API tools faq
paste
Advertisement
Guest User

cuda compile error

a guest
Jan 15th, 2022
117
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Bash 271.94 KB | None | 0 0
raw download clone embed print report
  1. (tensorflow2_p38) ubuntu@ip-172-31-40-250:~/instant-ngp$ cmake --build build --config RelWithDebInfo -j 16
  2. Consolidate compiler generated dependencies of target tiny-cuda-nn
  3. [ 29%] Built target glfw_objects
  4. [ 30%] Building CUDA object dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_mlp.cu.o
  5. [ 32%] Building CUDA object dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_resnet.cu.o
  6. [ 34%] Building CUDA object dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/encoding.cu.o
  7. [ 36%] Building CUDA object dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/object.cu.o
  8. [ 38%] Building CUDA object dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/fully_fused_mlp.cu.o
  9. nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
  10. nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
  11. nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
  12. nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
  13. nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
  14. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(400): error: explicit type is missing ("int" assumed)
  15.  
  16. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(400): error: expected a ")"
  17.  
  18. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(480): error: explicit type is missing ("int" assumed)
  19.  
  20. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(480): error: expected a ")"
  21.  
  22. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(595): error: no operator "*=" matches these operands
  23.             operand types are: __half *= float
  24.           detected during:
  25.             instantiation of "void tcnn::mult_scalar_kernel(uint32_t, T *, float) [with T=__half]"
  26. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/object.cu(59): here
  27.             instantiation of "void tcnn::mult(cudaStream_t, uint32_t, T *, float) [with T=__half]"
  28. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/object.cu(63): here
  29.  
  30. 1 error detected in the compilation of "/home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/object.cu".
  31. dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/build.make:187: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/object.cu.o' failed
  32. make[2]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/object.cu.o] Error 1
  33. make[2]: *** Waiting for unfinished jobs....
  34. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(305): error: no instance of overloaded function "atomicAdd" matches the argument list
  35.             argument types are: (__half2 *, __half2)
  36.           detected during:
  37.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=float, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  38. (537): here
  39.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=float, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  40. (537): here
  41.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=float, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  42. (537): here
  43.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=float, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  44. (814): here
  45.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=float, N_FEATURES_PER_LEVEL=1U]"
  46. (826): here
  47.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=float]"
  48. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  49.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=float]"
  50. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  51.  
  52. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  53.             operand types are: __half += __half
  54.           detected during:
  55.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  56. (600): here
  57.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  58. (537): here
  59.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  60. (537): here
  61.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  62. (537): here
  63.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=1U]"
  64. (814): here
  65.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=1U]"
  66. (826): here
  67.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  68. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  69.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  70. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  71.  
  72. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  73.             operand types are: __half += __half
  74.           detected during:
  75.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=1U]"
  76. (600): here
  77.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=1U]"
  78. (537): here
  79.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=1U]"
  80. (537): here
  81.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=1U]"
  82. (537): here
  83.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=1U]"
  84. (815): here
  85.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=1U]"
  86. (826): here
  87.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  88. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  89.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  90. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  91.  
  92. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  93.             operand types are: __half += __half
  94.           detected during:
  95.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  96. (600): here
  97.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  98. (537): here
  99.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  100. (537): here
  101.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  102. (537): here
  103.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  104. (814): here
  105.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=2U]"
  106. (827): here
  107.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  108. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  109.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  110. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  111.  
  112. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  113.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  114.           detected during:
  115.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U, N_FEATURES_PER_THREAD=2U]"
  116. (674): here
  117.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  118. (537): here
  119.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  120. (537): here
  121.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  122. (537): here
  123.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=2U]"
  124. (814): here
  125.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=2U]"
  126. (827): here
  127.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  128. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  129.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  130. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  131.  
  132. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  133.             operand types are: __half += __half
  134.           detected during:
  135.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  136. (600): here
  137.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  138. (537): here
  139.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  140. (537): here
  141.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  142. (537): here
  143.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  144. (815): here
  145.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=2U]"
  146. (827): here
  147.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  148. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  149.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  150. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  151.  
  152. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  153.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  154.           detected during:
  155.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U, N_FEATURES_PER_THREAD=2U]"
  156. (674): here
  157.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  158. (537): here
  159.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  160. (537): here
  161.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  162. (537): here
  163.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=2U]"
  164. (815): here
  165.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=2U]"
  166. (827): here
  167.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  168. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  169.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  170. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  171.  
  172. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  173.             operand types are: __half += __half
  174.           detected during:
  175.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  176. (600): here
  177.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  178. (537): here
  179.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  180. (537): here
  181.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  182. (537): here
  183.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  184. (814): here
  185.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=4U]"
  186. (828): here
  187.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  188. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  189.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  190. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  191.  
  192. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  193.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  194.           detected during:
  195.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U, N_FEATURES_PER_THREAD=2U]"
  196. (674): here
  197.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  198. (537): here
  199.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  200. (537): here
  201.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  202. (537): here
  203.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=4U]"
  204. (814): here
  205.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=4U]"
  206. (828): here
  207.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  208. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  209.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  210. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  211.  
  212. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  213.             operand types are: __half += __half
  214.           detected during:
  215.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  216. (600): here
  217.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  218. (537): here
  219.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  220. (537): here
  221.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  222. (537): here
  223.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  224. (815): here
  225.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=4U]"
  226. (828): here
  227.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  228. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  229.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  230. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  231.  
  232. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  233.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  234.           detected during:
  235.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U, N_FEATURES_PER_THREAD=2U]"
  236. (674): here
  237.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  238. (537): here
  239.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  240. (537): here
  241.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  242. (537): here
  243.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=4U]"
  244. (815): here
  245.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=4U]"
  246. (828): here
  247.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  248. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  249.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  250. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  251.  
  252. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  253.             operand types are: __half += __half
  254.           detected during:
  255.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  256. (600): here
  257.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  258. (537): here
  259.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  260. (537): here
  261.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  262. (537): here
  263.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  264. (814): here
  265.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=8U]"
  266. (829): here
  267.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  268. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  269.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  270. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  271.  
  272. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  273.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  274.           detected during:
  275.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U, N_FEATURES_PER_THREAD=2U]"
  276. (674): here
  277.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  278. (537): here
  279.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  280. (537): here
  281.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  282. (537): here
  283.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=2U, N_FEATURES_PER_LEVEL=8U]"
  284. (814): here
  285.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=8U]"
  286. (829): here
  287.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  288. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  289.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  290. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  291.  
  292. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(211): error: no operator "+=" matches these operands
  293.             operand types are: __half += __half
  294.           detected during:
  295.             instantiation of "void tcnn::kernel_grid<T,N_POS_DIMS,N_FEATURES_PER_LEVEL>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, float, const float *, tcnn::InterpolationType, tcnn::GridType, const T *, const float *, tcnn::vector_t<T, N_FEATURES_PER_LEVEL> *, float *) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  296. (600): here
  297.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::encode(cudaStream_t, uint32_t, tcnn::PitchedPtr<const float>, tcnn::PitchedPtr<T>, float *, __nv_bool) const [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  298. (537): here
  299.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  300. (537): here
  301.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  302. (537): here
  303.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  304. (815): here
  305.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=8U]"
  306. (829): here
  307.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  308. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  309.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  310. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  311.  
