004. gpuval = f(img, kern)
  7005. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7006. storage_map=self.fn.storage_map)
  7007. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7008. outputs = self.fn()
  7009. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7010. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7011. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, True, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85ddb1aa50>, Scalar(float32)]
  7012. Inputs shapes: [(1, 2, 2, 3), (1, 1, 4, 4), (1, 2, 5, 6), 'No shapes', ()]
  7013. Inputs strides: [(0, 6, 3, 1), (0, 0, 16, 2), (0, 30, 6, 1), 'No strides', ()]
  7014. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85dbdf1120>, 1.0]
  7015.  
  7016. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7017. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7018.  
  7019. ======================================================================
  7020. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 1, 8, 8), (2, 1, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 227, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  7021. ----------------------------------------------------------------------
  7022. Traceback (most recent call last):
  7023. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7024. self.test(*self.arg)
  7025. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7026. gpuval = f(img, kern)
  7027. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7028. storage_map=self.fn.storage_map)
  7029. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7030. outputs = self.fn()
  7031. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7032. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7033. Inputs types: [CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85dedfea90>, Scalar(float32)]
  7034. Inputs shapes: [(1, 2, 2, 3), (3, 1, 4, 4), (3, 2, 5, 6), 'No shapes', ()]
  7035. Inputs strides: [(0, 6, 3, 1), (64, 0, 16, 2), (60, 30, 6, 1), 'No strides', ()]
  7036. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a620>, 1.0]
  7037.  
  7038. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7039. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7040.  
  7041. ======================================================================
  7042. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (1, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 228, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  7043. ----------------------------------------------------------------------
  7044. Traceback (most recent call last):
  7045. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7046. self.test(*self.arg)
  7047. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7048. gpuval = f(img, kern)
  7049. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7050. storage_map=self.fn.storage_map)
  7051. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7052. outputs = self.fn()
  7053. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7054. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7055. Inputs types: [CudaNdarrayType(float32, (False, True, False, False)), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (False, True, False, False)), <theano.gof.type.CDataType object at 0x7f85af400090>, Scalar(float32)]
  7056. Inputs shapes: [(2, 1, 2, 3), (3, 2, 4, 4), (3, 1, 5, 6), 'No shapes', ()]
  7057. Inputs strides: [(6, 0, 3, 1), (128, 64, 16, 2), (30, 0, 6, 1), 'No strides', ()]
  7058. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a828>, 1.0]
  7059.  
  7060. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7061. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7062.  
  7063. ======================================================================
  7064. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((1, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 229, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  7065. ----------------------------------------------------------------------
  7066. Traceback (most recent call last):
  7067. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7068. self.test(*self.arg)
  7069. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7070. gpuval = f(img, kern)
  7071. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7072. storage_map=self.fn.storage_map)
  7073. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7074. outputs = self.fn()
  7075. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7076. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7077. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, (True, False, False, False)), CudaNdarrayType(float32, (True, False, False, False)), <theano.gof.type.CDataType object at 0x7f85abfd2190>, Scalar(float32)]
  7078. Inputs shapes: [(2, 2, 2, 3), (1, 2, 4, 4), (1, 2, 5, 6), 'No shapes', ()]
  7079. Inputs strides: [(12, 6, 3, 1), (0, 64, 16, 2), (0, 30, 6, 1), 'No strides', ()]
  7080. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a058>, 1.0]
  7081.  
  7082. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7083. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7084.  
  7085. ======================================================================
  7086. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((2, 2, 8, 8), (2, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 230, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  7087. ----------------------------------------------------------------------
  7088. Traceback (most recent call last):
  7089. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7090. self.test(*self.arg)
  7091. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7092. gpuval = f(img, kern)
  7093. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7094. storage_map=self.fn.storage_map)
  7095. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7096. outputs = self.fn()
  7097. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7098. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7099. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c73b4c10>, Scalar(float32)]
  7100. Inputs shapes: [(2, 2, 2, 3), (2, 2, 4, 4), (2, 2, 5, 6), 'No shapes', ()]
  7101. Inputs strides: [(12, 6, 3, 1), (128, 64, 16, 2), (60, 30, 6, 1), 'No strides', ()]
  7102. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973af30>, 1.0]
  7103.  
