API tools faq
paste
rajeswar

Untitled

rajeswar
Aug 8th, 2014
268
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Python 8.16 KB | None | 0 0
raw download clone embed print report
  1. Function profiling
  2. ==================
  3.   Message: Sum of all printed profiles at exit excluding Scan op profile.
  4.   Time in 590 calls to Function.__call__: 9.267533e+01s
  5.   Time in Function.fn.__call__: 9.263637e+01s (99.958%)
  6.   Time in thunks: 9.169430e+01s (98.941%)
  7.   Total compile time: 1.227890e+01s
  8.     Number of Apply nodes: 391
  9.     Theano Optimizer time: 7.960166e+00s
  10.        Theano validate time: 1.039941e-01s
  11.     Theano Linker time (includes C, CUDA code generation/compiling): 4.034438e+00s
  12.  
  13. Class
  14. ---
  15. <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Class name>
  16.   86.9%    86.9%      79.716s       1.98e-02s     C     4020      11   theano.sandbox.cuda.blas.GpuConv
  17.    7.0%    94.0%       6.448s       4.78e-03s     Py    1350       3   theano.tensor.nnet.conv3d2d.IncDiagonalSubtensor
  18.    2.4%    96.4%       2.183s       1.14e-04s     Py   19140      57   theano.sandbox.cuda.basic_ops.GpuReshape
  19.    1.6%    98.0%       1.473s       8.18e-04s     C     1800       4   theano.sandbox.cuda.basic_ops.GpuAlloc
  20.    0.5%    98.5%       0.485s       1.55e-04s     C     3120       9   theano.sandbox.cuda.basic_ops.GpuCAReduce
  21.    0.5%    99.0%       0.436s       3.80e-05s     C    11470      31   theano.sandbox.cuda.basic_ops.GpuElemwise
  22.    0.2%    99.2%       0.209s       1.18e-04s     C     1770       6   theano.sandbox.cuda.blas.GpuDownsampleFactorMax
  23.    0.2%    99.4%       0.165s       1.22e-04s     C     1350       3   theano.sandbox.cuda.blas.GpuDownsampleFactorMaxGrad
  24.    0.1%    99.5%       0.129s       6.18e-05s     C     2080       6   theano.sandbox.cuda.basic_ops.HostFromGpu
  25.    0.1%    99.6%       0.107s       5.15e-05s     C     2080       6   theano.sandbox.cuda.blas.GpuDot22
  26.    0.1%    99.7%       0.082s       5.53e-05s     C     1490       4   theano.tensor.elemwise.Sum
  27.    0.0%    99.8%       0.040s       2.24e-05s     C     1800       4   theano.sandbox.cuda.basic_ops.GpuFromHost
  28.    0.0%    99.8%       0.032s       3.52e-05s     C      900       2   theano.sandbox.cuda.blas.GpuGemm
  29.    0.0%    99.8%       0.028s       6.24e-05s     C      450       1   theano.sandbox.cuda.nnet.GpuCrossentropySoftmaxArgmax1HotWithBias
  30.    0.0%    99.9%       0.026s       1.48e-05s     Py    1770       6   theano.tensor.nnet.conv3d2d.DiagonalSubtensor
  31.    0.0%    99.9%       0.023s       3.86e-05s     Py     590       2   theano.sandbox.cuda.basic_ops.GpuFlatten
  32.    0.0%    99.9%       0.019s       6.36e-07s     C    30610      88   theano.compile.ops.Shape_i
  33.    0.0%    99.9%       0.016s       4.50e-06s     C     3540      12   theano.tensor.basic.Join
  34.    0.0%    99.9%       0.015s       1.18e-06s     C    12850      32   theano.sandbox.cuda.basic_ops.GpuDimShuffle
  35.    0.0%    99.9%       0.015s       3.31e-05s     C      450       1   theano.sandbox.cuda.nnet.GpuCrossentropySoftmax1HotWithBiasDx
  36.    ... (remaining 9 Classes account for   0.05%(0.05s) of the runtime)
  37.  