  312. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/grid.h(300): error: no instance of overloaded function "atomicAdd" matches the argument list
  313.             argument types are: (std::conditional_t<false, float, __half> *, __half)
  314.           detected during:
  315.             instantiation of "void tcnn::kernel_grid_backward<T,GRAD_T,N_POS_DIMS,N_FEATURES_PER_LEVEL,N_FEATURES_PER_THREAD>(uint32_t, uint32_t, const uint32_t *, uint32_t, float, float, const float *, __nv_bool, tcnn::InterpolationType, tcnn::GridType, GRAD_T *, const float *, const tcnn::vector_t<T, N_FEATURES_PER_THREAD> *) [with T=__half, GRAD_T=std::conditional_t<false, float, __half>, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U, N_FEATURES_PER_THREAD=2U]"
  316. (674): here
  317.             instantiation of "void tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::backward(cudaStream_t, uint32_t, tcnn::PitchedPtr<const T>, const float *, tcnn::PitchedPtr<float>, tcnn::PitchedPtr<const float>, __nv_bool) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  318. (537): here
  319.             implicit generation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::~GridEncodingTemplated() [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  320. (537): here
  321.             instantiation of class "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL> [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  322. (537): here
  323.             instantiation of "tcnn::GridEncodingTemplated<T, N_POS_DIMS, N_FEATURES_PER_LEVEL>::GridEncodingTemplated(uint32_t, uint32_t, uint32_t, float, __nv_bool, tcnn::InterpolationType, tcnn::GridType) [with T=__half, N_POS_DIMS=3U, N_FEATURES_PER_LEVEL=8U]"
  324. (815): here
  325.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding_templated<T,N_FEATURES_PER_LEVEL>(uint32_t, const tcnn::json &) [with T=__half, N_FEATURES_PER_LEVEL=8U]"
  326. (829): here
  327.             instantiation of "tcnn::GridEncoding<T> *tcnn::create_grid_encoding<T>(uint32_t, const tcnn::json &) [with T=__half]"
  328. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu(118): here
  329.             instantiation of "tcnn::Encoding<T> *tcnn::create_encoding<T>(uint32_t, const tcnn::json &, uint32_t) [with T=__half]"
  330. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/encodings/composite.h(84): here
  331.  
  332. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(617): error: name followed by "::" must be a class or namespace name
  333.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  334. (699): here
  335.  
  336. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(617): error: name followed by "::" must be a class or namespace name
  337.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  338. (699): here
  339.  
  340. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(527): error: identifier "output_layout" is undefined
  341.           detected during:
  342.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  343. (741): here
  344.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  345. (699): here
  346.  
  347. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(527): error: name followed by "::" must be a class or namespace name
  348.           detected during:
  349.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  350. (741): here
  351.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  352. (699): here
  353.  
  354. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(60): error: name must be a namespace name
  355.           detected during:
  356.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  357. (606): here
  358.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  359. (741): here
  360.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  361. (699): here
  362.  
  363. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(64): error: identifier "wmma" is undefined
  364.           detected during:
  365.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  366. (606): here
  367.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  368. (741): here
  369.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  370. (699): here
  371.  
  372. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(64): error: too few arguments for alias template "std::conditional_t"
  373.           detected during:
  374.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  375. (606): here
  376.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  377. (741): here
  378.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  379. (699): here
  380.  
  381. 15 errors detected in the compilation of "/home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/encoding.cu".
  382. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(64): error: expected a ";"
  383.           detected during:
  384.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  385. (606): here
  386.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  387. (741): here
  388.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  389. (699): here
  390.  
  391. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(67): error: name followed by "::" must be a class or namespace name
  392.           detected during:
  393.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  394. (606): here
  395.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  396. (741): here
  397.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  398. (699): here
  399.  
  400. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(67): error: type name is not allowed
  401.           detected during:
  402.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  403. (606): here
  404.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  405. (741): here
  406.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  407. (699): here
  408.  
  409. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(67): error: name followed by "::" must be a class or namespace name
  410.           detected during:
  411.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  412. (606): here
  413.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  414. (741): here
  415.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  416. (699): here
  417.  
  418. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(67): error: identifier "act_frag" is undefined
  419.           detected during:
  420.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  421. (606): here
  422.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  423. (741): here
  424.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  425. (699): here
  426.  
  427. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(68): error: name followed by "::" must be a class or namespace name
  428.           detected during:
  429.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  430. (606): here
  431.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  432. (741): here
  433.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  434. (699): here
  435.  
  436. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(68): error: type name is not allowed
  437.           detected during:
  438.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  439. (606): here
  440.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  441. (741): here
  442.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  443. (699): here
  444.  
  445. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(68): error: type name is not allowed
  446.           detected during:
  447.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  448. (606): here
  449.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  450. (741): here
  451.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  452. (699): here
  453.  
  454. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(68): error: identifier "weights_frag" is undefined
  455.           detected during:
  456.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  457. (606): here
  458.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  459. (741): here
  460.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  461. (699): here
  462.  
  463. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(69): error: name followed by "::" must be a class or namespace name
  464.           detected during:
  465.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  466. (606): here
  467.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  468. (741): here
  469.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  470. (699): here
  471.  
  472. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(69): error: type name is not allowed
  473.           detected during:
  474.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  475. (606): here
  476.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  477. (741): here
  478.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  479. (699): here
  480.  
  481. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(69): error: identifier "result_frag" is undefined
  482.           detected during:
  483.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  484. (606): here
  485.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  486. (741): here
  487.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  488. (699): here
  489.  
  490. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(88): error: name followed by "::" must be a class or namespace name
  491.           detected during:
  492.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  493. (606): here
  494.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  495. (741): here
  496.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  497. (699): here
  498.  
  499. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(90): error: name followed by "::" must be a class or namespace name
  500.           detected during:
  501.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  502. (606): here
  503.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  504. (741): here
  505.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  506. (699): here
  507.  
  508. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(96): error: name followed by "::" must be a class or namespace name
  509.           detected during:
  510.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  511. (606): here
  512.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  513. (741): here
  514.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  515. (699): here
  516.  
  517. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(101): error: name followed by "::" must be a class or namespace name
  518.           detected during:
  519.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  520. (606): here
  521.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  522. (741): here
  523.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  524. (699): here
  525.  
  526. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(102): error: name followed by "::" must be a class or namespace name
  527.           detected during:
  528.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  529. (606): here
  530.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  531. (741): here
  532.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  533. (699): here
  534.  
  535. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(108): error: name followed by "::" must be a class or namespace name
  536.           detected during:
  537.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  538. (606): here
  539.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  540. (741): here
  541.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  542. (699): here
  543.  
  544. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(119): error: name followed by "::" must be a class or namespace name
  545.           detected during:
  546.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  547. (606): here
  548.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  549. (741): here
  550.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  551. (699): here
  552.  
  553. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(119): error: name followed by "::" must be a class or namespace name
  554.           detected during:
  555.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  556. (606): here
  557.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  558. (741): here
  559.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  560. (699): here
  561.  
  562. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(322): error: name must be a namespace name
  563.           detected during:
  564.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  565. (606): here
  566.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  567. (741): here
  568.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  569. (699): here
  570.  
  571. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(325): error: name followed by "::" must be a class or namespace name
  572.           detected during:
  573.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  574. (606): here
  575.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  576. (741): here
  577.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  578. (699): here
  579.  
  580. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(325): error: type name is not allowed
  581.           detected during:
  582.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  583. (606): here
  584.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  585. (741): here
  586.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  587. (699): here
  588.  
  589. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(325): error: name followed by "::" must be a class or namespace name
  590.           detected during:
  591.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  592. (606): here
  593.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  594. (741): here
  595.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  596. (699): here
  597.  
  598. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(325): error: identifier "act_frag" is undefined
  599.           detected during:
  600.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  601. (606): here
  602.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  603. (741): here
  604.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  605. (699): here
  606.  
  607. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(326): error: name followed by "::" must be a class or namespace name
  608.           detected during:
  609.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  610. (606): here
  611.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  612. (741): here
  613.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  614. (699): here
  615.  
  616. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(326): error: type name is not allowed
  617.           detected during:
  618.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  619. (606): here
  620.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  621. (741): here
  622.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  623. (699): here
  624.  
  625. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(326): error: name followed by "::" must be a class or namespace name
  626.           detected during:
  627.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  628. (606): here
  629.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  630. (741): here
  631.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  632. (699): here
  633.  
  634. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(326): error: identifier "weights_frag" is undefined
  635.           detected during:
  636.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  637. (606): here
  638.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  639. (741): here
  640.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  641. (699): here
  642.  
  643. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(327): error: name followed by "::" must be a class or namespace name
  644.           detected during:
  645.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  646. (606): here
  647.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  648. (741): here
  649.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  650. (699): here
  651.  
  652. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(327): error: type name is not allowed
  653.           detected during:
  654.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  655. (606): here
  656.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  657. (741): here
  658.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  659. (699): here
  660.  
  661. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(327): error: identifier "result_frag" is undefined
  662.           detected during:
  663.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  664. (606): here
  665.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  666. (741): here
  667.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  668. (699): here
  669.  
  670. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(370): error: name followed by "::" must be a class or namespace name
  671.           detected during:
  672.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  673. (606): here
  674.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  675. (741): here
  676.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  677. (699): here
  678.  
