# Motivation
As mentioned in [[RFC] Intel GPU Runtime Upstreaming](https://github.com/pytorch/pytorch/issues/114842), The first runtime component we would like to upstream is `Device` which contains the device management functions of Intel GPU's runtime. To facilitate the code review, we split the code changes into 4 PRs. This is one of the 4 PRs and covers the changes under `c10`.
# Design
Intel GPU device is a wrapper of sycl device on which kernels can be executed. In our design, we will maintain a sycl device pool containing all the GPU devices of the current machine, and manage the status of the device pool by PyTorch. The thread local safe is considered in this design. The corresponding C++ files related to `Device` will be placed in c10/xpu folder. And we provide the c10 device runtime APIs, like
- `c10::xpu::device_count`
- `c10::xpu::set_device`
- ...
# Additional Context
In our plan, 4 PRs should be submitted to PyTorch for `Device`:
1. for c10
2. for aten
3. for python frontend
4. for lazy initialization shared with CUDA
Pull Request resolved: https://github.com/pytorch/pytorch/pull/116019
Approved by: https://github.com/gujinghui, https://github.com/jgong5, https://github.com/EikanWang, https://github.com/malfet
Related to #103973#110532#108404#94891
**Context:**
As commented in 6ae0554d11/cmake/Dependencies.cmake (L1198)
Kernel asserts are enabled by default for CUDA and disabled for ROCm.
However it is somewhat broken, and Kernel assert was still enabled for ROCm.
Disabling kernel assert is also needed for users who do not have PCIe atomics support. These community users have verified that disabling the kernel assert in PyTorch/ROCm platform fixed their pytorch workflow, like torch.sum script, stable-diffusion. (see the related issues)
**Changes:**
This pull request serves the following purposes:
* Refactor and clean up the logic, make it simpler for ROCm to enable and disable Kernel Asserts
* Fix the bug that Kernel Asserts for ROCm was not disabled by default.
Specifically,
- Renamed `TORCH_DISABLE_GPU_ASSERTS` to `C10_USE_ROCM_KERNEL_ASSERT` for the following reasons:
(1) This variable only applies to ROCm.
(2) The new name is more align with #define CUDA_KERNEL_ASSERT function.
(3) With USE_ in front of the name, we can easily control it with environment variable to turn on and off this feature during build (e.g. `USE_ROCM_KERNEL_ASSERT=1 python setup.py develop` will enable kernel assert for ROCm build).
- Get rid of the `ROCM_FORCE_ENABLE_GPU_ASSERTS' to simplify the logic and make it easier to understand and maintain
- Added `#cmakedefine` to carry over the CMake variable to C++
**Tests:**
(1) build with default mode and verify that USE_ROCM_KERNEL_ASSERT is OFF(0), and kernel assert is disabled:
```
python setup.py develop
```
Verify CMakeCache.txt has correct value.
```
/xxxx/pytorch/build$ grep USE_ROCM_KERNEL_ASSERT CMakeCache.txt
USE_ROCM_KERNEL_ASSERT:BOOL=0
```
Tested the following code in ROCm build and CUDA build, and expected the return code differently.
```
subprocess.call([sys.executable, '-c', "import torch;torch._assert_async(torch.tensor(0,device='cuda'));torch.cuda.synchronize()"])
```
This piece of code is adapted from below unit test to get around the limitation that this unit test now was skipped for ROCm. (We will check to enable this unit test in the future)
```
python test/test_cuda_expandable_segments.py -k test_fixed_cuda_assert_async
```
Ran the following script, expecting r ==0 since the CUDA_KERNEL_ASSERT is defined as nothing:
```
>> import sys
>>> import subprocess
>>> r=subprocess.call([sys.executable, '-c', "import torch;torch._assert_async(torch.tensor(0,device='cuda'));torch.cuda.synchronize()"])
>>> r
0
```
(2) Enable the kernel assert by building with USE_ROCM_KERNEL_ASSERT=1, or USE_ROCM_KERNEL_ASSERT=ON
```
USE_ROCM_KERNEL_ASSERT=1 python setup.py develop
```
Verify `USE_ROCM_KERNEL_ASSERT` is `1`
```
/xxxx/pytorch/build$ grep USE_ROCM_KERNEL_ASSERT CMakeCache.txt
USE_ROCM_KERNEL_ASSERT:BOOL=1
```
Run the assert test, and expected return code not equal to 0.
