[None][fix] Reduce host overhead for unified nvfp4 gemm tuning path.#10503
[None][fix] Reduce host overhead for unified nvfp4 gemm tuning path.#10503hyukn merged 1 commit intoNVIDIA:mainfrom
Conversation
📝 WalkthroughWalkthroughRefactored NVFP4GemmUnifiedRunner to represent tactics as tuples of (backend_name, sub_tactic) instead of simple integers. Updated get_valid_tactics() return type and forward() dispatch logic to work with the new tuple-based representation. Added conditional support for CuteDSL backend. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes 🚥 Pre-merge checks | ❌ 3❌ Failed checks (2 warnings, 1 inconclusive)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
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: 2
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py (1)
844-860: Bug: Incorrect membership check for tuple-based tactics.The check
"cutlass" in valid_tacticswill always fail becausevalid_tacticsis now aList[Tuple[str, int]](e.g.,[("cuda_core", 0), ("cutlass", 1)]). The string"cutlass"will never match a tuple.This causes the fallback to always use
valid_tactics[0]instead of preferring the cutlass backend.🐛 Proposed fix
if tactic == -1: # Get valid tactics and use first available from tensorrt_llm._torch.autotuner import OptimizationProfile valid_tactics = self.get_valid_tactics(inputs, OptimizationProfile()) if valid_tactics: # Prefer cutlass as fallback if available, otherwise use first valid tactic - tactic = ["cutlass", -1] if "cutlass" in valid_tactics else [ - valid_tactics[0], -1 - ] + cutlass_tactics = [t for t in valid_tactics if t[0] == "cutlass"] + if cutlass_tactics: + tactic = (cutlass_tactics[0][0], -1) # Use cutlass with fallback sub_tactic + else: + tactic = (valid_tactics[0][0], -1) # Use first available backend with fallback else:Additionally, the fallback tactic should be a tuple
(backend, sub_tactic), not a list[backend, sub_tactic], to maintain type consistency with the rest of the code.
🤖 Fix all issues with AI agents
In @tensorrt_llm/_torch/custom_ops/torch_custom_ops.py:
- Around line 836-838: The parameter 'tactic' is declared as Union[Tuple, int]
but its default is the string "cutlass"; update the signature for consistency by
either adding str to the type hint (Union[Tuple, int, str]) if string backends
are intentional, or change the default to a valid fallback like -1 (i.e.,
tactic: Union[Tuple, int] = -1) so the declared type matches the default; adjust
any downstream usages or docstrings of the 'tactic' parameter accordingly.
- Around line 863-870: The computed variable act_sf_unswizzled is dead
code—remove the torch.ops.trtllm.block_scale_interleave_reverse call and the
act_sf_unswizzled assignment in the branch where backend == "cuda_core" since
CudaCoreNVFP4Runner.forward already unswizzles internally; keep the m and
act_fp4 references only if used elsewhere, and simply return
CudaCoreNVFP4Runner(self.to_userbuffers, self.output_dtype)(inputs, sub_tactic)
without computing act_sf_unswizzled.