  679. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(374): error: name followed by "::" must be a class or namespace name
  680.           detected during:
  681.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  682. (606): here
  683.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  684. (741): here
  685.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  686. (699): here
  687.  
  688. dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/build.make:131: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/encoding.cu.o' failed
  689. make[2]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/encoding.cu.o] Error 1
  690. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(375): error: name followed by "::" must be a class or namespace name
  691.           detected during:
  692.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  693. (606): here
  694.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  695. (741): here
  696.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  697. (699): here
  698.  
  699. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(376): error: name followed by "::" must be a class or namespace name
  700.           detected during:
  701.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  702. (606): here
  703.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  704. (741): here
  705.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  706. (699): here
  707.  
  708. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(386): error: name followed by "::" must be a class or namespace name
  709.           detected during:
  710.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  711. (606): here
  712.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  713. (741): here
  714.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  715. (699): here
  716.  
  717. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(386): error: name followed by "::" must be a class or namespace name
  718.           detected during:
  719.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  720. (606): here
  721.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  722. (741): here
  723.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  724. (699): here
  725.  
  726. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(409): error: name must be a namespace name
  727.           detected during:
  728.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  729. (606): here
  730.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  731. (741): here
  732.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  733. (699): here
  734.  
  735. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(412): error: name followed by "::" must be a class or namespace name
  736.           detected during:
  737.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  738. (606): here
  739.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  740. (741): here
  741.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  742. (699): here
  743.  
  744. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(412): error: type name is not allowed
  745.           detected during:
  746.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  747. (606): here
  748.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  749. (741): here
  750.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  751. (699): here
  752.  
  753. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(412): error: name followed by "::" must be a class or namespace name
  754.           detected during:
  755.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  756. (606): here
  757.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  758. (741): here
  759.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  760. (699): here
  761.  
  762. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(412): error: identifier "act_frag" is undefined
  763.           detected during:
  764.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  765. (606): here
  766.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  767. (741): here
  768.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  769. (699): here
  770.  
  771. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(413): error: name followed by "::" must be a class or namespace name
  772.           detected during:
  773.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  774. (606): here
  775.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  776. (741): here
  777.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  778. (699): here
  779.  
  780. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(413): error: type name is not allowed
  781.           detected during:
  782.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  783. (606): here
  784.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  785. (741): here
  786.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  787. (699): here
  788.  
  789. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(413): error: name followed by "::" must be a class or namespace name
  790.           detected during:
  791.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  792. (606): here
  793.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  794. (741): here
  795.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  796. (699): here
  797.  
  798. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(413): error: identifier "weights_frag" is undefined
  799.           detected during:
  800.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  801. (606): here
  802.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  803. (741): here
  804.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  805. (699): here
  806.  
  807. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(414): error: name followed by "::" must be a class or namespace name
  808.           detected during:
  809.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  810. (606): here
  811.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  812. (741): here
  813.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  814. (699): here
  815.  
  816. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(414): error: type name is not allowed
  817.           detected during:
  818.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  819. (606): here
  820.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  821. (741): here
  822.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  823. (699): here
  824.  
  825. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(414): error: identifier "result_frag" is undefined
  826.           detected during:
  827.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  828. (606): here
  829.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  830. (741): here
  831.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  832. (699): here
  833.  
  834. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(436): error: name followed by "::" must be a class or namespace name
  835.           detected during:
  836.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  837. (606): here
  838.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  839. (741): here
  840.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  841. (699): here
  842.  
  843. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(440): error: name followed by "::" must be a class or namespace name
  844.           detected during:
  845.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  846. (606): here
  847.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  848. (741): here
  849.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  850. (699): here
  851.  
  852. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(444): error: name followed by "::" must be a class or namespace name
  853.           detected during:
  854.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  855. (606): here
  856.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  857. (741): here
  858.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  859. (699): here
  860.  
  861. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(445): error: name followed by "::" must be a class or namespace name
  862.           detected during:
  863.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  864. (606): here
  865.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  866. (741): here
  867.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  868. (699): here
  869.  
  870. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(450): error: identifier "output_layout" is undefined
  871.           detected during:
  872.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  873. (606): here
  874.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  875. (741): here
  876.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  877. (699): here
  878.  
  879. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(450): error: name followed by "::" must be a class or namespace name
  880.           detected during:
  881.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  882. (606): here
  883.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  884. (741): here
  885.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  886. (699): here
  887.  
  888. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(451): error: name followed by "::" must be a class or namespace name
  889.           detected during:
  890.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  891. (606): here
  892.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  893. (741): here
  894.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  895. (699): here
  896.  
  897. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(453): error: name followed by "::" must be a class or namespace name
  898.           detected during:
  899.             instantiation of "void tcnn::kernel_mlp_fused<WIDTH,BLOCK_DIM_Z,N_ITERS,OUT_T,ACTIVATION,INFERENCE>(tcnn::Activation, const __half *, const __half *, OUT_T *, OUT_T *, uint32_t, uint32_t, uint32_t, uint32_t, int) [with WIDTH=256, BLOCK_DIM_Z=1, N_ITERS=2, OUT_T=__half, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  900. (606): here
  901.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_forward<WIDTH,T,ACTIVATION,INFERENCE>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None, INFERENCE=true]"
  902. (741): here
  903.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  904. (699): here
  905.  
  906. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  907.             function "__half::operator float() const"
  908.             function "__half::operator short() const"
  909.             function "__half::operator unsigned short() const"
  910.             function "__half::operator int() const"
  911.             function "__half::operator unsigned int() const"
  912.             function "__half::operator long long() const"
  913.             function "__half::operator unsigned long long() const"
  914.             function "__half::operator __nv_bool() const"
  915.           detected during:
  916.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  917. (270): here
  918.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  919. (335): here
  920.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  921. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  922.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  923. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  924.  
  925. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  926.             function "__half::operator float() const"
  927.             function "__half::operator short() const"
  928.             function "__half::operator unsigned short() const"
  929.             function "__half::operator int() const"
  930.             function "__half::operator unsigned int() const"
  931.             function "__half::operator long long() const"
  932.             function "__half::operator unsigned long long() const"
  933.             function "__half::operator __nv_bool() const"
  934.           detected during:
  935.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  936. (270): here
  937.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  938. (335): here
  939.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  940. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  941.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  942. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  943.  
  944. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  945.             function "__half::operator float() const"
  946.             function "__half::operator short() const"
  947.             function "__half::operator unsigned short() const"
  948.             function "__half::operator int() const"
  949.             function "__half::operator unsigned int() const"
  950.             function "__half::operator long long() const"
  951.             function "__half::operator unsigned long long() const"
  952.             function "__half::operator __nv_bool() const"
  953.           detected during:
  954.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  955. (270): here
  956.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  957. (335): here
  958.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  959. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  960.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  961. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  962.  
  963. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  964.             function "__half::operator float() const"
  965.             function "__half::operator short() const"
  966.             function "__half::operator unsigned short() const"
  967.             function "__half::operator int() const"
  968.             function "__half::operator unsigned int() const"
  969.             function "__half::operator long long() const"
  970.             function "__half::operator unsigned long long() const"
  971.             function "__half::operator __nv_bool() const"
  972.           detected during:
  973.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  974. (270): here
  975.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  976. (335): here
  977.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  978. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  979.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  980. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  981.  
  982. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  983.             function "__half::operator float() const"
  984.             function "__half::operator short() const"
  985.             function "__half::operator unsigned short() const"
  986.             function "__half::operator int() const"
  987.             function "__half::operator unsigned int() const"
  988.             function "__half::operator long long() const"
  989.             function "__half::operator unsigned long long() const"
  990.             function "__half::operator __nv_bool() const"
  991.           detected during:
  992.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  993. (270): here
  994.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  995. (335): here
  996.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  997. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  998.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  999. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1000.  
  1001. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1002.             function "__half::operator float() const"
  1003.             function "__half::operator short() const"
  1004.             function "__half::operator unsigned short() const"
  1005.             function "__half::operator int() const"
  1006.             function "__half::operator unsigned int() const"
  1007.             function "__half::operator long long() const"
  1008.             function "__half::operator unsigned long long() const"
  1009.             function "__half::operator __nv_bool() const"
  1010.           detected during:
  1011.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1012. (270): here
  1013.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1014. (335): here
  1015.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1016. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  1017.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1018. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1019.  