```
>> import sys
>>> import subprocess
>>> r=subprocess.call([sys.executable, '-c', "import torch;torch._assert_async(torch.tensor(0,device='cuda'));torch.cuda.synchronize()"])
>>>/xxxx/pytorch/aten/src/ATen/native/hip/TensorCompare.hip:108: _assert_async_cuda_kernel: Device-side assertion `input[0] != 0' failed.
:0:rocdevice.cpp :2690: 2435301199202 us: [pid:206019 tid:0x7f6cf0a77700] Callback: Queue 0x7f64e8400000 aborting with error : HSA_STATUS_ERROR_EXCEPTION: An HSAIL operation resulted in a hardware exception. code: 0x1016
>>> r
-6
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/114660
Approved by: https://github.com/jeffdaily, https://github.com/malfet, https://github.com/jithunnair-amd
Fixes#113940. This vendors the relevant parts of [`packaging==23.2.0`]() to have access to `Version` and `InvalidVersion` without taking a runtime dependency on `setuptools` or `packaging`.
I didn't find any vendoring policy so I put it under `torch._vendor.packaging`. While I have only vendored the files we need, I have not touched or trimmed the files otherwise.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/114108
Approved by: https://github.com/malfet, https://github.com/albanD
Building with `USE_CUSTOM_DEBINFO=torch/csrc/Module.cpp python setup.py develop` for example will provide debug info only for this file.
This allows to enable debug symbols very fast from a non-debug build by doing a clean then develop (as long as you have ccache) and avoid very large binaries that take a very long time to load in gdb.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111748
Approved by: https://github.com/drisspg, https://github.com/ezyang, https://github.com/malfet
Removes the existing integration code & build of nvfuser in TorchScript.
Note that I intentionally left the part where we wipe out `third_party/nvfuser` repo. I'll do that in a separate PR.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111093
Approved by: https://github.com/albanD
Summary:
This PR adds a limited C shim layer for libtorch. The ultimate goal is to ban any direct reference to aten/c10 data structures or functions, to avoid ABI breakage by providing stable C interfaces.
To make the review and landing easier, we broke the changes into several steps. In this PR (a combination of https://github.com/pytorch/pytorch/pull/109022 and https://github.com/pytorch/pytorch/pull/109351), we add C interfaces for certain libtorch functions and modify the wrapper codegen to generate calls to those interfaces. There are a few other items to be addressed in future PRs:
* The AOTInductor runtime interface still takes lists of aten tensors as input and output
* The interaction with ProxyExecutor (general fallback support) needs to move away from aten tensor
* Remove all references to aten/c10 headers in the AOTInductor-generated code
Differential Revision: D49302669
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109391
Approved by: https://github.com/chenyang78
# Motivate
Without this PR, if we would like to include the header file like ```#include <ATen/native/ForeachUtils.h>``` in our C++ extension, it will raise a Error ```/home/xxx/torch/include/ATen/native/ForeachUtils.h:7:10: fatal error: 'ATen/native/utils/ParamsHash.h' file not found```. We should fix it.
# Solution
Add the ATen/native/utils header file in the build.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109013
Approved by: https://github.com/ezyang
Summary: Move AOTInductor runtime header files into its own subdirectory, to separate them from to-be-added libtorch C interface.
Reviewed By: frank-wei
Differential Revision: D48905038
Pull Request resolved: https://github.com/pytorch/pytorch/pull/108564
Approved by: https://github.com/frank-wei
The dependency was added twice before in CUDA and ROCm binaries, one as an installation dependency from builder and the later as an extra dependency for dynamo, for example:
```
Requires-Python: >=3.8.0
Description-Content-Type: text/markdown
License-File: LICENSE
License-File: NOTICE
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: pytorch-triton (==2.1.0+e6216047b8)
Provides-Extra: dynamo
Requires-Dist: pytorch-triton (==2.1.0+e6216047b8) ; extra == 'dynamo'
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'
```
In the previous release, we needed to remove this part from `setup.py` to build release binaries https://github.com/pytorch/pytorch/pull/96010. With this, that step isn't needed anymore because the dependency will come from builder.
