[https://nvbugs/5398180][feat] Improve Llama4 performance for small max_seqlen cases #6306
Conversation
Signed-off-by: Yilin Fan <[email protected]>
📝 WalkthroughWalkthroughA conditional check was added to the Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Suggested reviewers
Poem
Note ⚡️ Unit Test Generation is now available in beta!Learn more here, or try it out under "Finishing Touches" below. 📜 Recent review detailsConfiguration used: .coderabbit.yaml 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
⏰ 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)
✨ Finishing Touches
🧪 Generate unit tests
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. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
Documentation and Community
|
![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/models/modeling_llama.py (1)
72-76: LGTM! Logic correctly disables chunked attention for small sequences.The conditional check properly identifies when chunked attention provides no benefit (when
max_num_tokens < attention_chunk_size) and disables it to allow faster XQA kernel selection. This aligns well with the PR objective to improve performance for small sequence lengths.Consider adding a more descriptive comment explaining the performance rationale:
else: - # Disable chunked attention when max_num_tokens is smaller than attention_chunk_size - # TODO: Remove this after all attention kernels in TRTLLM backend support chunked attention + # Disable chunked attention when max_num_tokens < chunk_size to enable faster XQA kernels. + # XQA kernels don't support chunked attention but are faster for small sequences. + # TODO: Remove this after all attention kernels in TRTLLM backend support chunked attention if attention_chunk_size and model_config.max_num_tokens < attention_chunk_size: attention_chunk_size = None
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/models/modeling_llama.py(1 hunks)
🧰 Additional context used
🧠 Learnings (1)
tensorrt_llm/_torch/models/modeling_llama.py (1)
Learnt from: yechank-nvidia
PR: #6254
File: tensorrt_llm/_torch/pyexecutor/model_engine.py:1201-1204
Timestamp: 2025-07-22T09:22:14.726Z
Learning: In TensorRT-LLM's multimodal processing pipeline, shared tensor recovery using from_shared_tensor() is only needed during the context phase. Generation requests reuse the already-recovered tensor data and only need to call strip_for_generation() to remove unnecessary multimodal data while preserving the recovered tensors. This avoids redundant tensor recovery operations during generation.
⏰ 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 #13824 [ run ] triggered by Bot |
|
PR_Github #13824 [ run ] completed with state |
|
/bot run |
|
PR_Github #13830 [ run ] triggered by Bot |
|
PR_Github #13830 [ run ] completed with state |
|
/bot run |
|
PR_Github #14017 [ run ] triggered by Bot |
|
PR_Github #14017 [ run ] completed with state |
|
/bot run |
|
PR_Github #14038 [ run ] triggered by Bot |
|
PR_Github #14038 [ run ] completed with state |
Signed-off-by: Yilin Fan <[email protected]>
|
/bot run |
|
PR_Github #14165 [ run ] triggered by Bot |
|
PR_Github #14165 [ run ] completed with state |
nv-yilinf
changed the title
Signed-off-by: Yilin Fan <[email protected]>
nv-yilinf
changed the title
|
/bot run |
|
PR_Github #14625 [ run ] triggered by Bot |
|
PR_Github #14625 [ run ] completed with state |
|
/bot run |
|
PR_Github #14639 [ run ] triggered by Bot |
|
/bot run |
|
PR_Github #14642 [ run ] triggered by Bot |
|
PR_Github #14639 [ run ] completed with state |
|
/bot run |
1 similar comment
|
/bot run |
|
PR_Github #14644 [ run ] triggered by Bot |
|
PR_Github #14642 [ run ] completed with state |
|
PR_Github #14645 [ run ] triggered by Bot |
|
PR_Github #14644 [ run ] completed with state |
nv-yilinf
force-pushed
the
optimize-chunked-attention-for-small-max-seqlen
branch
from
a9d76c8 to
d3ef4f7
Compare
Signed-off-by: Yilin Fan <[email protected]>
nv-yilinf
force-pushed
the
optimize-chunked-attention-for-small-max-seqlen
branch
from
d3ef4f7 to
d41e955
Compare
|
/bot run |
|
PR_Github #14652 [ run ] triggered by Bot |
|
PR_Github #14645 [ run ] completed with state |
|
PR_Github #14652 [ run ] completed with state |
nv-yilinf
deleted the
optimize-chunked-attention-for-small-max-seqlen
branch
…6306) Signed-off-by: Yilin Fan <[email protected]>
Summary by CodeRabbit
Description
XQA does not support chunked attention at the moment, while MMHA supports chunked attention but is slower. In this PR we trick trtllm to select XQA kernels when we are sure that chunked attention will not be needed (i.e., max_seqlen < chunked_size).
This PR has to land after #6282
Test Coverage
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.