  1020. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1021.             function "__half::operator float() const"
  1022.             function "__half::operator short() const"
  1023.             function "__half::operator unsigned short() const"
  1024.             function "__half::operator int() const"
  1025.             function "__half::operator unsigned int() const"
  1026.             function "__half::operator long long() const"
  1027.             function "__half::operator unsigned long long() const"
  1028.             function "__half::operator __nv_bool() const"
  1029.           detected during:
  1030.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1031. (270): here
  1032.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1033. (335): here
  1034.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1035. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  1036.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1037. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1038.  
  1039. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1040.             function "__half::operator float() const"
  1041.             function "__half::operator short() const"
  1042.             function "__half::operator unsigned short() const"
  1043.             function "__half::operator int() const"
  1044.             function "__half::operator unsigned int() const"
  1045.             function "__half::operator long long() const"
  1046.             function "__half::operator unsigned long long() const"
  1047.             function "__half::operator __nv_bool() const"
  1048.           detected during:
  1049.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1050. (270): here
  1051.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1052. (335): here
  1053.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1054. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  1055.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1056. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1057.  
  1058. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1059.             function "__half::operator float() const"
  1060.             function "__half::operator short() const"
  1061.             function "__half::operator unsigned short() const"
  1062.             function "__half::operator int() const"
  1063.             function "__half::operator unsigned int() const"
  1064.             function "__half::operator long long() const"
  1065.             function "__half::operator unsigned long long() const"
  1066.             function "__half::operator __nv_bool() const"
  1067.           detected during:
  1068.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1069. (270): here
  1070.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1071. (335): here
  1072.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1073. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  1074.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1075. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1076.  
  1077. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1078.             function "__half::operator float() const"
  1079.             function "__half::operator short() const"
  1080.             function "__half::operator unsigned short() const"
  1081.             function "__half::operator int() const"
  1082.             function "__half::operator unsigned int() const"
  1083.             function "__half::operator long long() const"
  1084.             function "__half::operator unsigned long long() const"
  1085.             function "__half::operator __nv_bool() const"
  1086.           detected during:
  1087.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1088. (270): here
  1089.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1090. (335): here
  1091.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1092. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(820): here
  1093.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1094. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(982): here
  1095.  
  1096. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(521): error: more than one conversion function from "const tcnn::network_precision_t" to a built-in type applies:
  1097.             function "__half::operator float() const"
  1098.             function "__half::operator short() const"
  1099.             function "__half::operator unsigned short() const"
  1100.             function "__half::operator int() const"
  1101.             function "__half::operator unsigned int() const"
  1102.             function "__half::operator long long() const"
  1103.             function "__half::operator unsigned long long() const"
  1104.             function "__half::operator __nv_bool() const"
  1105.           detected during:
  1106.             instantiation of "void tcnn::add(uint32_t, const T *, T *) [with T=tcnn::network_precision_t]"
  1107. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(154): here
  1108.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1109. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(111): here
  1110.  
  1111. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(521): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1112.             function "__half::operator float() const"
  1113.             function "__half::operator short() const"
  1114.             function "__half::operator unsigned short() const"
  1115.             function "__half::operator int() const"
  1116.             function "__half::operator unsigned int() const"
  1117.             function "__half::operator long long() const"
  1118.             function "__half::operator unsigned long long() const"
  1119.             function "__half::operator __nv_bool() const"
  1120.           detected during:
  1121.             instantiation of "void tcnn::add(uint32_t, const T *, T *) [with T=tcnn::network_precision_t]"
  1122. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(154): here
  1123.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1124. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(111): here
  1125.  
  1126. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(75): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1127.             function "__half::operator float() const"
  1128.             function "__half::operator short() const"
  1129.             function "__half::operator unsigned short() const"
  1130.             function "__half::operator int() const"
  1131.             function "__half::operator unsigned int() const"
  1132.             function "__half::operator long long() const"
  1133.             function "__half::operator unsigned long long() const"
  1134.             function "__half::operator __nv_bool() const"
  1135.           detected during:
  1136.             instantiation of "void tcnn::warp_activation<T,fragment_t>(tcnn::Activation, const fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1137. (245): here
  1138.             instantiation of "void tcnn::kernel_activation(uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1139. (287): here
  1140.             instantiation of "void tcnn::activation_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t]"
  1141. (296): here
  1142.             instantiation of "void tcnn::activation_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1143. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(202): here
  1144.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::forward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1145. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1146.  
  1147. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(75): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1148.             function "__half::operator float() const"
  1149.             function "__half::operator short() const"
  1150.             function "__half::operator unsigned short() const"
  1151.             function "__half::operator int() const"
  1152.             function "__half::operator unsigned int() const"
  1153.             function "__half::operator long long() const"
  1154.             function "__half::operator unsigned long long() const"
  1155.             function "__half::operator __nv_bool() const"
  1156.           detected during:
  1157.             instantiation of "void tcnn::warp_activation<T,fragment_t>(tcnn::Activation, const fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1158. (245): here
  1159.             instantiation of "void tcnn::kernel_activation(uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1160. (287): here
  1161.             instantiation of "void tcnn::activation_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t]"
  1162. (296): here
  1163.             instantiation of "void tcnn::activation_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1164. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(202): here
  1165.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::forward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1166. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1167.  
  1168. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(301): error: name followed by "::" must be a class or namespace name
  1169.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1170. (982): here
  1171.  
  1172. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(302): error: name followed by "::" must be a class or namespace name
  1173.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1174. (982): here
  1175.  
  1176. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(302): error: expected an identifier
  1177.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1178. (982): here
  1179.  
  1180. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(302): error: expected a ";"
  1181.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1182. (982): here
  1183.  
  1184. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(304): error: name followed by "::" must be a class or namespace name
  1185.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1186. (982): here
  1187.  
  1188. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(305): error: name followed by "::" must be a class or namespace name
  1189.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1190. (982): here
  1191.  
  1192. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(305): error: expected an identifier
  1193.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1194. (982): here
  1195.  
  1196. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(305): error: expected a ";"
  1197.           detected during instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1198. (982): here
  1199.  
  1200. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1201.             function "__half::operator float() const"
  1202.             function "__half::operator short() const"
  1203.             function "__half::operator unsigned short() const"
  1204.             function "__half::operator int() const"
  1205.             function "__half::operator unsigned int() const"
  1206.             function "__half::operator long long() const"
  1207.             function "__half::operator unsigned long long() const"
  1208.             function "__half::operator __nv_bool() const"
  1209.           detected during:
  1210.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1211. (270): here
  1212.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1213. (335): here
  1214.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1215. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1216.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1217. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1218.  
  1219. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1220.             function "__half::operator float() const"
  1221.             function "__half::operator short() const"
  1222.             function "__half::operator unsigned short() const"
  1223.             function "__half::operator int() const"
  1224.             function "__half::operator unsigned int() const"
  1225.             function "__half::operator long long() const"
  1226.             function "__half::operator unsigned long long() const"
  1227.             function "__half::operator __nv_bool() const"
  1228.           detected during:
  1229.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1230. (270): here
  1231.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1232. (335): here
  1233.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1234. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1235.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1236. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1237.  
  1238. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1239.             function "__half::operator float() const"
  1240.             function "__half::operator short() const"
  1241.             function "__half::operator unsigned short() const"
  1242.             function "__half::operator int() const"
  1243.             function "__half::operator unsigned int() const"
  1244.             function "__half::operator long long() const"
  1245.             function "__half::operator unsigned long long() const"
  1246.             function "__half::operator __nv_bool() const"
  1247.           detected during:
  1248.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1249. (270): here
  1250.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1251. (335): here
  1252.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1253. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1254.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1255. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1256.  
  1257. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1258.           detected during:
  1259.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None]"
  1260. (860): here
  1261.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1262. (982): here
  1263.  
  1264. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1265.             function "__half::operator float() const"
  1266.             function "__half::operator short() const"
  1267.             function "__half::operator unsigned short() const"
  1268.             function "__half::operator int() const"
  1269.             function "__half::operator unsigned int() const"
  1270.             function "__half::operator long long() const"
  1271.             function "__half::operator unsigned long long() const"
  1272.             function "__half::operator __nv_bool() const"
  1273.           detected during:
  1274.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1275. (270): here
  1276.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1277. (335): here
  1278.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1279. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1280.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1281. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1282.  
