Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/67050
This PR moves init_multi_gpu_helper to common_distributed so that it could be shared by different distributed tests.
ghstack-source-id: 141370119
Test Plan: wait for ci.
Reviewed By: mrshenli
Differential Revision: D31842644
fbshipit-source-id: c7bad25d6cef9bdce7ad1fb6c60c1cad4b765702
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66904
Add more FSDP unit tests to cover core logic, freezing weights and flatten parameter wrappe, these unit tests are refactored to be aligned with PyTorch commonly used test classes
ghstack-source-id: 141335614
Test Plan: unit tests
Reviewed By: mrshenli
Differential Revision: D31779565
fbshipit-source-id: c727110d1d7570c0ec49e42cadfc9e9a5e440073
Summary:
CAFFE2 has been deprecated for a while, but still included in every PyTorch build.
We should stop building it by default, although CI should still validate that caffe2 code is buildable.
Build even fewer dependencies when compiling mobile builds without Caffe2
Introduce `TEST_CAFFE2` in torch.common.utils
Skip `TestQuantizedEmbeddingOps` and `TestJit.test_old_models_bc` is code is compiled without Caffe2
Should be landed after https://github.com/pytorch/builder/pull/864
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66658
Reviewed By: driazati, seemethere, janeyx99
Differential Revision: D31669156
Pulled By: malfet
fbshipit-source-id: 1cc45e2d402daf913a4685eb9f841cc3863e458d
Summary:
Adds `torch.argwhere` as an alias to `torch.nonzero`
Currently, `torch.nonzero` is actually provides equivalent functionality to `np.argwhere`.
From NumPy docs,
> np.argwhere(a) is almost the same as np.transpose(np.nonzero(a)), but produces a result of the correct shape for a 0D array.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/64257
Reviewed By: dagitses
Differential Revision: D31474901
Pulled By: saketh-are
fbshipit-source-id: 335327a4986fa327da74e1fb8624cc1e56959c70
Summary:
Fixes https://github.com/pytorch/pytorch/issues/62533.
In very rare cases, the decorator for detecting memory leak is throwing assertion, even when the test is passing, and the memory is being freed with a tiny delay. The issue is not being reproduced in internal testing, but shows up sometimes in CI environment.
Reducing the severity of such detection to warning, so as not to fail the CI tests, as the actual test is not failing, rather only the check inside the decorator is failing.
Limiting the change to ROCM only for now.
cc jeffdaily sunway513 jithunnair-amd ROCmSupport
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65973
Reviewed By: anjali411
Differential Revision: D31776154
Pulled By: malfet
fbshipit-source-id: 432199fca17669648463c4177c62adb553cacefd
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66554
In native_functions.yaml, the schemas for batch_norm and instance_norm
are incorrect: the inputs `running_mean` and `running_var` are mutated,
but are not marked as such in the function schema. Since `(a!)?`
annotations are currently not working (see #65760), this instead adds a
special case to `alias_anaysis.cpp`. If the value of `training` or
`use_input_stats` is known to be `false`, then `alias_analysis` will
mark the input as _not_ being written to.
Test Plan:
Removed the `skip` annotation on the following test, and added a special
exception in `check_alias_annotations`:
```
python test/test_ops.py -k test_variant_consistency_jit_nn_functional_batch_norm
```
Also:
```
./build/bin/test_jit --gtest_filter="*BatchAndInstanceNormFixture*"
```
Imported from OSS
Reviewed By: eellison
Differential Revision: D31612339
fbshipit-source-id: 12ca61b782b9e41e06883ba080a276209dc435bb
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66603
Found the issue here: https://github.com/pytorch/pytorch/issues/66281 by make the test cases more complicated.
By closely reading the code again, it turns out my original understanding is also wrong. Let's use the example mentioned in the issue to explain:
If the placement is like:
```
"rank:3/cuda:3",
"rank:0/cuda:0",
"rank:1/cuda:1",
"rank:2/cuda:2",
```
First, we split the column or row by the order of [3, 0, 1, 2].
In the case of column-wise sharding:
We get to reaggrage the result from rank0-4.
