[None][feat] Add torch compile support for cuda core GEMM OP #8261
Conversation
Signed-off-by: Dylan Chen <[email protected]>
DylanChen-NV
force-pushed
the
fix_torch_compile_cudacoregemm
branch
from
0d27741 to
dbc7b14
Compare
📝 WalkthroughWalkthroughAdds a new Torch fake-op registration trtllm::cuda_scaled_mm in tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py. It defines a function mirroring cublas_scaled_mm with optional bias and out_dtype, plus userbuffers_id. The implementation infers output shape from inputs, allocates the output tensor with optional dtype, and returns it. Existing cublas op unchanged. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor U as Caller
participant T as PyTorch Dispatcher
participant O as trtllm::cuda_scaled_mm (fake-op)
U->>T: call trtllm::cuda_scaled_mm(mat_a, mat_b, scale_a, scale_b, [bias], [out_dtype], [userbuffers_id])
T->>O: dispatch to fake-op
Note over O: Infer output shape from mat_a x mat_b<br/>Select dtype (out_dtype or inferred)
O-->>T: return allocated output tensor
T-->>U: output tensor
sequenceDiagram
autonumber
actor U as Caller
participant T as PyTorch Dispatcher
participant C as trtllm::cublas_scaled_mm (existing)
participant N as trtllm::cuda_scaled_mm (new)
alt Use cuBLAS path
U->>T: trtllm::cublas_scaled_mm(...)
T->>C: dispatch
C-->>U: output tensor
else Use CUDA path
U->>T: trtllm::cuda_scaled_mm(...)
T->>N: dispatch
N-->>U: output tensor
end
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
📝 WalkthroughWalkthroughAdds a new fake Torch operator registration trtllm::cuda_scaled_mm in _register_fake, mirroring trtllm::cublas_scaled_mm. It defines shape/dtype inference: output shape copies mat_a with last dim set to mat_b.shape[-1], and returns a new tensor of dtype out_dtype. No other files changed. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py (1)
124-137: Approve fake operator registration fortrtllm::cuda_scaled_mm—shape/dtype inference is correct and C++ registration in cudaScaledMM.cpp matches the schema. Optional: extract the shared fake implementation to eliminate duplication.
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py(1 hunks)
🧰 Additional context used
📓 Path-based instructions (3)
**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Use only spaces, no tabs; indent with 4 spaces.
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Python code must target Python 3.8+.
Indent Python code with 4 spaces; do not use tabs.
Maintain module namespace when importing; prefer 'from package.subpackage import foo' then 'foo.SomeClass()' instead of importing the class directly.
Python filenames should be snake_case (e.g., some_file.py).
Python classes use PascalCase names.
Functions and methods use snake_case names.
Local variables use snake_case; prefix 'k' for variables that start with a number (e.g., k_99th_percentile).
Global variables use upper SNAKE_CASE prefixed with 'G' (e.g., G_MY_GLOBAL).
Constants use upper SNAKE_CASE (e.g., MY_CONSTANT).
Avoid shadowing variables from an outer scope.
Initialize all externally visible members of a class in the constructor.
Prefer docstrings for interfaces that may be used outside a file; comments for in-function or file-local interfaces.
Use Google-style docstrings for classes and functions (Sphinx-parsable).
Document attributes and variables inline so they render under the class/function docstring.
Avoid reflection when a simpler, explicit approach suffices (e.g., avoid dict(**locals()) patterns).
In try/except, catch the most specific exceptions possible.
For duck-typing try/except, keep the try body minimal and use else for the main logic.
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Prepend the NVIDIA Apache-2.0 copyright header with current year to the top of all source files (e.g., .cpp, .h, .cu, .py).
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
🧬 Code graph analysis (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py (1)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py (9)
_(249-302)_(378-386)_(464-474)_(643-670)_(703-713)_(787-797)_(887-903)_(984-992)_(1025-1036)
🪛 Ruff (0.13.3)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
128-128: Unused function argument: scale_a
(ARG001)
129-129: Unused function argument: scale_b
(ARG001)
130-130: Unused function argument: bias
(ARG001)
132-132: Unused function argument: userbuffers_id
(ARG001)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Pre-commit Check
|
/bot run |
|
PR_Github #21011 [ run ] triggered by Bot |
|
PR_Github #21011 [ run ] completed with state |
Signed-off-by: Dylan Chen <[email protected]>
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py (1)
109-122: Consider adding type annotations for consistency.For consistency with the newly added
cuda_scaled_mm, consider adding proper type annotations tocublas_scaled_mm:@torch.library.register_fake("trtllm::cublas_scaled_mm") def _( mat_a: torch.Tensor, mat_b: torch.Tensor, scale_a: torch.Tensor, scale_b: torch.Tensor, - bias, - out_dtype, - userbuffers_id=False, + bias: Optional[torch.Tensor] = None, + out_dtype: Optional[torch.dtype] = None, + userbuffers_id: bool = False, ):
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py(1 hunks)
🧰 Additional context used
📓 Path-based instructions (3)