  1283. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1284.             function "__half::operator float() const"
  1285.             function "__half::operator short() const"
  1286.             function "__half::operator unsigned short() const"
  1287.             function "__half::operator int() const"
  1288.             function "__half::operator unsigned int() const"
  1289.             function "__half::operator long long() const"
  1290.             function "__half::operator unsigned long long() const"
  1291.             function "__half::operator __nv_bool() const"
  1292.           detected during:
  1293.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1294. (270): here
  1295.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1296. (335): here
  1297.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1298. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1299.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1300. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1301.  
  1302. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1303.             function "__half::operator float() const"
  1304.             function "__half::operator short() const"
  1305.             function "__half::operator unsigned short() const"
  1306.             function "__half::operator int() const"
  1307.             function "__half::operator unsigned int() const"
  1308.             function "__half::operator long long() const"
  1309.             function "__half::operator unsigned long long() const"
  1310.             function "__half::operator __nv_bool() const"
  1311.           detected during:
  1312.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1313. (270): here
  1314.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1315. (335): here
  1316.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1317. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1318.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1319. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1320.  
  1321. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1322.             function "__half::operator float() const"
  1323.             function "__half::operator short() const"
  1324.             function "__half::operator unsigned short() const"
  1325.             function "__half::operator int() const"
  1326.             function "__half::operator unsigned int() const"
  1327.             function "__half::operator long long() const"
  1328.             function "__half::operator unsigned long long() const"
  1329.             function "__half::operator __nv_bool() const"
  1330.           detected during:
  1331.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1332. (270): here
  1333.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1334. (335): here
  1335.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1336. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1337.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1338. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1339.  
  1340. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1341.           detected during:
  1342.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Exponential]"
  1343. (861): here
  1344.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1345. (982): here
  1346.  
  1347. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1348.             function "__half::operator float() const"
  1349.             function "__half::operator short() const"
  1350.             function "__half::operator unsigned short() const"
  1351.             function "__half::operator int() const"
  1352.             function "__half::operator unsigned int() const"
  1353.             function "__half::operator long long() const"
  1354.             function "__half::operator unsigned long long() const"
  1355.             function "__half::operator __nv_bool() const"
  1356.           detected during:
  1357.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1358. (270): here
  1359.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1360. (335): here
  1361.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1362. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1363.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1364. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1365.  
  1366. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1367.             function "__half::operator float() const"
  1368.             function "__half::operator short() const"
  1369.             function "__half::operator unsigned short() const"
  1370.             function "__half::operator int() const"
  1371.             function "__half::operator unsigned int() const"
  1372.             function "__half::operator long long() const"
  1373.             function "__half::operator unsigned long long() const"
  1374.             function "__half::operator __nv_bool() const"
  1375.           detected during:
  1376.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1377. (270): here
  1378.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1379. (335): here
  1380.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1381. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1382.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1383. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1384.  
  1385. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1386.             function "__half::operator float() const"
  1387.             function "__half::operator short() const"
  1388.             function "__half::operator unsigned short() const"
  1389.             function "__half::operator int() const"
  1390.             function "__half::operator unsigned int() const"
  1391.             function "__half::operator long long() const"
  1392.             function "__half::operator unsigned long long() const"
  1393.             function "__half::operator __nv_bool() const"
  1394.           detected during:
  1395.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1396. (270): here
  1397.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1398. (335): here
  1399.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1400. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(267): here
  1401.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1402. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1403.  
  1404. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1405.           detected during:
  1406.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Sigmoid]"
  1407. (862): here
  1408.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1409. (982): here
  1410.  
  1411. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1412.           detected during:
  1413.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::ReLU]"
  1414. (863): here
  1415.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1416. (982): here
  1417.  
  1418. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1419.           detected during:
  1420.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Squareplus]"
  1421. (864): here
  1422.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1423. (982): here
  1424.  
  1425. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1426.           detected during:
  1427.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=256, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Softplus]"
  1428. (865): here
  1429.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=256]"
  1430. (982): here
  1431.  
  1432. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(130): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1433.             function "__half::operator float() const"
  1434.             function "__half::operator short() const"
  1435.             function "__half::operator unsigned short() const"
  1436.             function "__half::operator int() const"
  1437.             function "__half::operator unsigned int() const"
  1438.             function "__half::operator long long() const"
  1439.             function "__half::operator unsigned long long() const"
  1440.             function "__half::operator __nv_bool() const"
  1441.           detected during:
  1442.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1443. (257): here
  1444.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1445. (311): here
  1446.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1447. (320): here
  1448.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1449. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1450.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1451. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1452.  
  1453. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(130): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1454.             function "__half::operator float() const"
  1455.             function "__half::operator short() const"
  1456.             function "__half::operator unsigned short() const"
  1457.             function "__half::operator int() const"
  1458.             function "__half::operator unsigned int() const"
  1459.             function "__half::operator long long() const"
  1460.             function "__half::operator unsigned long long() const"
  1461.             function "__half::operator __nv_bool() const"
  1462.           detected during:
  1463.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1464. (257): here
  1465.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1466. (311): here
  1467.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1468. (320): here
  1469.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1470. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1471.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1472. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1473.  
  1474. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(136): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1475.             function "__half::operator float() const"
  1476.             function "__half::operator short() const"
  1477.             function "__half::operator unsigned short() const"
  1478.             function "__half::operator int() const"
  1479.             function "__half::operator unsigned int() const"
  1480.             function "__half::operator long long() const"
  1481.             function "__half::operator unsigned long long() const"
  1482.             function "__half::operator __nv_bool() const"
  1483.           detected during:
  1484.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1485. (257): here
  1486.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1487. (311): here
  1488.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1489. (320): here
  1490.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1491. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1492.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1493. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1494.  
  1495. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(136): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1496.             function "__half::operator float() const"
  1497.             function "__half::operator short() const"
  1498.             function "__half::operator unsigned short() const"
  1499.             function "__half::operator int() const"
  1500.             function "__half::operator unsigned int() const"
  1501.             function "__half::operator long long() const"
  1502.             function "__half::operator unsigned long long() const"
  1503.             function "__half::operator __nv_bool() const"
  1504.           detected during:
  1505.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1506. (257): here
  1507.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1508. (311): here
  1509.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1510. (320): here
  1511.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1512. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1513.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1514. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1515.  
  1516. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(142): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1517.             function "__half::operator float() const"
  1518.             function "__half::operator short() const"
  1519.             function "__half::operator unsigned short() const"
  1520.             function "__half::operator int() const"
  1521.             function "__half::operator unsigned int() const"
  1522.             function "__half::operator long long() const"
  1523.             function "__half::operator unsigned long long() const"
  1524.             function "__half::operator __nv_bool() const"
  1525.           detected during:
  1526.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1527. (257): here
  1528.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1529. (311): here
  1530.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1531. (320): here
  1532.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1533. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1534.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1535. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1536.  
  1537. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(142): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1538.             function "__half::operator float() const"
  1539.             function "__half::operator short() const"
  1540.             function "__half::operator unsigned short() const"
  1541.             function "__half::operator int() const"
  1542.             function "__half::operator unsigned int() const"
  1543.             function "__half::operator long long() const"
  1544.             function "__half::operator unsigned long long() const"
  1545.             function "__half::operator __nv_bool() const"
  1546.           detected during:
  1547.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1548. (257): here
  1549.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1550. (311): here
  1551.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1552. (320): here
  1553.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1554. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1555.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1556. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1557.  
  1558. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(149): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1559.             function "__half::operator float() const"
  1560.             function "__half::operator short() const"
  1561.             function "__half::operator unsigned short() const"
  1562.             function "__half::operator int() const"
  1563.             function "__half::operator unsigned int() const"
  1564.             function "__half::operator long long() const"
  1565.             function "__half::operator unsigned long long() const"
  1566.             function "__half::operator __nv_bool() const"
  1567.           detected during:
  1568.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1569. (257): here
  1570.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1571. (311): here
  1572.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1573. (320): here
  1574.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1575. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1576.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1577. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1578.  
