search

ROCm / pytorch

Tensors and Dynamic neural networks in Python with strong GPU acceleration
http://pytorch.org
Other
219 stars 50 forks source link

[NO CP] Update the hipsparse sampled addmm condition for release/2.2 #1414

Closed dnikolaev-amd closed 2 months ago

dnikolaev-amd commented 2 months ago

fix sparse_sampled_addmm_cuda on rocm use AT_USE_HIPSPARSE_GENERIC_API instead of AT_USE_HIPSPARSE_GENERIC_52_API Cherry-picked form https://github.com/ROCm/pytorch/commit/e6e5c48a2eb940a84598c69a4f03bcf8811b6bb6 except .ci/docker/common/install_rocm_magma.sh Fixes: https://ontrack-internal.amd.com/browse/SWDEV-452817 (partial, only test_ops) https://ontrack-internal.amd.com/browse/SWDEV-450965

Built successful test_fake_crossref_backward_no_amp_sparse_sampled_addmm_cuda_float32 passed

# python test/test_ops.py -v -k test_fake_crossref_backward_no_amp_sparse_sampled_addmm_cuda_float32
Test results will be stored in test-reports/python-unittest/test.test_ops

Running tests...
----------------------------------------------------------------------
  test_fake_crossref_backward_no_amp_sparse_sampled_addmm_cuda_float32 (__main__.TestFakeTensorCUDA) ... /opt/conda/envs/py_3.9/lib/python3.9/site-packages/torch/testing/_internal/common_methods_invocations.py:1186: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /var/lib/jenkins/pytorch/aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
  torch.eye(m, n, device=device, dtype=dtype)
ok (1.095s)

----------------------------------------------------------------------
Ran 1 test in 1.400s

OK