pytorch icon indicating copy to clipboard operation
pytorch copied to clipboard

[ROCm] Improve softmax performance.

Open doru1004 opened this issue 1 year ago • 41 comments

This patch improves the performance of softmax for 2D tensors by:

  • using a softmax calculation which eliminates the increase of shared memory usage with the size of the tensor and relies on global memory accesses for the tensor data accesses while still using shared memory for the actual reduction step (the shared memory used for the reduction is constant and does not increase with tensor size).
  • for the final computation replacing the division by the sum with the multiplication of 1/sum. The 1/sum is computed as the last step of the warp reduction.
  • replace the use of the exp function with the __expf function.

The impact on numerical accuracy is within a 1e-5 for half precision and 1e-7 for full precision.

The impact on performance for MI300X is between 22% and 50% percentage improvement over current runtimes.

doru1004 avatar Nov 21 '24 18:11 doru1004

@doru1004 Plz run pytorch UTs as well.

UTs that reference softmax: https://github.com/search?q=repo%3Apytorch%2Fpytorch+path%3A%2F%5Etest%5C%2F%2F+softmax&type=code

After that you can also try to run the full pytorch UT test suite CONTINUE_THROUGH_ERROR=1 .ci/pytorch/test.sh

jerrymannil avatar Nov 21 '24 19:11 jerrymannil

Jenkins build for 211a9130ea8724d3b356368e73fd64a9a3157c83 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

Warning: Unused direct dependencies:
	/var/lib/jenkins/pytorch/build/lib/libshm.so
	/opt/rocm/lib/libhsa-runtime64.so.1
	/lib/x86_64-linux-gnu/libm.so.6
[7981/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/./torch_hip_generated_flash_api.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/flash_attn/flash_api.hip:57:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Nov 21 '24 20:11 okakarpa

@jerrymannil , You can trigger a CI build for the branch - improve-softmax-performance-2.4 and verify the UTs.

pruthvistony avatar Nov 21 '24 21:11 pruthvistony

Jenkins build for 211a9130ea8724d3b356368e73fd64a9a3157c83 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7940/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_WeightNorm.hip.o
[7941/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_scaled_modified_bessel_k0.hip.o
[7942/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_scaled_modified_bessel_k1.hip.o
[7943/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_modified_bessel_k1.hip.o
[7944/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/./torch_hip_generated_flash_api.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/flash_attn/torch_hip_generated_flash_api.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/flash_attn/flash_api.hip:57:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Nov 22 '24 18:11 okakarpa

Jenkins build for 211a9130ea8724d3b356368e73fd64a9a3157c83 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7943/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adam_impl.hip.o
[7944/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_modified_bessel_k0.hip.o
[7945/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adam_amsgrad_impl.hip.o
[7946/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adamw_amsgrad_impl.hip.o
[7947/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention.hip:84:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Nov 23 '24 10:11 okakarpa

Jenkins build for 211a9130ea8724d3b356368e73fd64a9a3157c83 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

	/opt/rocm/lib/libhsa-runtime64.so.1
	/lib/x86_64-linux-gnu/libm.so.6
[7981/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_BinaryMiscOpsKernels.hip.o
[7982/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_UnaryGeometricAcoshKernel.hip.o
[7983/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention_backward.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention_backward.hip:49:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Nov 26 '24 06:11 okakarpa

Jenkins build for 211a9130ea8724d3b356368e73fd64a9a3157c83 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

	/var/lib/jenkins/pytorch/build/lib/libshm.so
	/opt/rocm/lib/libhsa-runtime64.so.1
	/lib/x86_64-linux-gnu/libm.so.6
[7981/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_shifted_chebyshev_polynomial_w.hip.o
[7982/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention.hip:84:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Nov 26 '24 06:11 okakarpa

Jenkins build for 19324cb093099ea7d73b28f8a2155ae6b2d0280e commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7866/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_ActivationMishKernel.hip.o
[7867/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_RecordStream.hip.o
[7868/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_MultinomialKernel.hip.o
[7869/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_MultiLabelMarginCriterion.hip.o
[7870/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/./torch_hip_generated_SoftMax.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o.cmake
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/SoftMax.hip:1206:3: error: no matching function for call to 'host_softmax'
 1206 |   host_softmax<LogSoftMaxForwardEpilogue,true>(input, dim, half_to_float, output);
      |   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/SoftMax.hip:940:8: note: candidate template ignored: invalid explicitly-specified argument for template parameter 'EpilogueWithMul'

okakarpa avatar Nov 27 '24 17:11 okakarpa

Jenkins build for 19324cb093099ea7d73b28f8a2155ae6b2d0280e commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