  1579. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(149): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1580.             function "__half::operator float() const"
  1581.             function "__half::operator short() const"
  1582.             function "__half::operator unsigned short() const"
  1583.             function "__half::operator int() const"
  1584.             function "__half::operator unsigned int() const"
  1585.             function "__half::operator long long() const"
  1586.             function "__half::operator unsigned long long() const"
  1587.             function "__half::operator __nv_bool() const"
  1588.           detected during:
  1589.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1590. (257): here
  1591.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1592. (311): here
  1593.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1594. (320): here
  1595.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1596. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1597.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1598. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1599.  
  1600. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(157): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1601.             function "__half::operator float() const"
  1602.             function "__half::operator short() const"
  1603.             function "__half::operator unsigned short() const"
  1604.             function "__half::operator int() const"
  1605.             function "__half::operator unsigned int() const"
  1606.             function "__half::operator long long() const"
  1607.             function "__half::operator unsigned long long() const"
  1608.             function "__half::operator __nv_bool() const"
  1609.           detected during:
  1610.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1611. (257): here
  1612.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1613. (311): here
  1614.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1615. (320): here
  1616.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1617. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1618.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1619. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1620.  
  1621. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(157): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1622.             function "__half::operator float() const"
  1623.             function "__half::operator short() const"
  1624.             function "__half::operator unsigned short() const"
  1625.             function "__half::operator int() const"
  1626.             function "__half::operator unsigned int() const"
  1627.             function "__half::operator long long() const"
  1628.             function "__half::operator unsigned long long() const"
  1629.             function "__half::operator __nv_bool() const"
  1630.           detected during:
  1631.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1632. (257): here
  1633.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1634. (311): here
  1635.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1636. (320): here
  1637.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1638. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1639.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1640. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1641.  
  1642. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(164): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1643.             function "__half::operator float() const"
  1644.             function "__half::operator short() const"
  1645.             function "__half::operator unsigned short() const"
  1646.             function "__half::operator int() const"
  1647.             function "__half::operator unsigned int() const"
  1648.             function "__half::operator long long() const"
  1649.             function "__half::operator unsigned long long() const"
  1650.             function "__half::operator __nv_bool() const"
  1651.           detected during:
  1652.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1653. (257): here
  1654.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1655. (311): here
  1656.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1657. (320): here
  1658.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1659. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1660.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1661. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1662.  
  1663. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(164): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1664.             function "__half::operator float() const"
  1665.             function "__half::operator short() const"
  1666.             function "__half::operator unsigned short() const"
  1667.             function "__half::operator int() const"
  1668.             function "__half::operator unsigned int() const"
  1669.             function "__half::operator long long() const"
  1670.             function "__half::operator unsigned long long() const"
  1671.             function "__half::operator __nv_bool() const"
  1672.           detected during:
  1673.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1674. (257): here
  1675.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1676. (311): here
  1677.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1678. (320): here
  1679.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1680. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(315): here
  1681.             instantiation of "void tcnn::CutlassResNet<T, input_activation>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, input_activation=tcnn::Activation::None]"
  1682. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu(407): here
  1683.  
  1684. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(75): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1685.             function "__half::operator float() const"
  1686.             function "__half::operator short() const"
  1687.             function "__half::operator unsigned short() const"
  1688.             function "__half::operator int() const"
  1689.             function "__half::operator unsigned int() const"
  1690.             function "__half::operator long long() const"
  1691.             function "__half::operator unsigned long long() const"
  1692.             function "__half::operator __nv_bool() const"
  1693.           detected during:
  1694.             instantiation of "void tcnn::warp_activation<T,fragment_t>(tcnn::Activation, const fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1695. (245): here
  1696.             instantiation of "void tcnn::kernel_activation(uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1697. (287): here
  1698.             instantiation of "void tcnn::activation_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t]"
  1699. (296): here
  1700.             instantiation of "void tcnn::activation_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1701. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(147): here
  1702.             instantiation of "__nv_bool tcnn::compute_layer<CutlassLayer,T>(cudaStream_t, __nv_bool, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with CutlassLayer=tcnn::LastLayer, T=tcnn::network_precision_t]"
  1703. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(161): here
  1704.             instantiation of "__nv_bool tcnn::compute_inference_layer<CutlassLayer,T>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with CutlassLayer=tcnn::LastLayer, T=tcnn::network_precision_t]"
  1705. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(190): here
  1706.             instantiation of "void tcnn::CutlassMLP<T>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t]"
  1707. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(120): here
  1708.  
  1709. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(75): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1710.             function "__half::operator float() const"
  1711.             function "__half::operator short() const"
  1712.             function "__half::operator unsigned short() const"
  1713.             function "__half::operator int() const"
  1714.             function "__half::operator unsigned int() const"
  1715.             function "__half::operator long long() const"
  1716.             function "__half::operator unsigned long long() const"
  1717.             function "__half::operator __nv_bool() const"
  1718.           detected during:
  1719.             instantiation of "void tcnn::warp_activation<T,fragment_t>(tcnn::Activation, const fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1720. (245): here
  1721.             instantiation of "void tcnn::kernel_activation(uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1722. (287): here
  1723.             instantiation of "void tcnn::activation_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, T *) [with T=tcnn::network_precision_t]"
  1724. (296): here
  1725.             instantiation of "void tcnn::activation_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  1726. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(147): here
  1727.             instantiation of "__nv_bool tcnn::compute_layer<CutlassLayer,T>(cudaStream_t, __nv_bool, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with CutlassLayer=tcnn::LastLayer, T=tcnn::network_precision_t]"
  1728. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(161): here
  1729.             instantiation of "__nv_bool tcnn::compute_inference_layer<CutlassLayer,T>(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with CutlassLayer=tcnn::LastLayer, T=tcnn::network_precision_t]"
  1730. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(190): here
  1731.             instantiation of "void tcnn::CutlassMLP<T>::inference_mixed_precision(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrixDynamic<T> &, __nv_bool) [with T=tcnn::network_precision_t]"
  1732. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(120): here
  1733.  
  1734. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1735.           detected during:
  1736.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None]"
  1737. (860): here
  1738.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1739. (983): here
  1740.  
  1741. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1742.           detected during:
  1743.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Exponential]"
  1744. (861): here
  1745.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1746. (983): here
  1747.  
  1748. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1749.           detected during:
  1750.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Sigmoid]"
  1751. (862): here
  1752.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1753. (983): here
  1754.  
  1755. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1756.           detected during:
  1757.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::ReLU]"
  1758. (863): here
  1759.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1760. (983): here
  1761.  
  1762. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1763.           detected during:
  1764.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Squareplus]"
  1765. (864): here
  1766.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1767. (983): here
  1768.  
  1769. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1770.           detected during:
  1771.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=128, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Softplus]"
  1772. (865): here
  1773.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=128]"
  1774. (983): here
  1775.  
  1776. 26 errors detected in the compilation of "/home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_resnet.cu".
  1777. dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/build.make:117: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_resnet.cu.o' failed
  1778. make[2]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_resnet.cu.o] Error 1
  1779. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1780.           detected during:
  1781.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None]"
  1782. (860): here
  1783.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1784. (984): here
  1785.  
  1786. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1787.           detected during:
  1788.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Exponential]"
  1789. (861): here
  1790.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1791. (984): here
  1792.  
  1793. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1794.           detected during:
  1795.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Sigmoid]"
  1796. (862): here
  1797.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1798. (984): here
  1799.  
  1800. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1801.           detected during:
  1802.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::ReLU]"
  1803. (863): here
  1804.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1805. (984): here
  1806.  
  1807. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1808.           detected during:
  1809.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Squareplus]"
  1810. (864): here
  1811.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1812. (984): here
  1813.  
  1814. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  1815.           detected during:
  1816.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=64, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Softplus]"
  1817. (865): here
  1818.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=64]"
  1819. (984): here
  1820.  
  1821. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1822.             function "__half::operator float() const"
  1823.             function "__half::operator short() const"
  1824.             function "__half::operator unsigned short() const"
  1825.             function "__half::operator int() const"
  1826.             function "__half::operator unsigned int() const"
  1827.             function "__half::operator long long() const"
  1828.             function "__half::operator unsigned long long() const"
  1829.             function "__half::operator __nv_bool() const"
  1830.           detected during:
  1831.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1832. (270): here
  1833.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1834. (335): here
  1835.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1836. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1837.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1838. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1839.  