### Testing
Using the draft https://github.com/pytorch/pytorch/pull/108374 for testing and manually inspect the wheels artifact at https://github.com/pytorch/pytorch/actions/runs/6045878399 (don't want to go through all `ciflow/binaries` again)
* torch-2.1.0.dev20230901+cu121-cp39-cp39-linux_x86_64
```
Requires-Python: >=3.8.0
Description-Content-Type: text/markdown
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: pytorch-triton (==2.1.0+e6216047b8) <-- This will be 2.1.0 on the release branch after https://github.com/pytorch/builder/pull/1515
Provides-Extra: dynamo
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'
```
* torch-2.1.0.dev20230901+cu121.with.pypi.cudnn-cp39-cp39-linux_x86_64
```
Requires-Python: >=3.8.0
Description-Content-Type: text/markdown
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: pytorch-triton (==2.1.0+e6216047b8)
Requires-Dist: nvidia-cuda-nvrtc-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cuda-runtime-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cuda-cupti-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cudnn-cu12 (==8.9.2.26) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cublas-cu12 (==12.1.3.1) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cufft-cu12 (==11.0.2.54) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-curand-cu12 (==10.3.2.106) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cusolver-cu12 (==11.4.5.107) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-cusparse-cu12 (==12.1.0.106) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-nccl-cu12 (==2.18.1) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: nvidia-nvtx-cu12 (==12.1.105) ; platform_system == "Linux" and platform_machine == "x86_64"
Requires-Dist: triton (==2.1.0) ; platform_system == "Linux" and platform_machine == "x86_64" <--This is 2.1.0 because it already has https://github.com/pytorch/pytorch/pull/108423, but the package doesn't exist yet atm
Provides-Extra: dynamo
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'
```
* torch-2.1.0.dev20230901+rocm5.6-cp38-cp38-linux_x86_64
```
Requires-Python: >=3.8.0
Description-Content-Type: text/markdown
Requires-Dist: filelock
Requires-Dist: typing-extensions
Requires-Dist: sympy
Requires-Dist: networkx
Requires-Dist: jinja2
Requires-Dist: fsspec
Requires-Dist: pytorch-triton-rocm (==2.1.0+34f8189eae) <-- This will be 2.1.0 on the release branch after https://github.com/pytorch/builder/pull/1515
Provides-Extra: dynamo
Requires-Dist: jinja2 ; extra == 'dynamo'
Provides-Extra: opt-einsum
Requires-Dist: opt-einsum (>=3.3) ; extra == 'opt-einsum'
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/108424
Approved by: https://github.com/atalman
PR #90689 replaces NVTX with NVTX3. However, the torch::nvtoolsext is created only when the third party NVTX is used.
This is clear a logical error. We now move the creation code out of the branch to cover all cases. This should fix the issues reported in the comments of #90689.
It would be better to move configurations of the failed FRL jobs to CI tests so that we can find such issues early before merging.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/97582
Approved by: https://github.com/peterbell10
Summary:
This stack of PR's integrates cuSPARSELt into PyTorch.
This PR adds support for cuSPARSELt into the build process.
It adds in a new flag, USE_CUSPARSELT that defaults to false.
When USE_CUSPASRELT=1 is specified, the user can also specify
CUSPASRELT_ROOT, which defines the path to the library.
Compiling pytorch with cusparselt support can be done as follows:
``
USE_CUSPARSELT=1
CUSPARSELT_ROOT=/path/to/cusparselt
python setup.py develop
```
Test Plan:
Reviewers:
Subscribers:
Tasks:
Tags:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/103700
Approved by: https://github.com/albanD
This change adds the TensorPipe header files to `torch_package_data` if `USE_DISTRIBUTED` is set to `ON` in the CMake cache. The TensorPipe library and CMake config is already available in the Torch wheel, but the headers are not. This resolves issue where out-of-tree backends could not implement TensorPipe converters, because the definition of the `tensorpipe::Message` struct is defined in the TensorPipe headers.
Fixes#105224.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105521
Approved by: https://github.com/albanD
Summary:
Original PR at https://github.com/pytorch/pytorch/pull/104977. Landing from fbcode instead.
