[AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c_shim for XPU. #136742
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced Sep 26, 2024
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/136742
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 78b994e with merge base 2ede4c9 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
EikanWang
approved these changes
Oct 30, 2024
torchgen/gen_aoti_c_shim.py
Outdated
| func_group_mapping: dict[OperatorName, NativeFunctionsGroup], | ||
| dispatch_key: DispatchKey, | ||
| backend_indices: dict[DispatchKey, BackendIndex], | ||
| aoti_extend: bool, |
Collaborator
There was a problem hiding this comment.
Refine the variable name a little bit.
torchgen/gen.py
Outdated
| parser.add_argument( | ||
| "--aoti-extend", | ||
| action="store_true", | ||
| help="Update AOTInductor C shim after adding an entry to inductor_fallback_ops in torchgen/aoti/fallback_ops.py. " |
Collaborator
There was a problem hiding this comment.
Update the help message.
torchgen/gen_aoti_c_shim.py
Outdated
| else: | ||
| # for the extend out-of-tree kernels, we don't need to | ||
| # duplicatly create C shim wrappers for other dispatch keys | ||
| if aoti_extend: |
Collaborator
There was a problem hiding this comment.
Suggested change
| if aoti_extend: | |
| if aoti_extend and diapatch_key == "XPU": |
Collaborator
Author
There was a problem hiding this comment.
The flag aoti_extend, which now is refined as extend_aoti_c_shim, is by design used for extending the c shim of an existing in-tree/partially-in-tree backend, eg, XPU, those totally out-of-tree backend can generate their c shim normally without this flag. So here we don't need to specify if aoti_extend and diapatch_key == "XPU"
Contributor
|
@desertfire should handle this review |
desertfire
approved these changes
Nov 7, 2024
| ) | ||
|
|
||
| set(GENERATED_CXX_TORCH_XPU | ||
| "${TORCH_SRC_DIR}/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp" |
Contributor
There was a problem hiding this comment.
NVM: after reading a few more lines, I realized you actually have two parts of c_shim_xpu,
Collaborator
Author
There was a problem hiding this comment.
@desertfire yes, as I descripted in the PR description above, there are
torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.h
torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp
torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.h
torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.cpp
The extend part is generated from out-of-tree torch-xpu-ops, which call python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim --aoti-insall-dir=torch/include/torch/csrc/inductor/aoti_torch/generated/extend/, and the generated extend/c_shim_xpu.cpp is built into libtorch_xpu_ops.a. So extend/c_shim_xpu.cpp is added in cmake file of torch-xpu-ops, not in pytorch cmake file.
Collaborator
Author
|
@pytorchbot merge |
Collaborator
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
atalman
pushed a commit
to atalman/pytorch
that referenced
this pull request
Nov 11, 2024
… make it easy to produce c shims for out-of-tree OP kernels as well. Add c_shim for XPU. (pytorch#136742) [AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c shim for XPU. ### Motivation Since the current c shim codegen will only produce C wrappers for Op's registered in `aten/src/ATen/native/native_functions.yaml`, for the same backend, when a portion of out-of-tree OP's are not registered in that file, but are registered externally. For example, `third_party/torch-xpu-ops/yaml/native_functions.yaml` , in this case, the existing codegen can't fulfill the need to do extensions for the c shims from the out-of-tree OPs for the in-tree that has already been produced. ### Design To extend the c shim with more OP for a backend from out-of-tree. The PR provided a bool option `--aoti-extend` to indicate the codegen is to extend c shim from out-of-tree. The generated c shim is stored in the `extend` subdirectory , for example: ``` torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.cpp ``` example usage: `python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim ` `--xpu`: generate c shim for XPU `--aoti-extend `: this is an out-of-tree OPs(defined in `third_party/torch-xpu-ops/yaml/native_functions.yaml`) extend for in-tree ops(defined in `aten/src/ATen/native/native_functions.yaml`) `--update-aoti-c-shim`: always generate c_shim_xpu.h for the extend c_shim. Pull Request resolved: pytorch#136742 Approved by: https://github.com/EikanWang, https://github.com/desertfire ghstack dependencies: pytorch#139025
Ryo-not-rio
pushed a commit
to Ryo-not-rio/pytorch
that referenced
this pull request
Dec 2, 2024
… make it easy to produce c shims for out-of-tree OP kernels as well. Add c_shim for XPU. (pytorch#136742) [AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c shim for XPU. ### Motivation Since the current c shim codegen will only produce C wrappers for Op's registered in `aten/src/ATen/native/native_functions.yaml`, for the same backend, when a portion of out-of-tree OP's are not registered in that file, but are registered externally. For example, `third_party/torch-xpu-ops/yaml/native_functions.yaml` , in this case, the existing codegen can't fulfill the need to do extensions for the c shims from the out-of-tree OPs for the in-tree that has already been produced. ### Design To extend the c shim with more OP for a backend from out-of-tree. The PR provided a bool option `--aoti-extend` to indicate the codegen is to extend c shim from out-of-tree. The generated c shim is stored in the `extend` subdirectory , for example: ``` torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.cpp ``` example usage: `python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim ` `--xpu`: generate c shim for XPU `--aoti-extend `: this is an out-of-tree OPs(defined in `third_party/torch-xpu-ops/yaml/native_functions.yaml`) extend for in-tree ops(defined in `aten/src/ATen/native/native_functions.yaml`) `--update-aoti-c-shim`: always generate c_shim_xpu.h for the extend c_shim. Pull Request resolved: pytorch#136742 Approved by: https://github.com/EikanWang, https://github.com/desertfire ghstack dependencies: pytorch#139025