  1840. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(188): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1841.             function "__half::operator float() const"
  1842.             function "__half::operator short() const"
  1843.             function "__half::operator unsigned short() const"
  1844.             function "__half::operator int() const"
  1845.             function "__half::operator unsigned int() const"
  1846.             function "__half::operator long long() const"
  1847.             function "__half::operator unsigned long long() const"
  1848.             function "__half::operator __nv_bool() const"
  1849.           detected during:
  1850.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1851. (270): here
  1852.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1853. (335): here
  1854.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1855. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1856.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1857. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1858.  
  1859. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1860.             function "__half::operator float() const"
  1861.             function "__half::operator short() const"
  1862.             function "__half::operator unsigned short() const"
  1863.             function "__half::operator int() const"
  1864.             function "__half::operator unsigned int() const"
  1865.             function "__half::operator long long() const"
  1866.             function "__half::operator unsigned long long() const"
  1867.             function "__half::operator __nv_bool() const"
  1868.           detected during:
  1869.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1870. (270): here
  1871.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1872. (335): here
  1873.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1874. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1875.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1876. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1877.  
  1878. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(194): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1879.             function "__half::operator float() const"
  1880.             function "__half::operator short() const"
  1881.             function "__half::operator unsigned short() const"
  1882.             function "__half::operator int() const"
  1883.             function "__half::operator unsigned int() const"
  1884.             function "__half::operator long long() const"
  1885.             function "__half::operator unsigned long long() const"
  1886.             function "__half::operator __nv_bool() const"
  1887.           detected during:
  1888.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1889. (270): here
  1890.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1891. (335): here
  1892.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1893. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1894.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1895. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1896.  
  1897. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1898.             function "__half::operator float() const"
  1899.             function "__half::operator short() const"
  1900.             function "__half::operator unsigned short() const"
  1901.             function "__half::operator int() const"
  1902.             function "__half::operator unsigned int() const"
  1903.             function "__half::operator long long() const"
  1904.             function "__half::operator unsigned long long() const"
  1905.             function "__half::operator __nv_bool() const"
  1906.           detected during:
  1907.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1908. (270): here
  1909.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1910. (335): here
  1911.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1912. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1913.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1914. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1915.  
  1916. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(205): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1917.             function "__half::operator float() const"
  1918.             function "__half::operator short() const"
  1919.             function "__half::operator unsigned short() const"
  1920.             function "__half::operator int() const"
  1921.             function "__half::operator unsigned int() const"
  1922.             function "__half::operator long long() const"
  1923.             function "__half::operator unsigned long long() const"
  1924.             function "__half::operator __nv_bool() const"
  1925.           detected during:
  1926.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1927. (270): here
  1928.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1929. (335): here
  1930.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1931. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1932.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1933. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1934.  
  1935. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1936.             function "__half::operator float() const"
  1937.             function "__half::operator short() const"
  1938.             function "__half::operator unsigned short() const"
  1939.             function "__half::operator int() const"
  1940.             function "__half::operator unsigned int() const"
  1941.             function "__half::operator long long() const"
  1942.             function "__half::operator unsigned long long() const"
  1943.             function "__half::operator __nv_bool() const"
  1944.           detected during:
  1945.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1946. (270): here
  1947.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1948. (335): here
  1949.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1950. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1951.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1952. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1953.  
  1954. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(212): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1955.             function "__half::operator float() const"
  1956.             function "__half::operator short() const"
  1957.             function "__half::operator unsigned short() const"
  1958.             function "__half::operator int() const"
  1959.             function "__half::operator unsigned int() const"
  1960.             function "__half::operator long long() const"
  1961.             function "__half::operator unsigned long long() const"
  1962.             function "__half::operator __nv_bool() const"
  1963.           detected during:
  1964.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1965. (270): here
  1966.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1967. (335): here
  1968.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1969. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1970.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1971. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1972.  
  1973. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1974.             function "__half::operator float() const"
  1975.             function "__half::operator short() const"
  1976.             function "__half::operator unsigned short() const"
  1977.             function "__half::operator int() const"
  1978.             function "__half::operator unsigned int() const"
  1979.             function "__half::operator long long() const"
  1980.             function "__half::operator unsigned long long() const"
  1981.             function "__half::operator __nv_bool() const"
  1982.           detected during:
  1983.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  1984. (270): here
  1985.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  1986. (335): here
  1987.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  1988. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  1989.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  1990. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  1991.  
  1992. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(218): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  1993.             function "__half::operator float() const"
  1994.             function "__half::operator short() const"
  1995.             function "__half::operator unsigned short() const"
  1996.             function "__half::operator int() const"
  1997.             function "__half::operator unsigned int() const"
  1998.             function "__half::operator long long() const"
  1999.             function "__half::operator unsigned long long() const"
  2000.             function "__half::operator __nv_bool() const"
  2001.           detected during:
  2002.             instantiation of "void tcnn::warp_activation_backward<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2003. (270): here
  2004.             instantiation of "void tcnn::kernel_activation_backward_output(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2005. (335): here
  2006.             instantiation of "void tcnn::activation_backward_output_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2007. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(282): here
  2008.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2009. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2010.  
  2011. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2012.           detected during:
  2013.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None]"
  2014. (860): here
  2015.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2016. (985): here
  2017.  
  2018. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2019.           detected during:
  2020.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Exponential]"
  2021. (861): here
  2022.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2023. (985): here
  2024.  
  2025. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2026.           detected during:
  2027.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Sigmoid]"
  2028. (862): here
  2029.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2030. (985): here
  2031.  
  2032. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2033.           detected during:
  2034.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::ReLU]"
  2035. (863): here
  2036.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2037. (985): here
  2038.  
  2039. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2040.           detected during:
  2041.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Squareplus]"
  2042. (864): here
  2043.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2044. (985): here
  2045.  
  2046. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2047.           detected during:
  2048.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=32, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Softplus]"
  2049. (865): here
  2050.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=32]"
  2051. (985): here
  2052.  
  2053. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2054.           detected during:
  2055.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::None]"
  2056. (860): here
  2057.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2058. (986): here
  2059.  
  2060. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2061.           detected during:
  2062.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Exponential]"
  2063. (861): here
  2064.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2065. (986): here
  2066.  
  2067. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2068.           detected during:
  2069.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Sigmoid]"
  2070. (862): here
  2071.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2072. (986): here
  2073.  
  2074. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2075.           detected during:
  2076.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::ReLU]"
  2077. (863): here
  2078.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2079. (986): here
  2080.  
  2081. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2082.           detected during:
  2083.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Squareplus]"
  2084. (864): here
  2085.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2086. (986): here
  2087.  
  2088. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu(291): warning: variable "threads" was declared but never referenced
  2089.           detected during:
  2090.             instantiation of "std::enable_if_t<std::is_same<__half, T>::value, void> tcnn::mlp_fused_backward<WIDTH,T,ACTIVATION>(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::RowMajor> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, uint32_t) [with WIDTH=16, T=tcnn::network_precision_t, ACTIVATION=tcnn::Activation::Softplus]"
  2091. (865): here
  2092.             instantiation of "void tcnn::FullyFusedMLP<T, WIDTH>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t, WIDTH=16]"
  2093. (986): here
  2094.  
  2095. 87 errors detected in the compilation of "/home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/fully_fused_mlp.cu".
  2096. dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/build.make:145: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/fully_fused_mlp.cu.o' failed
  2097. make[2]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/fully_fused_mlp.cu.o] Error 1
  2098. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(130): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2099.             function "__half::operator float() const"
  2100.             function "__half::operator short() const"
  2101.             function "__half::operator unsigned short() const"
  2102.             function "__half::operator int() const"
  2103.             function "__half::operator unsigned int() const"
  2104.             function "__half::operator long long() const"
  2105.             function "__half::operator unsigned long long() const"
  2106.             function "__half::operator __nv_bool() const"
  2107.           detected during:
  2108.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2109. (257): here
  2110.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2111. (311): here
  2112.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2113. (320): here
  2114.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2115. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2116.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2117. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2118.  