22 warnings generated when compiling for host.
[7866/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_Reduce.hip.o
[7867/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_RecordStream.hip.o
[7868/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_MultinomialKernel.hip.o
[7869/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/./torch_hip_generated_SoftMax.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o.cmake
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/SoftMax.hip:1206:3: error: no matching function for call to 'host_softmax'
 1206 |   host_softmax<LogSoftMaxForwardEpilogue,true>(input, dim, half_to_float, output);
      |   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/SoftMax.hip:940:8: note: candidate template ignored: invalid explicitly-specified argument for template parameter 'EpilogueWithMul'

okakarpa avatar Nov 27 '24 18:11 okakarpa

Jenkins build for 5515a5aabcd2138110071c46419f695b16631db5 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7940/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_ForeachBinaryOpScalarList.hip.o
[7941/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_WeightNorm.hip.o
[7942/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_scaled_modified_bessel_k0.hip.o
[7943/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adam_amsgrad_impl.hip.o
[7944/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention_backward.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention_backward.hip:49:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Dec 02 '24 18:12 okakarpa

Jenkins build for da48662953ffe5890a4baec41fc0454a0e26f477 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

/opt/rocm-6.2.3/lib/llvm/bin/../../../include/hip/hip_runtime_api.h:580:41: note: expanded from macro 'DEPRECATED'
  580 | #define DEPRECATED(msg) __attribute__ ((deprecated(msg)))
      |                                         ^
1 warning generated when compiling for gfx908.
[7837/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/./torch_hip_generated_SoftMax.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_SoftMax.hip.o.cmake
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/SoftMax.hip:976:15: error: use of undeclared identifier 'ekse'; did you mean 'else'?
  976 |             } ekse {
      |               ^~~~
      |               else

okakarpa avatar Dec 02 '24 22:12 okakarpa

Jenkins build for da48662953ffe5890a4baec41fc0454a0e26f477 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/DistributionTemplates.h:182:17: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/var/lib/jenkins/pytorch/aten/src/ATen/native/hip/DistributionTemplates.h:182:17: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
12 warnings generated when compiling for gfx942.
[7982/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_UnaryGeometricAcoshKernel.hip.o
[7983/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention.hip:84:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Dec 02 '24 22:12 okakarpa

Jenkins build for 5a4ae3de47cf3499a6d55025e3733e3bd49a8a8b commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

	/var/lib/jenkins/pytorch/build/lib/libshm.so
	/opt/rocm/lib/libhsa-runtime64.so.1
	/lib/x86_64-linux-gnu/libm.so.6
[7981/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_spherical_bessel_j0.hip.o
[7982/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention.hip:84:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

okakarpa avatar Dec 03 '24 01:12 okakarpa

Jenkins build for 5a4ae3de47cf3499a6d55025e3733e3bd49a8a8b commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

	/lib/x86_64-linux-gnu/libm.so.6
[7979/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_BinaryMiscOpsKernels.hip.o
[7980/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_spherical_bessel_j0.hip.o
[7981/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_shifted_chebyshev_polynomial_w.hip.o
[7982/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention_backward.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention_backward.hip:49:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

rocm-mici avatar Dec 03 '24 18:12 rocm-mici

Jenkins build for 5a4ae3de47cf3499a6d55025e3733e3bd49a8a8b commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7936/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_modified_bessel_k1.hip.o
[7937/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_modified_bessel_i0.hip.o
[7938/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_modified_bessel_k0.hip.o
[7939/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_legendre_polynomial_p.hip.o
[7940/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention.hip:84:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

rocm-mici avatar Dec 04 '24 06:12 rocm-mici

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Detected error during Pytorch building:

[7941/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adam_amsgrad_impl.hip.o
[7942/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adamw_amsgrad_impl.hip.o
[7943/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_ActivationGeluKernel.hip.o
[7944/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/hip/torch_hip_generated_fused_adam_impl.hip.o
[7945/8635] Building HIPCC object caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o
FAILED: caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o 
cd /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip && /opt/conda/envs/py_3.10/bin/cmake -E make_directory /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/. && /opt/conda/envs/py_3.10/bin/cmake -D verbose:BOOL=OFF -D build_configuration:STRING=RELEASE -D generated_file:STRING=/var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/./torch_hip_generated_attention_backward.hip.o -P /var/lib/jenkins/pytorch/build/caffe2/CMakeFiles/torch_hip.dir/__/aten/src/ATen/native/transformers/hip/torch_hip_generated_attention_backward.hip.o.cmake
In file included from /var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/attention_backward.hip:49:
/var/lib/jenkins/pytorch/aten/src/ATen/native/transformers/hip/aotriton_adapter.h:120:10: error: no matching constructor for initialization of 'aotriton::TensorView<0>'
  120 |   return aotriton::TensorView<0>(reinterpret_cast<intptr_t>(q.data_ptr()),
      |          ^                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as ABORTED Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as SUCCESS Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as ABORTED Links: Blue Ocean view / Build artifacts

Jenkins build for 70a25e21a46ec6f8685e0b86d1a975f4aa248409 commit finished as FAILURE Links: Blue Ocean view / Build artifacts