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: The code developed for TensorRT-LLM should conform to Python 3.8+
Indent Python code with 4 spaces. Do not use tabs
Always maintain the namespace when importing Python modules, even if only one class or function from a module is used
Python filenames should use snake_case (e.g.,some_file.py)
Python classes should use PascalCase (e.g.,class SomeClass)
Python functions and methods should use snake_case (e.g.,def my_awesome_function():)
Python local variables should use snake_case, with prefixkfor variable names that start with a number (e.g.,k_99th_percentile)
Python global variables should use upper snake_case with prefixG(e.g.,G_MY_GLOBAL)
Python constants should use upper snake_case (e.g.,MY_CONSTANT)
Avoid shadowing variables declared in an outer scope in Python
Initialize all externally visible members of a Python class in the constructor
For Python interfaces that may be used outside a file, prefer docstrings over comments
Use comments in Python for code within a function, or interfaces that are local to a file
Use Google-style docstrings for Python classes and functions, which can be parsed by Sphinx
Python attributes and variables can be documented inline with the format"""<type>: Description"""
Avoid using reflection in Python when functionality can be easily achieved without reflection
When using try-except blocks in Python, limit the except clause to the smallest set of errors possible
When using try-except blocks in Python to handle multiple possible variable types (duck-typing), keep the body of the try as small as possible and use the else block for the main logic
Files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
**/*.{cpp,cc,cxx,h,hpp,hxx,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
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
Files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
🧠 Learnings (6)
📚 Learning: 2025-11-14T11:22:03.729Z
Learnt from: nzmora-nvidia
Repo: NVIDIA/TensorRT-LLM PR: 9163
File: tensorrt_llm/_torch/auto_deploy/custom_ops/quant.py:107-113
Timestamp: 2025-11-14T11:22:03.729Z
Learning: In TensorRT-LLM AutoDeploy custom ops, when adding hardware capability checks to select between kernel implementations (e.g., cuBLAS vs. CUDA kernel), use descriptive variable names that identify the specific GPU architectures or families being targeted (e.g., `is_blackwell_geforce_or_ada`) rather than generic names like `enable_cuda_core`. This makes it clear that the code is selecting an implementation path based on hardware capabilities, not enabling/disabling hardware features.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
📚 Learning: 2025-08-21T21:48:35.135Z
Learnt from: djns99
Repo: NVIDIA/TensorRT-LLM PR: 7104
File: cpp/tensorrt_llm/cutlass_extensions/include/cutlass_extensions/epilogue/fusion/sm90_visitor_scatter.hpp:399-417
Timestamp: 2025-08-21T21:48:35.135Z
Learning: CUTLASS extensions in TensorRT-LLM (located under cpp/tensorrt_llm/cutlass_extensions/) are designed to integrate with and extend functionality in the external CUTLASS repository. When analyzing these extensions, their consumers and functionality wiring may exist in the CUTLASS codebase rather than within TensorRT-LLM itself.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
📚 Learning: 2025-08-22T01:54:35.850Z
Learnt from: djns99
Repo: NVIDIA/TensorRT-LLM PR: 7104
File: cpp/tensorrt_llm/kernels/cutlass_kernels/include/moe_kernels.h:999-1000
Timestamp: 2025-08-22T01:54:35.850Z
Learning: The `internal_cutlass_kernels` directory in TensorRT-LLM is a mirror of an internal NVIDIA repository and maintains its own implementation and API that may diverge from the public `cutlass_kernels` version. API inconsistencies between these two directories are intentional and by design, not bugs to be fixed.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
📚 Learning: 2025-10-20T16:54:09.824Z
Learnt from: nvchenghaoz
Repo: NVIDIA/TensorRT-LLM PR: 8469
File: tensorrt_llm/_torch/auto_deploy/custom_ops/rms_norm.py:6-6
Timestamp: 2025-10-20T16:54:09.824Z
Learning: In tensorrt_llm/_torch/auto_deploy/custom_ops/rms_norm.py, the import `from ...modules.mamba.layernorm_gated import _layer_norm_fwd` is correct and should not be changed to modules.fla.layernorm_gated. The _layer_norm_fwd function exists in both modules/mamba/layernorm_gated.py and modules/fla/layernorm_gated.py, but the mamba version is the intended implementation for this use case.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
📚 Learning: 2025-09-23T14:58:05.372Z
Learnt from: nv-lschneider
Repo: NVIDIA/TensorRT-LLM PR: 7910
File: cpp/tensorrt_llm/kernels/nccl_device/config.cu:42-49
Timestamp: 2025-09-23T14:58:05.372Z
Learning: In TensorRT-LLM NCCL device kernels (cpp/tensorrt_llm/kernels/nccl_device/), the token partitioning intentionally uses ceil-like distribution (same token_per_rank for all ranks) to ensure all ranks launch the same number of blocks. This is required for optimal NCCL device API barrier performance, even though it may launch extra blocks for non-existent tokens on later ranks. Runtime bounds checking in the kernel (blockID validation) handles the overshoot cases.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
📚 Learning: 2025-12-12T10:07:31.564Z
Learnt from: lirundong
Repo: NVIDIA/TensorRT-LLM PR: 9725
File: tensorrt_llm/_torch/custom_ops/cuda_tile_custom_ops.py:110-178
Timestamp: 2025-12-12T10:07:31.564Z
Learning: In PyTorch custom operators registered with torch.library.custom_op, mutable operators that return None and specify mutates_args do not require a register_fake decorator. Mutation tracking is handled automatically without needing a FakeTensor kernel. This applies to Python custom op definitions in tensorrt_llm/_torch/custom_ops that use mutates_args and return None; verify you are not relying on register_fake in these cases.