  2119. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(130): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2120.             function "__half::operator float() const"
  2121.             function "__half::operator short() const"
  2122.             function "__half::operator unsigned short() const"
  2123.             function "__half::operator int() const"
  2124.             function "__half::operator unsigned int() const"
  2125.             function "__half::operator long long() const"
  2126.             function "__half::operator unsigned long long() const"
  2127.             function "__half::operator __nv_bool() const"
  2128.           detected during:
  2129.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2130. (257): here
  2131.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2132. (311): here
  2133.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2134. (320): here
  2135.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2136. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2137.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2138. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2139.  
  2140. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(136): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2141.             function "__half::operator float() const"
  2142.             function "__half::operator short() const"
  2143.             function "__half::operator unsigned short() const"
  2144.             function "__half::operator int() const"
  2145.             function "__half::operator unsigned int() const"
  2146.             function "__half::operator long long() const"
  2147.             function "__half::operator unsigned long long() const"
  2148.             function "__half::operator __nv_bool() const"
  2149.           detected during:
  2150.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2151. (257): here
  2152.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2153. (311): here
  2154.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2155. (320): here
  2156.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2157. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2158.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2159. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2160.  
  2161. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(136): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2162.             function "__half::operator float() const"
  2163.             function "__half::operator short() const"
  2164.             function "__half::operator unsigned short() const"
  2165.             function "__half::operator int() const"
  2166.             function "__half::operator unsigned int() const"
  2167.             function "__half::operator long long() const"
  2168.             function "__half::operator unsigned long long() const"
  2169.             function "__half::operator __nv_bool() const"
  2170.           detected during:
  2171.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2172. (257): here
  2173.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2174. (311): here
  2175.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2176. (320): here
  2177.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2178. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2179.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2180. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2181.  
  2182. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(142): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2183.             function "__half::operator float() const"
  2184.             function "__half::operator short() const"
  2185.             function "__half::operator unsigned short() const"
  2186.             function "__half::operator int() const"
  2187.             function "__half::operator unsigned int() const"
  2188.             function "__half::operator long long() const"
  2189.             function "__half::operator unsigned long long() const"
  2190.             function "__half::operator __nv_bool() const"
  2191.           detected during:
  2192.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2193. (257): here
  2194.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2195. (311): here
  2196.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2197. (320): here
  2198.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2199. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2200.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2201. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2202.  
  2203. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(142): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2204.             function "__half::operator float() const"
  2205.             function "__half::operator short() const"
  2206.             function "__half::operator unsigned short() const"
  2207.             function "__half::operator int() const"
  2208.             function "__half::operator unsigned int() const"
  2209.             function "__half::operator long long() const"
  2210.             function "__half::operator unsigned long long() const"
  2211.             function "__half::operator __nv_bool() const"
  2212.           detected during:
  2213.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2214. (257): here
  2215.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2216. (311): here
  2217.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2218. (320): here
  2219.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2220. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2221.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2222. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2223.  
  2224. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(149): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2225.             function "__half::operator float() const"
  2226.             function "__half::operator short() const"
  2227.             function "__half::operator unsigned short() const"
  2228.             function "__half::operator int() const"
  2229.             function "__half::operator unsigned int() const"
  2230.             function "__half::operator long long() const"
  2231.             function "__half::operator unsigned long long() const"
  2232.             function "__half::operator __nv_bool() const"
  2233.           detected during:
  2234.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2235. (257): here
  2236.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2237. (311): here
  2238.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2239. (320): here
  2240.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2241. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2242.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2243. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2244.  
  2245. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(149): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2246.             function "__half::operator float() const"
  2247.             function "__half::operator short() const"
  2248.             function "__half::operator unsigned short() const"
  2249.             function "__half::operator int() const"
  2250.             function "__half::operator unsigned int() const"
  2251.             function "__half::operator long long() const"
  2252.             function "__half::operator unsigned long long() const"
  2253.             function "__half::operator __nv_bool() const"
  2254.           detected during:
  2255.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2256. (257): here
  2257.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2258. (311): here
  2259.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2260. (320): here
  2261.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2262. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2263.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2264. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2265.  
  2266. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(157): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2267.             function "__half::operator float() const"
  2268.             function "__half::operator short() const"
  2269.             function "__half::operator unsigned short() const"
  2270.             function "__half::operator int() const"
  2271.             function "__half::operator unsigned int() const"
  2272.             function "__half::operator long long() const"
  2273.             function "__half::operator unsigned long long() const"
  2274.             function "__half::operator __nv_bool() const"
  2275.           detected during:
  2276.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2277. (257): here
  2278.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2279. (311): here
  2280.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2281. (320): here
  2282.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2283. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2284.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2285. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2286.  
  2287. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(157): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2288.             function "__half::operator float() const"
  2289.             function "__half::operator short() const"
  2290.             function "__half::operator unsigned short() const"
  2291.             function "__half::operator int() const"
  2292.             function "__half::operator unsigned int() const"
  2293.             function "__half::operator long long() const"
  2294.             function "__half::operator unsigned long long() const"
  2295.             function "__half::operator __nv_bool() const"
  2296.           detected during:
  2297.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2298. (257): here
  2299.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2300. (311): here
  2301.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2302. (320): here
  2303.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2304. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2305.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2306. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2307.  
  2308. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(164): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2309.             function "__half::operator float() const"
  2310.             function "__half::operator short() const"
  2311.             function "__half::operator unsigned short() const"
  2312.             function "__half::operator int() const"
  2313.             function "__half::operator unsigned int() const"
  2314.             function "__half::operator long long() const"
  2315.             function "__half::operator unsigned long long() const"
  2316.             function "__half::operator __nv_bool() const"
  2317.           detected during:
  2318.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2319. (257): here
  2320.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2321. (311): here
  2322.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2323. (320): here
  2324.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2325. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2326.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2327. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2328.  
  2329. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/include/tiny-cuda-nn/common_device.h(164): error: more than one conversion function from "tcnn::network_precision_t" to a built-in type applies:
  2330.             function "__half::operator float() const"
  2331.             function "__half::operator short() const"
  2332.             function "__half::operator unsigned short() const"
  2333.             function "__half::operator int() const"
  2334.             function "__half::operator unsigned int() const"
  2335.             function "__half::operator long long() const"
  2336.             function "__half::operator unsigned long long() const"
  2337.             function "__half::operator __nv_bool() const"
  2338.           detected during:
  2339.             instantiation of "void tcnn::warp_activation_backward_in<T,fragment_t,forward_fragment_t>(tcnn::Activation, const fragment_t &, const forward_fragment_t &, fragment_t &) [with T=tcnn::network_precision_t, fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>, forward_fragment_t=tcnn::vector_fragment_t<tcnn::network_precision_t, 8U>]"
  2340. (257): here
  2341.             instantiation of "void tcnn::kernel_activation_backward(uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t, N=8U]"
  2342. (311): here
  2343.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, uint32_t, tcnn::Activation, const T *, const T *, T *) [with T=tcnn::network_precision_t]"
  2344. (320): here
  2345.             instantiation of "void tcnn::activation_backward_gpu(cudaStream_t, tcnn::Activation, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrixDynamic<T> &) [with T=tcnn::network_precision_t]"
  2346. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(334): here
  2347.             instantiation of "void tcnn::CutlassMLP<T>::backward(cudaStream_t, const tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> &, const tcnn::GPUMatrixDynamic<T> &, const tcnn::GPUMatrixDynamic<T> &, tcnn::GPUMatrix<T, tcnn::MatrixLayout::ColumnMajor> *, __nv_bool, __nv_bool) [with T=tcnn::network_precision_t]"
  2348. /home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu(452): here
  2349.  
  2350. 24 errors detected in the compilation of "/home/ubuntu/instant-ngp/dependencies/tiny-cuda-nn/src/cutlass_mlp.cu".
  2351. dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/build.make:103: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_mlp.cu.o' failed
  2352. make[2]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/cutlass_mlp.cu.o] Error 1
  2353. CMakeFiles/Makefile2:305: recipe for target 'dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/all' failed
  2354. make[1]: *** [dependencies/tiny-cuda-nn/src/CMakeFiles/tiny-cuda-nn.dir/all] Error 2
  2355. Makefile:90: recipe for target 'all' failed
  2356. make: *** [all] Error 2
  2357. (tensorflow2_p38) ubuntu@ip-172-31-40-250:~/instant-ngp$
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!