Add an aot_inductor backend (Export+AOTInductor) in the benchmarking harness. Note it is not a dynamo backend.
Moved files from torch/_inductor/aot_inductor_include to torch/csrc/inductor as a more standard way for exposing headers
Created a caching function in benchmarks/dynamo/common.py for compiling, loading and caching the .so file, as a proxy for a pure C++ deployment, but easier for benchmarking.
Differential Revision: D47452591
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105221
Approved by: https://github.com/jansel
This PR combines the C++ code for the AOTInductor's model and interface with Bin Bao's changes to AOTInductor codegen.
It adds a number of AOTInductor C interfaces that can be used by an inference runtime. Under the hood of the interfaces, the model code generated by the AOTInductor's codegen is wrapped into a class, AOTInductorModel, which manages tensors and run the model inference.
On top of AOTInductorModel, we provide one more abstract layer, AOTInductorModelContainer, which allows the user to have multiple inference runs concurrently for the same model.
This PR also adjusts the compilation options for AOT codegen, particularly some fbcode-related changes such as libs to be linked and header-file search paths.
Note that this is the very first version of the AOTInductor model and interface, so many features (e.g. dynamic shape) are incomplete. We will support those missing features in in future PRs.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/104202
Approved by: https://github.com/desertfire
Now, when you do an inplace mutation and the view is naughty, you get this message:
```
RuntimeError: A view was created in no_grad mode and is being modified inplace with grad mode enabled. Given that this use case is ambiguous and error-prone, it is forbidden. You can clarify your code by moving both the view and the inplace either both inside the no_grad block (if you don't want the inplace to be tracked) or both outside (if you want the inplace to be tracked). To find out where this view was allocated, run your entire forward region under anomaly mode (torch.autograd.detect_anomaly(check_nan=False)).
```
When you run under anomaly mode, you get:
```
RuntimeError: A view was created in no_grad mode and is being modified inplace with grad mode enabled. Given that this use case is ambiguous and error-prone, it is forbidden. You can clarify your code by moving both the view and the inplace either both inside the no_grad block (if you don't want the inplace to be tracked) or both outside (if you want the inplace to be tracked). This view was allocated at:
File "/data/users/ezyang/c/pytorch/test/test_autograd.py", line 4299, in arglebargle
File "/data/users/ezyang/c/pytorch/test/test_autograd.py", line 4306, in test_anomaly_gives_view_stack
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/case.py", line 549, in _callTestMethod
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/case.py", line 591, in run
File "/data/users/ezyang/c/pytorch/torch/testing/_internal/common_utils.py", line 2266, in _run_with_retry
File "/data/users/ezyang/c/pytorch/torch/testing/_internal/common_utils.py", line 2337, in run
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/case.py", line 650, in __call__
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/suite.py", line 122, in run
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/suite.py", line 84, in __call__
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/suite.py", line 122, in run
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/suite.py", line 84, in __call__
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/runner.py", line 184, in run
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/main.py", line 271, in runTests
File "/home/ezyang/local/c/pytorch-env/lib/python3.10/unittest/main.py", line 101, in __init__
File "/data/users/ezyang/c/pytorch/torch/testing/_internal/common_utils.py", line 894, in run_tests
File "/data/users/ezyang/c/pytorch/test/test_autograd.py", line 11209, in <module>
```
Signed-off-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/103185
Approved by: https://github.com/zdevito
Updating the pin to the same hash as https://github.com/pytorch/pytorch/pull/100922
On the XLA side, build have switch from CMake to bazel, which requires number of changes on PyTorch side:
- Copy installed headers back to the `torch/` folder before starting the build
- Install `torch/csrc/lazy/python/python_utils.h`
- Define `LD_LIBRARY_PATH`
TODO:
- Enable bazel caching
- Pass CXX11_ABI flag to `//test/cpp:all` to reuse build artifacts from `//:_XLAC.so`
<!--
copilot:poem
-->
### <samp>🤖 Generated by Copilot at cd4768b</samp>
> _To fix the XLA tests that were failing_
> _We updated the submodule and scaling_
> _We added `python_util.h`_
> _And copied `torch` as well_
> _And set `LD_LIBRARY_PATH` for linking_
Pull Request resolved: https://github.com/pytorch/pytorch/pull/102446
Approved by: https://github.com/huydhn
Failing mechanism on #95424 :
In dynamo mode, when passing numpy.int_ to 'shape' like param (Sequence[Union[int, symint]]) is wrapped as list with FakeTensor. However, in python_arg_parser, parser expect int in symint_list but got FakeTensor.