  7104. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7105. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7106.  
  7107. ======================================================================
  7108. ERROR: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.test_dnn_full((3, 2, 8, 8), (4, 2, 2, 3), 'full', (1, 1), (2, 2), (1, 1), -1, 0, True, None, 231, 1e-05, 1e-08, 0, False, None, <theano.compile.mode.Mode object at 0x7f85d37e8410>, <class 'theano.sandbox.cuda.dnn.DnnBase'>)
  7109. ----------------------------------------------------------------------
  7110. Traceback (most recent call last):
  7111. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7112. self.test(*self.arg)
  7113. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 205, in _params_allgood
  7114. gpuval = f(img, kern)
  7115. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7116. storage_map=self.fn.storage_map)
  7117. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7118. outputs = self.fn()
  7119. RuntimeError: GpuDnnConvGradI: error doing operation: CUDNN_STATUS_NOT_SUPPORTED
  7120. Apply node that caused the error: GpuDnnConvGradI{inplace=True}(GpuContiguous.0, CopyOnNegativeStrides.0, GpuAlloc{memset_0=True}.0, GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='cross'}.0, Constant{1.0})
  7121. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85d88e2d10>, Scalar(float32)]
  7122. Inputs shapes: [(2, 4, 2, 3), (3, 2, 4, 4), (3, 4, 5, 6), 'No shapes', ()]
  7123. Inputs strides: [(24, 6, 3, 1), (128, 64, 16, 2), (120, 30, 6, 1), 'No strides', ()]
  7124. Inputs values: ['not shown', 'not shown', 'not shown', <PyCObject object at 0x7f85a973a760>, 1.0]
  7125.  
  7126. HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created. This can be done with by setting the Theano flag 'optimizer=fast_compile'. If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
  7127. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7128.  
  7129. ======================================================================
  7130. ERROR: theano.sandbox.cuda.tests.test_dnn.test_pooling
  7131. ----------------------------------------------------------------------
  7132. Traceback (most recent call last):
  7133. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7134. self.test(*self.arg)
  7135. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_dnn.py", line 151, in test_pooling
  7136. g_out = fg(data)
  7137. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 606, in __call__
  7138. storage_map=self.fn.storage_map)
  7139. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 595, in __call__
  7140. outputs = self.fn()
  7141. RuntimeError: GpuDnnPoolGrad: error doing operation: CUDNN_STATUS_BAD_PARAM. input.shape=(1, 1, 2, 1) input_grad.shape=(1, 1, 1, 1) output.shape=(1, 1, 2, 2) output_grad.shape=(1, 1, 2, 2)
  7142. Apply node that caused the error: GpuDnnPoolGrad(GpuContiguous.0, GpuDnnPool.0, GpuContiguous.0, GpuDnnPoolDesc{ws=(2, 2), stride=(2, 2), mode='max', pad=(1, 0)}.0)
  7143. Inputs types: [CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), CudaNdarrayType(float32, 4D), <theano.gof.type.CDataType object at 0x7f85c7571e90>]
  7144. Inputs shapes: [(1, 1, 2, 2), (1, 1, 2, 1), (1, 1, 1, 1), 'No shapes']
  7145. Inputs strides: [(0, 0, 2, 1), (0, 0, 1, 0), (0, 0, 0, 0), 'No strides']
  7146. Inputs values: [<CudaNdarray object at 0x7f85af40f130>, <CudaNdarray object at 0x7f85dbb4f270>, <CudaNdarray object at 0x7f85a99b4370>, <PyCObject object at 0x7f85a973ae90>]
  7147.  
