[TRTLLM-9771][feat] Make refit compatible with CUDA Graph#11267
[TRTLLM-9771][feat] Make refit compatible with CUDA Graph#11267shuyixiong wants to merge 4 commits intoNVIDIA:mainfrom
Conversation
Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
…n sm100 GPUs Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
|
/bot run --disable-fail-fast |
|
PR_Github #34968 [ run ] triggered by Bot. Commit: |
|
PR_Github #34968 [ run ] completed with state |
📝 WalkthroughWalkthroughThe changes introduce SM_100f-specific weight resmoothing and parameter management optimizations for MoE quantization pipelines. Meta tensor extraction from metadata dictionaries is added to weight loading workflows across modules. A utility function for parameter replacement is enhanced to support persistent metadata tracking and incremental data syncing. New multi-GPU weight update tests with quantization are added, and existing tests receive Hopper-specific skip decorators. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~22 minutes 🚥 Pre-merge checks | ✅ 1 | ❌ 2❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ 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
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (3)
tensorrt_llm/_torch/modules/linear.py (1)
1-1:⚠️ Potential issue | 🟠 MajorAdd NVIDIA copyright header with the latest year.
This source file is missing the required NVIDIA header.As per coding guidelines, All TensorRT-LLM source files (.cpp, .h, .cu, .py, and other source files) should contain an NVIDIA copyright header with the year of latest meaningful modification.
tensorrt_llm/_torch/utils.py (1)
1-1:⚠️ Potential issue | 🟠 MajorAdd NVIDIA copyright header with the latest year.
This source file is missing the required NVIDIA header.As per coding guidelines, All TensorRT-LLM source files (.cpp, .h, .cu, .py, and other source files) should contain an NVIDIA copyright header with the year of latest meaningful modification.
tensorrt_llm/_torch/modules/fused_moe/quantization.py (1)
1-5:⚠️ Potential issue | 🟠 MajorAdd the NVIDIA copyright header (latest year).
This file has no NVIDIA copyright header even though it was modified; please add the standard TensorRT-LLM header with the year 2026 at the top.As per coding guidelines, All TensorRT-LLM source files (.cpp, .h, .cu, .py, and other source files) should contain an NVIDIA copyright header with the year of latest meaningful modification.
🧹 Nitpick comments (4)
tests/unittest/_torch/ray_orchestrator/single_gpu/test_llm_update_weights.py (1)
12-12: Keep utils.util import namespaced.
Project style prefers module‑qualified imports; switch tofrom utils import utiland update the decorators accordingly.♻️ Suggested update
-from utils.util import skip_no_hopper +from utils import util -@skip_no_hopper +@util.skip_no_hopperAs per coding guidelines, Always maintain the namespace when importing Python modules, even if only one class or function from a module is used.
tensorrt_llm/_torch/utils.py (1)
419-428: Use Google‑style docstring format for replace_parameter_and_save_metadata.
Please convert the new description to an Args/Returns layout to keep Sphinx parsing consistent.As per coding guidelines, Use Google-style docstrings for Python classes and functions, which can be parsed by Sphinx.
tests/unittest/_torch/ray_orchestrator/multi_gpu/test_llm_update_weights_multi_gpu.py (1)
1-13: Keep cross‑test imports namespaced.
Prefer module‑qualified imports for the single‑GPU helpers andutils.utilto match project style.♻️ Suggested update
-from _torch.ray_orchestrator.single_gpu.test_llm_update_weights import ( - RefHFModelWithIPCHandles, - compare_logits, - run_generate, -) -from utils.util import skip_pre_blackwell +from _torch.ray_orchestrator.single_gpu import test_llm_update_weights as single_gpu_update_weights +from utils import util -@skip_pre_blackwell +@util.skip_pre_blackwell -hf_model = RefHFModelWithIPCHandles(fp8_model_dir, num_hidden_layers=num_hidden_layers) +hf_model = single_gpu_update_weights.RefHFModelWithIPCHandles( + fp8_model_dir, num_hidden_layers=num_hidden_layers +)Apply the same namespace pattern to
run_generate/compare_logitscalls in both tests.As per coding guidelines, Always maintain the namespace when importing Python modules, even if only one class or function from a module is used.
tensorrt_llm/_torch/modules/fused_moe/quantization.py (1)
1028-1064: Avoid reflection for constant attribute names.
Line 1028 onward usesgetattr/setattrwith constant strings; direct attribute access is clearer and aligns with project guidance.♻️ Suggested refactor
- local_shared_w3_w1_tensors = getattr( - module, 'local_shared_w3_w1_tensors') - local_shared_w3_w1_scale_tensors = getattr( - module, 'local_shared_w3_w1_scale_tensors') - local_shared_w2_tensors = getattr( - module, 'local_shared_w2_tensors') - local_shared_w2_scale_tensors = getattr( - module, 'local_shared_w2_scale_tensors') + local_shared_w3_w1_tensors = module.local_shared_w3_w1_tensors + local_shared_w3_w1_scale_tensors = module.local_shared_w3_w1_scale_tensors + local_shared_w2_tensors = module.local_shared_w2_tensors + local_shared_w2_scale_tensors = module.local_shared_w2_scale_tensors ... - setattr(module, 'local_shared_w3_w1_tensors', - resmoothed_shared_w3_w1_weight.cpu()) - setattr(module, 'local_shared_w3_w1_scale_tensors', - transformed_shared_w3_w1_scale.cpu()) + module.local_shared_w3_w1_tensors = resmoothed_shared_w3_w1_weight.cpu() + module.local_shared_w3_w1_scale_tensors = transformed_shared_w3_w1_scale.cpu() ... - setattr(module, 'local_shared_w2_tensors', - resmoothed_shared_w2_weight.cpu()) - setattr(module, 'local_shared_w2_scale_tensors', - transformed_shared_w2_scale.cpu()) + module.local_shared_w2_tensors = resmoothed_shared_w2_weight.cpu() + module.local_shared_w2_scale_tensors = transformed_shared_w2_scale.cpu()As per coding guidelines, Avoid using reflection in Python when functionality can be easily achieved without reflection.
2 participants
Summary by CodeRabbit
Tests
Refactor
Description
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
Update tava architecture diagram if there is a significant design change in PR.
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.
Details
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.