Following #85759, this PR allow tensor element in symint_list when in dynamo mode
This PR also fix below test with similar failing mechanism
pytest ./generated/test_huggingface_diffusers.py -k test_016
pytest ./generated/test_ustcml_RecStudio.py -k test_036
Fixes #ISSUE_NUMBER
Pull Request resolved: https://github.com/pytorch/pytorch/pull/97508
Approved by: https://github.com/yanboliang
implementation of DataPtr context for copy-on-write tensors
Summary:
Copy-on-write storage
=====================
This library adds support for copy-on-write storage, i.e. lazy copies,
to tensors. The design maintains the PyTorch invariant that tensors
alias if and only if they share a storage. Thus, tensors that are lazy
copies of one another will have distinct storages that share a data
allocation.
Thread-safety
-------------
The correctness of this design hinges on the pre-existing PyTorch user
requirement (and general default programming assumption) that users
are responsible for guaranteeing that writes do not take places
concurrently with reads and other writes.
Lazily copied tensors add a complication to this programming model
because users are not required to know if lazy copies exist and are
not required to serialize writes across lazy copies. For example: two
tensors with distinct storages that share a copy-on-write data context
may be given to different threads that may do whatever they wish to
them, and the runtime is required to guarantee its safety.
It turns out that this is not that difficult to protect because, due
to the copy-on-write requirement, we just need to materialize a tensor
upon writing. This could be done entirely without synchronization if
we materialized each copy, however, we have a common-sense
optimization to elide the copy for the last remaining reference. This
requires waiting for any pending copies.
### Thread-safety detailed design
There are two operations that affect the copy-on-write details of a
tensor:
1) lazy-clone (e.g. an explicit call or a hidden implementation detail
added through an operator like reshape)
2) materialization (i.e. any write to the tensor)
The key insight that we exploit is that lazy-clone is logically a read
operation and materialization is logically a write operation. This
means that, for a given set of tensors that share a storage, if
materialization is taking place, no other read operation, including
lazy-clone, can be concurrent with it.
However, this insight only applies within a set of tensors that share
a storage. We also have to be concerned with tensors with different
storages that share a copy-on-write context. In this world,
materialization can race with lazy-clone or even other
materializations. _However_, in order for this to be the case, there
must be _at least_ two references to the context. This means that the
context _can not_ vanish out from under you if you are performing a
lazy-clone, and hence, it only requires an atomic refcount bump.
The most complicated case is that all lazy-copies are concurrently
materializing. In this case, because a write is occurring, there are
no in-flight lazy-copies taking place. We must simply ensure that all
lazy-copies are able to materialize (read the data) concurrently. If
we didn't have the aforementioned optimization where the last copy
steals the data, we could get away with no locking whatsoever: each
makes a copy and decrements the refcount. However, because of the
optimization, we require the loser of the materializing race wait for
the pending copies to finish, and then steal the data without copying
it.
We implement this by taking a shared lock when copying the data and
taking an exclusive lock when stealing the data. The exclusive lock
acquisition ensures that all pending shared locks are finished before
we steal the data.
Test Plan: 100% code coverage.
---
Stack created with [Sapling](https://sapling-scm.com). Best reviewed with [ReviewStack](https://reviewstack.dev/pytorch/pytorch/pull/100818).
* #100821
* #100820
* #100819
* __->__ #100818
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100818
Approved by: https://github.com/ezyang
Follow up for https://github.com/pytorch/pytorch/pull/96532. Including this in setup.py so the package will be available for CI.
Fsspec package size:
```
du -h /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg
264K /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg/fsspec/__pycache__
58K /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg/fsspec/implementations/__pycache__
377K /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg/fsspec/implementations
1017K /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg/fsspec
96K /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg/EGG-INFO
1.2M /fsx/users/irisz/conda/envs/pytorch/lib/python3.9/site-packages/fsspec-2023.3.0-py3.9.egg
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/99768
Approved by: https://github.com/kit1980