  7148. Backtrace when the node is created:
  7149. File "/home/ubuntu/repositories/theano/theano/gradient.py", line 1241, in access_grad_cache
  7150. term = access_term_cache(node)[idx]
  7151.  
  7152. HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
  7153.  
  7154. ======================================================================
  7155. ERROR: test_GpuCumsum2D (theano.sandbox.cuda.tests.test_extra_ops.TestGpuCumsum)
  7156. ----------------------------------------------------------------------
  7157. Traceback (most recent call last):
  7158. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_extra_ops.py", line 143, in test_GpuCumsum2D
  7159. f = theano.function([x], cumsum(x, axis=axis), mode=self.mode)
  7160. File "/home/ubuntu/repositories/theano/theano/compile/function.py", line 265, in function
  7161. profile=profile)
  7162. File "/home/ubuntu/repositories/theano/theano/compile/pfunc.py", line 511, in pfunc
  7163. on_unused_input=on_unused_input)
  7164. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1557, in orig_function
  7165. defaults)
  7166. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1415, in create
  7167. input_storage=input_storage_lists)
  7168. File "/home/ubuntu/repositories/theano/theano/gof/link.py", line 525, in make_thunk
  7169. output_storage=output_storage)[:3]
  7170. File "/home/ubuntu/repositories/theano/theano/gof/vm.py", line 897, in make_all
  7171. no_recycling))
  7172. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/__init__.py", line 256, in make_thunk
  7173. compute_map, no_recycling)
  7174. File "/home/ubuntu/repositories/theano/theano/gof/op.py", line 739, in make_thunk
  7175. output_storage=node_output_storage)
  7176. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1072, in make_thunk
  7177. keep_lock=keep_lock)
  7178. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1014, in __compile__
  7179. keep_lock=keep_lock)
  7180. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1433, in cthunk_factory
  7181. key = self.cmodule_key()
  7182. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1153, in cmodule_key
  7183. compile_args=self.compile_args(),
  7184. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 862, in compile_args
  7185. ret += x.c_compile_args()
  7186. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1786, in __hide
  7187. raise MethodNotDefined()
  7188. NameError: ('The following error happened while compiling the node', GpuFlatten{1}(GpuFromHost.0), '\n', "global name 'MethodNotDefined' is not defined")
  7189.  
  7190. ======================================================================
  7191. ERROR: test_GpuCumsum3D (theano.sandbox.cuda.tests.test_extra_ops.TestGpuCumsum)
  7192. ----------------------------------------------------------------------
  7193. Traceback (most recent call last):
  7194. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_extra_ops.py", line 182, in test_GpuCumsum3D
  7195. f = theano.function([x], cumsum(x, axis=axis), mode=self.mode)
  7196. File "/home/ubuntu/repositories/theano/theano/compile/function.py", line 265, in function
  7197. profile=profile)
  7198. File "/home/ubuntu/repositories/theano/theano/compile/pfunc.py", line 511, in pfunc
  7199. on_unused_input=on_unused_input)
  7200. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1557, in orig_function
  7201. defaults)
  7202. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1415, in create
  7203. input_storage=input_storage_lists)
  7204. File "/home/ubuntu/repositories/theano/theano/gof/link.py", line 525, in make_thunk
  7205. output_storage=output_storage)[:3]
  7206. File "/home/ubuntu/repositories/theano/theano/gof/vm.py", line 897, in make_all
  7207. no_recycling))
  7208. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/__init__.py", line 256, in make_thunk
  7209. compute_map, no_recycling)
  7210. File "/home/ubuntu/repositories/theano/theano/gof/op.py", line 739, in make_thunk
  7211. output_storage=node_output_storage)
  7212. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1072, in make_thunk
  7213. keep_lock=keep_lock)
  7214. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1014, in __compile__
  7215. keep_lock=keep_lock)
  7216. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1433, in cthunk_factory
  7217. key = self.cmodule_key()
  7218. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1153, in cmodule_key
  7219. compile_args=self.compile_args(),
  7220. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 862, in compile_args
  7221. ret += x.c_compile_args()
  7222. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1786, in __hide
  7223. raise MethodNotDefined()
  7224. NameError: ('The following error happened while compiling the node', GpuFlatten{1}(GpuFromHost.0), '\n', "global name 'MethodNotDefined' is not defined")
  7225.  