Step 1: we split the output based on the original sharding strategy, aka, rank3 gets the 1st shard, rank0 get the 2nd shard, etc.
Step 2: we need to rearrange the result from rank0-4 by ordering them following the order of [3, 0, 1, 2], aka, the result from rank3 needs to be put in the front, and so forth.
In the case of row-wise sharding:
We need to rearrange the input being sent to rank0-4.
Step 1: we reorder the input and follow the map of [3, 0, 1, 2]. For example, the first shard goes to rank 3 so we need to put in the 3rd part, the second shard goes to rank 0, so we put it in the 2nd part, and so on.
Step 2: the size of the sharding for each rank is decided by the original placement: [3, 0, 1, 2], aka, rank 3 gets the first shard and its size, etc.
Update the unit test to reflect this change.
Also, correct some format and comments in the sharded linear.
ghstack-source-id: 141055689
Test Plan: unit test and wait for CI.
Reviewed By: pritamdamania87, bowangbj
Differential Revision: D31634590
fbshipit-source-id: 677a9c2b42da1e2c63220523ed2c004565bbecc7
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66680
Closes https://github.com/pytorch/pytorch/issues/66215. Tracks models
with sync BN so we can find workflows that use them and target for perf
optimization.
ghstack-source-id: 140875182
Test Plan: CI
Reviewed By: pritamdamania87
Differential Revision: D31679477
fbshipit-source-id: 0e68cd1a7aabbc5b26227895c53d33b8e98bfb8e
Summary:
Skip failing tests when LAPACK and MAGMA are not available for ` test_linalg.py` and ` test_ops.py`.
Note that there's no CI without LAPACK or MAGMA. I verified locally that now it works as expected, but in the future we have no guards against tests failing again for this situation.
<details>
<summary> test_ops.py failures that are fixed</summary>
```
FAILED test/test_ops.py::TestCommonCPU::test_out_linalg_tensorinv_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_reference_testing_linalg_tensorinv_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_reference_testing_linalg_tensorinv_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_variant_consistency_eager_linalg_tensorinv_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_variant_consistency_eager_linalg_tensorinv_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_variant_consistency_eager_triangular_solve_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestCommonCPU::test_variant_consistency_eager_triangular_solve_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_grad_linalg_tensorinv_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_grad_linalg_tensorinv_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_grad_triangular_solve_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_grad_triangular_solve_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_linalg_tensorinv_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_linalg_tensorinv_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_triangular_solve_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_triangular_solve_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_forward_mode_AD_linalg_tensorinv_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_forward_mode_AD_linalg_tensorinv_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_forward_mode_AD_triangular_solve_cpu_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestGradientsCPU::test_forward_mode_AD_triangular_solve_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestJitCPU::test_variant_consistency_jit_linalg_tensorinv_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestJitCPU::test_variant_consistency_jit_triangular_solve_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestJitCPU::test_variant_consistency_jit_triangular_solve_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestMathBitsCPU::test_conj_view_linalg_tensorinv_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestMathBitsCPU::test_conj_view_triangular_solve_cpu_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestMathBitsCPU::test_neg_view_linalg_tensorinv_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_ops.py::TestMathBitsCPU::test_neg_view_triangular_solve_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
```
</details>
<details>
<summary> test_linalg.py failures that are fixed</summary>
```
FAILED test/test_linalg.py::TestLinalgCPU::test_norm_dtype_cpu - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCPU::test_norm_matrix_cpu_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCPU::test_norm_matrix_cpu_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCPU::test_nuclear_norm_axes_small_brute_force_old_cpu - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_eigh_hermitian_grad_meta_complex128 - RuntimeError: Calling torch.linalg.eigh or eigvalsh on a CPU tensor requires compiling PyTorch with LAPACK. Please use PyTorch built with LAPACK support.
FAILED test/test_linalg.py::TestLinalgMETA::test_eigh_hermitian_grad_meta_float64 - RuntimeError: Calling torch.linalg.eigh or eigvalsh on a CPU tensor requires compiling PyTorch with LAPACK. Please use PyTorch built with LAPACK support.