Applied to files:
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
🧬 Code graph analysis (1)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py (1)
tensorrt_llm/_torch/custom_ops/cute_dsl_custom_ops.py (6)
CuteDSLNVFP4BlackwellLinear(334-739)get_valid_tactics(372-493)get_valid_tactics(858-893)get_valid_tactics(1148-1188)get_valid_tactics(1536-1571)get_valid_tactics(1852-1892)
🪛 Ruff (0.14.10)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py
734-734: Unused method argument: kwargs
(ARG002)
867-867: Local variable act_sf_unswizzled is assigned to but never used
Remove assignment to unused variable act_sf_unswizzled
(F841)
⏰ 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 (4)
tensorrt_llm/_torch/custom_ops/torch_custom_ops.py (4)
26-28: LGTM!The conditional import is correctly guarded behind
IS_CUTLASS_DSL_AVAILABLE, preventingImportErrorwhen CuteDSL is not compiled/linked.
755-784: LGTM!The tactic tuple construction correctly wraps each backend's sub-tactics with the backend name, enabling unified dispatch. The pattern is consistent across all backends (cuda_core, cutlass, cublaslt).
812-813: LGTM!CuteDSL tactic construction follows the same tuple pattern as other backends.
871-881: No action required. TheCuteDSLNVFP4BlackwellLinearconstructor hasto_userbuffers: bool = Falseas an optional parameter, so both call patterns are valid: instantiation with onlyoutput_dtype(line 805) uses the default value, and explicit provision ofto_userbuffers(line 879) is also supported. This is standard Python design and will not cause failures.Likely an incorrect or invalid review comment.
|
/bot run --disable-fail-fast |
|
PR_Github #30971 [ run ] triggered by Bot. Commit: |
|
/bot run --disable-fail-fast |
|
/bot run --disable-fail-fast |
hyukn
force-pushed
the
fix/fp4_gemm_tuning_host_overhead
branch
from
8842a90 to
63ebe75
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #30980 [ run ] triggered by Bot. Commit: |
|
PR_Github #30971 [ run ] completed with state |
|
PR_Github #30983 [ run ] triggered by Bot. Commit: |
|
PR_Github #30980 [ run ] completed with state |
|
PR_Github #30984 [ run ] triggered by Bot. Commit: |
|
PR_Github #30983 [ run ] completed with state |
hyukn
force-pushed
the
fix/fp4_gemm_tuning_host_overhead
branch
from
63ebe75 to
a549e35
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #31035 [ run ] triggered by Bot. Commit: |
|
PR_Github #31035 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #31157 [ run ] triggered by Bot. Commit: |
|
PR_Github #31157 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #31448 [ run ] triggered by Bot. Commit: |
|
PR_Github #31448 [ run ] completed with state
|
Signed-off-by: Yukun He <23156053+hyukn@users.noreply.github.com>
hyukn
force-pushed
the
fix/fp4_gemm_tuning_host_overhead
branch
from
a549e35 to
9550c96
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #31518 [ run ] triggered by Bot. Commit: |
|
PR_Github #31518 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #31644 [ run ] triggered by Bot. Commit: |
|
PR_Github #31644 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #31719 [ run ] triggered by Bot. Commit: |
|
PR_Github #31719 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #31766 [ run ] triggered by Bot. Commit: |
|
PR_Github #31766 [ run ] completed with state |
…VIDIA#10503) Signed-off-by: Yukun He <23156053+hyukn@users.noreply.github.com>
5 participants
Summary by CodeRabbit
Release Notes
New Features
Refactor
✏️ Tip: You can customize this high-level summary in your review settings.
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.