  7226. ======================================================================
  7227. ERROR: test_Strides2D (theano.sandbox.cuda.tests.test_extra_ops.TestGpuCumsum)
  7228. ----------------------------------------------------------------------
  7229. Traceback (most recent call last):
  7230. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_extra_ops.py", line 76, in test_Strides2D
  7231. mode=self.mode)
  7232. File "/home/ubuntu/repositories/theano/theano/compile/function.py", line 265, in function
  7233. profile=profile)
  7234. File "/home/ubuntu/repositories/theano/theano/compile/pfunc.py", line 511, in pfunc
  7235. on_unused_input=on_unused_input)
  7236. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1557, in orig_function
  7237. defaults)
  7238. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1415, in create
  7239. input_storage=input_storage_lists)
  7240. File "/home/ubuntu/repositories/theano/theano/gof/link.py", line 525, in make_thunk
  7241. output_storage=output_storage)[:3]
  7242. File "/home/ubuntu/repositories/theano/theano/gof/vm.py", line 897, in make_all
  7243. no_recycling))
  7244. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/__init__.py", line 256, in make_thunk
  7245. compute_map, no_recycling)
  7246. File "/home/ubuntu/repositories/theano/theano/gof/op.py", line 739, in make_thunk
  7247. output_storage=node_output_storage)
  7248. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1072, in make_thunk
  7249. keep_lock=keep_lock)
  7250. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1014, in __compile__
  7251. keep_lock=keep_lock)
  7252. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1433, in cthunk_factory
  7253. key = self.cmodule_key()
  7254. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1153, in cmodule_key
  7255. compile_args=self.compile_args(),
  7256. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 862, in compile_args
  7257. ret += x.c_compile_args()
  7258. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1786, in __hide
  7259. raise MethodNotDefined()
  7260. NameError: ('The following error happened while compiling the node', GpuFlatten{1}(GpuFromHost.0), '\n', "global name 'MethodNotDefined' is not defined")
  7261.  
  7262. ======================================================================
  7263. ERROR: test_Strides3D (theano.sandbox.cuda.tests.test_extra_ops.TestGpuCumsum)
  7264. ----------------------------------------------------------------------
  7265. Traceback (most recent call last):
  7266. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_extra_ops.py", line 99, in test_Strides3D
  7267. mode=self.mode)
  7268. File "/home/ubuntu/repositories/theano/theano/compile/function.py", line 265, in function
  7269. profile=profile)
  7270. File "/home/ubuntu/repositories/theano/theano/compile/pfunc.py", line 511, in pfunc
  7271. on_unused_input=on_unused_input)
  7272. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1557, in orig_function
  7273. defaults)
  7274. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1415, in create
  7275. input_storage=input_storage_lists)
  7276. File "/home/ubuntu/repositories/theano/theano/gof/link.py", line 525, in make_thunk
  7277. output_storage=output_storage)[:3]
  7278. File "/home/ubuntu/repositories/theano/theano/gof/vm.py", line 897, in make_all
  7279. no_recycling))
  7280. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/__init__.py", line 256, in make_thunk
  7281. compute_map, no_recycling)
  7282. File "/home/ubuntu/repositories/theano/theano/gof/op.py", line 739, in make_thunk
  7283. output_storage=node_output_storage)
  7284. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1072, in make_thunk
  7285. keep_lock=keep_lock)
  7286. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1014, in __compile__
  7287. keep_lock=keep_lock)
  7288. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1433, in cthunk_factory
  7289. key = self.cmodule_key()
  7290. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1153, in cmodule_key
  7291. compile_args=self.compile_args(),
  7292. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 862, in compile_args
  7293. ret += x.c_compile_args()
  7294. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1786, in __hide
  7295. raise MethodNotDefined()
  7296. NameError: ('The following error happened while compiling the node', GpuFlatten{1}(GpuFromHost.0), '\n', "global name 'MethodNotDefined' is not defined")
  7297.  