  38. Ops
  39. ---
  40. <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Op name>
  41.   73.9%    73.9%      67.803s       1.51e-01s     C      450        1   GpuConv{valid, (1, 1), None, (54, 72), False, (90, 60, 80), (54, 72)}
  42.    7.0%    81.0%       6.448s       4.78e-03s     Py    1350        3   IncDiagonalSubtensor
  43.    4.7%    85.7%       4.296s       7.28e-03s     C      590        2   GpuConv{valid, (1, 1), None, (7, 9), True, (1, 60, 80), (7, 9)}
  44.    2.1%    87.7%       1.883s       3.19e-03s     C      590        2   GpuConv{valid, (1, 1), None, (4, 6), True, (10, 4, 6), (4, 6)}
  45.    2.1%    89.8%       1.882s       1.06e-03s     Py    1770        6   GpuReshape{3}
  46.    1.8%    91.6%       1.655s       3.68e-03s     C      450        1   GpuConv{full, (1, 1), None, (7, 7), True, (40, 12, 18), (7, 7)}
  47.    1.6%    93.2%       1.473s       8.18e-04s     C     1800        4   GpuAlloc
  48.    1.4%    94.6%       1.280s       2.84e-03s     C      450        1   GpuConv{valid, (1, 1), None, (12, 18), False, (60, 18, 24), (12, 18)}
  49.    1.3%    95.9%       1.233s       2.74e-03s     C      450        1   GpuConv{full, (1, 1), None, (4, 6), True, (384, 1, 1), (4, 6)}
  50.    1.1%    97.1%       1.046s       1.77e-03s     C      590        2   GpuConv{valid, (1, 1), None, (7, 7), True, (5, 18, 24), (7, 7)}
  51.    0.6%    97.6%       0.521s       1.16e-03s     C      450        1   GpuConv{valid, (1, 1), None, (1, 1), False, (30, 4, 6), (1, 1)}
  52.    0.5%    98.1%       0.439s       2.48e-04s     C     1770        6   GpuCAReduce{add}{0,1,0}
  53.    0.2%    98.3%       0.208s       2.74e-05s     Py    7590       21   GpuReshape{5}
  54.    0.2%    98.5%       0.183s       1.55e-04s     C     1180        4   GpuDownsampleFactorMax{(3, 3),True}
  55.    0.2%    98.7%       0.143s       1.59e-04s     C      900        2   GpuDownsampleFactorMaxGrad{(3, 3),True}
  56.    0.1%    98.8%       0.129s       6.18e-05s     C     2080        6   HostFromGpu
  57.    0.1%    99.0%       0.113s       8.40e-05s     C     1350        3   GpuElemwise{add,no_inplace}
  58.    0.1%    99.1%       0.107s       5.15e-05s     C     2080        6   GpuDot22
  59.    0.1%    99.2%       0.086s       2.40e-05s     C     3600        8   GpuElemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)]
  60.    0.1%    99.3%       0.082s       6.09e-05s     C     1350        3   Sum{0, 1, 3, 4}
  61.    ... (remaining 61 Ops account for   0.75%(0.68s) of the runtime)
  62.  
  63. Apply
  64. ------
  65. <% time> <sum %> <apply time> <time per call> <#call> <id> <Apply name>
  66.   73.9%    73.9%      67.803s       1.51e-01s    450   254   GpuConv{valid, (1, 1), None, (54, 72), False, (90, 60, 80), (54, 72)}(GpuDimShuff
  67.    6.2%    80.1%       5.645s       1.25e-02s    450   250   IncDiagonalSubtensor(CudaNdarrayConstant{[[[[[[ 0.  0.  0. ...,  0.  0.  0.]
  68.    