**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Use only spaces, no tabs; indent with 4 spaces.
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Python code must target Python 3.8+.
Indent Python code with 4 spaces; do not use tabs.
Maintain module namespace when importing; prefer 'from package.subpackage import foo' then 'foo.SomeClass()' instead of importing the class directly.
Python filenames should be snake_case (e.g., some_file.py).
Python classes use PascalCase names.
Functions and methods use snake_case names.
Local variables use snake_case; prefix 'k' for variables that start with a number (e.g., k_99th_percentile).
Global variables use upper SNAKE_CASE prefixed with 'G' (e.g., G_MY_GLOBAL).
Constants use upper SNAKE_CASE (e.g., MY_CONSTANT).
Avoid shadowing variables from an outer scope.
Initialize all externally visible members of a class in the constructor.
Prefer docstrings for interfaces that may be used outside a file; comments for in-function or file-local interfaces.
Use Google-style docstrings for classes and functions (Sphinx-parsable).
Document attributes and variables inline so they render under the class/function docstring.
Avoid reflection when a simpler, explicit approach suffices (e.g., avoid dict(**locals()) patterns).
In try/except, catch the most specific exceptions possible.
For duck-typing try/except, keep the try body minimal and use else for the main logic.
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Prepend the NVIDIA Apache-2.0 copyright header with current year to the top of all source files (e.g., .cpp, .h, .cu, .py).
Files:
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
🧬 Code graph analysis (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py (1)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py (9)
_(249-302)_(378-386)_(464-474)_(643-670)_(703-713)_(787-797)_(887-903)_(984-992)_(1025-1036)
🪛 Ruff (0.13.3)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py
128-128: Unused function argument: scale_a
(ARG001)
129-129: Unused function argument: scale_b
(ARG001)
130-130: Unused function argument: bias
(ARG001)
132-132: Unused function argument: userbuffers_id
(ARG001)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Pre-commit Check
🔇 Additional comments (1)
tensorrt_llm/_torch/custom_ops/cpp_custom_ops.py (1)
124-137: LGTM! Proper type annotations added.The new fake op registration for
cuda_scaled_mmcorrectly implements the output shape inference with proper type annotations for all parameters. This addresses the past review feedback about completing type annotations forbiasandout_dtype.The static analysis warnings about unused parameters are false positives—fake op registrations only need to compute output shapes for torch.compile tracing, not perform actual computations.
|
/bot reuse-pipeline |
|
PR_Github #21137 [ reuse-pipeline ] triggered by Bot |
|
PR_Github #21137 [ reuse-pipeline ] completed with state |
3 participants
Summary by CodeRabbit
Description
Add torch compile support for cuda core GEMM OP.
Test Coverage
PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
GitHub Bot Help
/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...Provide a user friendly way for developers to interact with a Jenkins server.
Run
/bot [-h|--help]to print this help message.See details below for each supported subcommand.
run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]Launch build/test pipelines. All previously running jobs will be killed.
--reuse-test (optional)pipeline-id(OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.--disable-reuse-test(OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.--disable-fail-fast(OPTIONAL) : Disable fail fast on build/tests/infra failures.--skip-test(OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.--stage-list "A10-PyTorch-1, xxx"(OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.--gpu-type "A30, H100_PCIe"(OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.--test-backend "pytorch, cpp"(OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.--only-multi-gpu-test(OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.--disable-multi-gpu-test(OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.--add-multi-gpu-test(OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.--post-merge(OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx"(OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".--detailed-log(OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.--debug(OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in thestage-listparameter to access the appropriate container environment. Note: Does NOT update GitHub check status.For guidance on mapping tests to stage names, see
docs/source/reference/ci-overview.mdand the
scripts/test_to_stage_mapping.pyhelper.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip testing for latest commit on pull request.
--comment "Reason for skipping build/test"is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.reuse-pipeline
reuse-pipelineReuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.