FAILED test/test_linalg.py::TestLinalgMETA::test_inverse_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_inverse_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_inverse_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_inverse_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_broadcasting_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_broadcasting_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_broadcasting_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_broadcasting_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_non_contiguous_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_non_contiguous_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_non_contiguous_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_batched_non_contiguous_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_lu_solve_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_broadcasting_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_broadcasting_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_broadcasting_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_broadcasting_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_non_contiguous_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_non_contiguous_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_non_contiguous_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_batched_non_contiguous_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_old_solve_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_batched_non_contiguous_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_batched_non_contiguous_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_batched_non_contiguous_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_batched_non_contiguous_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_solve_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_col_maj_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_col_maj_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_meta_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_meta_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_square_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_all_col_maj_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_all_col_maj_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_all_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_all_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_some_col_maj_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_some_col_maj_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_some_meta_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgMETA::test_svd_tall_some_meta_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_inverse_cuda_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_inverse_cuda_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_inverse_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_inverse_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_lowrank_cuda_float64 - RuntimeError: Calling torch.lu on a CUDA tensor requires compiling PyTorch with MAGMA. lease rebuild with MAGMA.
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_col_maj_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_col_maj_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_cuda_complex128 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_cuda_complex64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_square_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_all_col_maj_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_all_col_maj_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_all_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_all_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_some_col_maj_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_some_col_maj_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_some_cuda_float32 - RuntimeError: svd: LAPACK library not found in compilation
FAILED test/test_linalg.py::TestLinalgCUDA::test_svd_tall_some_cuda_float64 - RuntimeError: svd: LAPACK library not found in compilation
```
</details>
Fixes https://github.com/pytorch/pytorch/issues/59662
cc mruberry jianyuh nikitaved pearu walterddr IvanYashchuk xwang233 Lezcano
Pull Request resolved: https://github.com/pytorch/pytorch/pull/64930
Reviewed By: zou3519
Differential Revision: D31739416
Pulled By: mruberry
fbshipit-source-id: 153c40d8eeeb094b06816882a7cbb28c681509a9
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66798
get_cycles_per_ms is copied and used in a few places, move it to common_utils so that it can be used as a shared util function
ghstack-source-id: 140790599
Test Plan: unit tests
Reviewed By: pritamdamania87
Differential Revision: D31706870
fbshipit-source-id: e8dccecb13862646a19aaadd7bad7c8f414fd4ab
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/64181
This PR replaces all the calls to:
- `transpose(-2, -1)` or `transpose(-1, -2)` by `mT()` in C++ and `mT` in Python
- `conj().transpose(-2, -1)` or `transpose(-2, -1).conj()` or `conj().transpose(-1, -2)` or `transpose(-1, -2).conj()` by `mH()` in C++ and `mH` in Python.
It also simplifies two pieces of code, and fixes one bug where a pair
of parentheses were missing in the function `make_symmetric_matrices`.
Test Plan: Imported from OSS
Reviewed By: H-Huang
Differential Revision: D31692896
Pulled By: anjali411
fbshipit-source-id: e9112c42343663d442dc5bd53ff2b492094b434a
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65855
This adjusted our test base to support non-nccl backend like gloo/mpi, so that we could test sharding on CPU with gloo/mpi backend.
ghstack-source-id: 140840866
Test Plan: wait for the CI for existing tests, also adding tests in the stacked diff above.
Reviewed By: pritamdamania87, bowangbj
Differential Revision: D31287162
fbshipit-source-id: d48dfc8ef886a4d34b1de42f3ce6b600b5c9a617
Summary:
Fixes https://github.com/pytorch/pytorch/issues/64883
Adds a `warn_only` kwarg to `use_deterministic_algorithms`. When enabled, calling an operation that does not have a deterministic implementation will raise a warning, rather than an error.