  69.    3.6%    83.7%       3.298s       7.33e-03s    450    44   GpuConv{valid, (1, 1), None, (7, 9), True, (1, 60, 80), (7, 9)}(GpuReshape{4}.0,
  70.    1.8%    85.5%       1.655s       3.68e-03s    450   230   GpuConv{full, (1, 1), None, (7, 7), True, (40, 12, 18), (7, 7)}(GpuReshape{4}.0,
  71.    1.6%    87.1%       1.436s       3.19e-03s    450   131   GpuConv{valid, (1, 1), None, (4, 6), True, (10, 4, 6), (4, 6)}(GpuElemwise{tanh,n
  72.    1.4%    88.5%       1.280s       2.84e-03s    450   235   GpuConv{valid, (1, 1), None, (12, 18), False, (60, 18, 24), (12, 18)}(GpuDimShuff
  73.    1.4%    89.8%       1.246s       2.77e-03s    450    62   GpuReshape{3}(DiagonalSubtensor{inplace}.0, MakeVector.0)
  74.    1.3%    91.2%       1.237s       2.75e-03s    450   249   GpuAlloc(GpuDimShuffle{0,1,2,3,4,5}.0, Shape_i{0}.0, Shape_i{1}.0, Shape_i{2}.0,
  75.    1.3%    92.5%       1.233s       2.74e-03s    450   208   GpuConv{full, (1, 1), None, (4, 6), True, (384, 1, 1), (4, 6)}(GpuReshape{4}.0, G
  76.    1.1%    93.6%       0.998s       7.13e-03s    140    16   GpuConv{valid, (1, 1), None, (7, 9), True, (1, 60, 80), (7, 9)}(GpuReshape{4}.0,
  77.    0.9%    94.5%       0.797s       1.77e-03s    450    88   GpuConv{valid, (1, 1), None, (7, 7), True, (5, 18, 24), (7, 7)}(GpuElemwise{tanh,
  78.    0.8%    95.2%       0.707s       1.57e-03s    450   228   IncDiagonalSubtensor(CudaNdarrayConstant{[[[[[[ 0.  0.  0. ...,  0.  0.  0.]
  79.    
  80.    0.6%    95.8%       0.521s       1.16e-03s    450   213   GpuConv{valid, (1, 1), None, (1, 1), False, (30, 4, 6), (1, 1)}(GpuDimShuffle{1,0
  81.    0.5%    96.3%       0.447s       3.19e-03s    140    88   GpuConv{valid, (1, 1), None, (4, 6), True, (10, 4, 6), (4, 6)}(GpuElemwise{Tanh}[
  82.    0.4%    96.7%       0.390s       2.79e-03s    140    32   GpuReshape{3}(DiagonalSubtensor{inplace}.0, MakeVector.0)
  83.    0.3%    97.0%       0.248s       1.77e-03s    140    52   GpuConv{valid, (1, 1), None, (7, 7), True, (5, 18, 24), (7, 7)}(GpuElemwise{Tanh}
  84.    0.3%    97.3%       0.242s       5.38e-04s    450    64   GpuCAReduce{add}{0,1,0}(GpuReshape{3}.0)
  85.    0.2%    97.4%       0.153s       3.40e-04s    450   227   GpuAlloc(GpuDimShuffle{0,1,2,3,4,5}.0, Shape_i{0}.0, Shape_i{1}.0, Shape_i{2}.0,
  86.    0.2%    97.6%       0.149s       3.32e-04s    450   105   GpuReshape{3}(DiagonalSubtensor{inplace}.0, MakeVector.0)
  87.    0.1%    97.7%       0.116s       2.58e-04s    450   240   GpuDownsampleFactorMaxGrad{(3, 3),True}(GpuReshape{4}.0, GpuDownsampleFactorMax{(
  88.    ... (remaining 371 Apply instances account for 2.28%(2.09s) of the runtime)
  89.  
  90.  
  91. TACC: Shutdown complete. Exiting.
Advertisement
Add Comment
Please, Sign In to add comment
Public Pastes
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!