pobin6
pushed a commit
to pobin6/pytorch
that referenced
this pull request
Dec 5, 2024
… make it easy to produce c shims for out-of-tree OP kernels as well. Add c_shim for XPU. (pytorch#136742) [AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c shim for XPU. ### Motivation Since the current c shim codegen will only produce C wrappers for Op's registered in `aten/src/ATen/native/native_functions.yaml`, for the same backend, when a portion of out-of-tree OP's are not registered in that file, but are registered externally. For example, `third_party/torch-xpu-ops/yaml/native_functions.yaml` , in this case, the existing codegen can't fulfill the need to do extensions for the c shims from the out-of-tree OPs for the in-tree that has already been produced. ### Design To extend the c shim with more OP for a backend from out-of-tree. The PR provided a bool option `--aoti-extend` to indicate the codegen is to extend c shim from out-of-tree. The generated c shim is stored in the `extend` subdirectory , for example: ``` torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.cpp ``` example usage: `python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim ` `--xpu`: generate c shim for XPU `--aoti-extend `: this is an out-of-tree OPs(defined in `third_party/torch-xpu-ops/yaml/native_functions.yaml`) extend for in-tree ops(defined in `aten/src/ATen/native/native_functions.yaml`) `--update-aoti-c-shim`: always generate c_shim_xpu.h for the extend c_shim. Pull Request resolved: pytorch#136742 Approved by: https://github.com/EikanWang, https://github.com/desertfire ghstack dependencies: pytorch#139025
fmo-mt
pushed a commit
to fmo-mt/pytorch
that referenced
this pull request
Dec 11, 2024
… make it easy to produce c shims for out-of-tree OP kernels as well. Add c_shim for XPU. (pytorch#136742) [AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c shim for XPU. ### Motivation Since the current c shim codegen will only produce C wrappers for Op's registered in `aten/src/ATen/native/native_functions.yaml`, for the same backend, when a portion of out-of-tree OP's are not registered in that file, but are registered externally. For example, `third_party/torch-xpu-ops/yaml/native_functions.yaml` , in this case, the existing codegen can't fulfill the need to do extensions for the c shims from the out-of-tree OPs for the in-tree that has already been produced. ### Design To extend the c shim with more OP for a backend from out-of-tree. The PR provided a bool option `--aoti-extend` to indicate the codegen is to extend c shim from out-of-tree. The generated c shim is stored in the `extend` subdirectory , for example: ``` torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/c_shim_xpu.cpp torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.h torch/include/torch/csrc/inductor/aoti_torch/generated/extend/c_shim_xpu.cpp ``` example usage: `python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim ` `--xpu`: generate c shim for XPU `--aoti-extend `: this is an out-of-tree OPs(defined in `third_party/torch-xpu-ops/yaml/native_functions.yaml`) extend for in-tree ops(defined in `aten/src/ATen/native/native_functions.yaml`) `--update-aoti-c-shim`: always generate c_shim_xpu.h for the extend c_shim. Pull Request resolved: pytorch#136742 Approved by: https://github.com/EikanWang, https://github.com/desertfire ghstack dependencies: pytorch#139025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):
[AOTI] Introduce an extensibility mechanism for the c shim codegen to make it easy to produce c shims for out-of-tree OP kernels as well. Add c shim for XPU.
Motivation
Since the current c shim codegen will only produce C wrappers for Op's registered in
aten/src/ATen/native/native_functions.yaml, for the same backend, when a portion of out-of-tree OP's are not registered in that file, but are registered externally. For example,third_party/torch-xpu-ops/yaml/native_functions.yaml, in this case, the existing codegen can't fulfill the need to do extensions for the c shims from the out-of-tree OPs for the in-tree that has already been produced.Design
To extend the c shim with more OP for a backend from out-of-tree.
The PR provided a bool option
--aoti-extendto indicate the codegen is to extend c shim from out-of-tree.The generated c shim is stored in the
extendsubdirectory , for example:example usage:
python -m torchgen.gen --source-path third_party/torch-xpu-ops/yaml/ --xpu --aoti-extend --update-aoti-c-shim--xpu: generate c shim for XPU--aoti-extend: this is an out-of-tree OPs(defined inthird_party/torch-xpu-ops/yaml/native_functions.yaml) extend for in-tree ops(defined inaten/src/ATen/native/native_functions.yaml)--update-aoti-c-shim: always generate c_shim_xpu.h for the extend c_shim.cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang @aakhundov