  7298. ======================================================================
  7299. ERROR: test_infer_shape (theano.sandbox.cuda.tests.test_extra_ops.TestGpuCumsum)
  7300. ----------------------------------------------------------------------
  7301. Traceback (most recent call last):
  7302. File "/home/ubuntu/repositories/theano/theano/tensor/tests/test_extra_ops.py", line 51, in test_infer_shape
  7303. self.op_class)
  7304. File "/home/ubuntu/repositories/theano/theano/tests/unittest_tools.py", line 237, in _compile_and_check
  7305. outputs_function = theano.function(inputs, outputs, mode=mode)
  7306. File "/home/ubuntu/repositories/theano/theano/compile/function.py", line 265, in function
  7307. profile=profile)
  7308. File "/home/ubuntu/repositories/theano/theano/compile/pfunc.py", line 511, in pfunc
  7309. on_unused_input=on_unused_input)
  7310. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1557, in orig_function
  7311. defaults)
  7312. File "/home/ubuntu/repositories/theano/theano/compile/function_module.py", line 1415, in create
  7313. input_storage=input_storage_lists)
  7314. File "/home/ubuntu/repositories/theano/theano/gof/link.py", line 525, in make_thunk
  7315. output_storage=output_storage)[:3]
  7316. File "/home/ubuntu/repositories/theano/theano/gof/vm.py", line 897, in make_all
  7317. no_recycling))
  7318. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/__init__.py", line 256, in make_thunk
  7319. compute_map, no_recycling)
  7320. File "/home/ubuntu/repositories/theano/theano/gof/op.py", line 739, in make_thunk
  7321. output_storage=node_output_storage)
  7322. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1072, in make_thunk
  7323. keep_lock=keep_lock)
  7324. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1014, in __compile__
  7325. keep_lock=keep_lock)
  7326. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1433, in cthunk_factory
  7327. key = self.cmodule_key()
  7328. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1153, in cmodule_key
  7329. compile_args=self.compile_args(),
  7330. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 862, in compile_args
  7331. ret += x.c_compile_args()
  7332. File "/home/ubuntu/repositories/theano/theano/gof/cc.py", line 1786, in __hide
  7333. raise MethodNotDefined()
  7334. NameError: ('The following error happened while compiling the node', GpuFlatten{1}(GpuFromHost.0), '\n', "global name 'MethodNotDefined' is not defined")
  7335.  
  7336. ======================================================================
  7337. FAIL: theano.sandbox.cuda.tests.test_conv_cuda_ndarray.TestConvWithPadding.test_numeric_value((5, 10, 8, 8), (12, 10, 4, 4), 3, <function dnn_conv at 0x7f85fbdf3050>)
  7338. ----------------------------------------------------------------------
  7339. Traceback (most recent call last):
  7340. File "/usr/local/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
  7341. self.test(*self.arg)
  7342. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py", line 692, in _run_onecase
  7343. Y = op(X, kern, border_mode=border_mode)
  7344. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/dnn.py", line 702, in dnn_conv
  7345. subsample)
  7346. File "/home/ubuntu/repositories/theano/theano/sandbox/cuda/dnn.py", line 471, in get_out_shape
  7347. assert border_mode == 'valid'
  7348. AssertionError
  7349.  
  7350. ----------------------------------------------------------------------
  7351. Ran 21080 tests in 6898.320s
  7352.  
  7353. FAILED (KNOWNFAIL=19, SKIP=45, errors=330, failures=1)
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!