`torch.testing._internal.common_device_type.expectedAlertNondeterministic` is also refactored and documented in this PR to make it easier to use and understand.
cc mruberry kurtamohler
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66233
Reviewed By: bdhirsh
Differential Revision: D31616481
Pulled By: mruberry
fbshipit-source-id: 059634a82d54407492b1d8df08f059c758d0a420
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65772
Looking at some workloads and it would be useful to have this info.
ghstack-source-id: 140555200
Test Plan: CI
Reviewed By: zhaojuanmao, wayi1
Differential Revision: D31224417
fbshipit-source-id: 14eeb053aced87c7ca43b6879f81f54bd0a42b76
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65771
Fixes some logging around monitored_barrier to make it cleaner.
ghstack-source-id: 140555204
Test Plan: CI
Reviewed By: zhaojuanmao, wayi1
Differential Revision: D31222881
fbshipit-source-id: 77d6f072ce98a9b31192e0d48ea0f8cbd8f216fe
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66393
Third try!
Fixes:
- test_nccl_timeout can be flaky because of 1s timeout, bump up the timeout to resolve the flakiness. But in general we should not have been relying on time.sleep for this test, filed https://github.com/pytorch/pytorch/issues/66354 to track that.
- ciflow/all did not actually run tests due to a bug causing multigpu tests to not be run. This has since been fixed.
ghstack-source-id: 140560113
Test Plan: CI
Reviewed By: mrshenli
Differential Revision: D31534735
fbshipit-source-id: 8b7e0f4fed3972b7a77cbcda28876c9eefb0c7e2
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66600
Sparse RPC functionality added in
https://github.com/pytorch/pytorch/pull/62794 works only for TensorPipe and is
broken for other agent types.
Moving these tests to a TensorPipe only class.
ghstack-source-id: 140553147
Test Plan: waitforbuildbot
Reviewed By: rohan-varma
Differential Revision: D31633305
fbshipit-source-id: 37d94cb9ed5565a72a6d512c2a9db75a497d5b95
Summary:
All of the pooling modules except MaxUnpool and LPPool return either a
Tensor or [Tensor, Tensor]. The current type annotations are inaccurate,
and prevent scripting the module if return_indices is set as True in the
module.
There's not a great way to make this agree with mypy because the
overload is dependent on the value of return_indices, an attribute.
I tried changing the annotations from `Tensor` to
`Union[Tensor, Tuple[Tensor, Tensor]]`, but that breaks a bunch of uses
that have return_indices=False.
For example, this breaks:
4e94e84f65/torch/nn/modules/container.py (L139)
Also clean up how test names were being constructed in test_jit, since
otherwise we were getting name collisions when there were two tests on
the same nn.Module.
Fixes https://github.com/pytorch/pytorch/issues/45904
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65847
Reviewed By: ZolotukhinM
Differential Revision: D31462517
Pulled By: eellison
fbshipit-source-id: 6f9e8df1be6c75e5e1e9bae07cf3ad3603ba59bd
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/64282
OpInfos for:
- Tensor.bfloat16, Tensor.bool, Tensor.bypte, Tensor.char
- Tensor.double, Tensor.float, Tensor.half, Tensor.int
- Tensor.short, Tensor.long
None of these are supported by TorchScript. Also, the OpInfo autograd
test runner assumes that the operation is not allowed to change the
dtype of the argument, so only Tensor.double has
`supports_autograd=True` (in theory Tensor.bfloat16, Tensor.float,
Tensor.half should be differentiable).
Test Plan: - run tests
Reviewed By: dagitses
Differential Revision: D31452627
Pulled By: zou3519
fbshipit-source-id: b7f272e558558412c47aefe947af7f060dfb45c5
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66579
Didn't commit this file in the PR that open sources fx2trt tests
Test Plan: ci
Reviewed By: 842974287
Differential Revision: D31623354
fbshipit-source-id: 6cedbe0f229da40499b83e6df28e16caca392d9c
Summary:
Fixes https://github.com/pytorch/pytorch/issues/20972
log_sigmoid calculates something like `log(1 + x)` where x is always a
positive number less than one. This wastes floating point precision
because the exponent always becomes zero. Instead, using
`log1p(x)` gives the full mantissa precision around `x=0`.
This also fixes infinity propagation because the old code does,
`exp(in - in)` when `in` is negative. Which for infinity, results in a
NaN instead of 0.
cc albanD mruberry jbschlosser walterddr
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66441
Reviewed By: bdhirsh
Differential Revision: D31619630
Pulled By: albanD
fbshipit-source-id: e7867f3459a91e944b92f8ca42b6e0697b13f89b
