[TRTLLM-6854][feat] Enable guided decoding with CUDA graph padding and draft model chunked prefill #6774
Conversation
Signed-off-by: Enwei Zhu <[email protected]>
📝 WalkthroughWalkthroughUpdates guided decoder draft-advancement conditions and relax logits-size assertion to allow CUDA-graph dummy logits. Integration tests enable CUDA graph padding and chunked prefill for guided decoding scenarios. Changes
Sequence Diagram(s)sequenceDiagram
participant Driver as Decode Driver
participant Decoder as GuidedDecoder
participant Matcher as GrammarMatcher
Driver->>Decoder: draft_step(request)
Decoder->>Decoder: check state flags<br/>(is_context_init_state, is_last_context_chunk,<br/>is_generation_in_progress_state)
alt Allowed to advance
Decoder->>Matcher: advance()
Matcher-->>Decoder: advanced
else Not allowed
Decoder-->>Driver: no-advance
end
sequenceDiagram
participant Exec as execute()
participant CUDA as CUDA Graph Runner
Exec->>CUDA: run step (may include dummy requests)
CUDA-->>Exec: logits (may include dummy padding)
Exec->>Exec: assert offset <= logits.size(0)
Exec-->>CUDA: continue
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Possibly related PRs
Suggested labels
Suggested reviewers
✨ 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
|
|
/bot run --disable-fail-fast |
|
PR_Github #14736 [ run ] triggered by Bot |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)
350-350: CUDA graph padding enabled — consider specifying capture sizesSame note as above: consider batch_sizes=[1] or an explicit max_batch_size for clearer capture behavior in CI.
tensorrt_llm/_torch/pyexecutor/guided_decoder.py (1)
196-198: Relaxed logits-size assert to allow CUDA-graph padding — consider adding diagnosticsoffset <= logits.size(0) is the right direction to tolerate dummy rows from CUDA-graph padding. For easier debugging when padding is present, consider a lightweight debug when offset < logits.size(0) to confirm the surplus rows are indeed dummy and won’t be consumed.
Example:
- # Dummy logits may exist for CUDA graph dummy requests. - assert offset <= logits.size(0) + # Dummy logits may exist for CUDA graph dummy requests. + assert offset <= logits.size(0) + if offset < logits.size(0): + logger.debug(f"GuidedDecoder.execute: {logits.size(0) - offset} padded rows detected in logits.")
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
tensorrt_llm/_torch/pyexecutor/guided_decoder.py(2 hunks)tests/integration/defs/accuracy/test_llm_api_pytorch.py(3 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.py: Python code should conform to Python 3.8+.
Indent Python code with 4 spaces. Do not use tabs.
Always maintain the namespace when importing in Python, 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. Prefix k for variable names that start with a number (e.g., k_99th_percentile).
Python global variables should use upper snake_case and prefix G (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 interfaces that may be used outside a Python file, prefer docstrings over comments.
Comments in Python should be reserved 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.
Attributes and variables in Python can be documented inline; attribute docstrings will be rendered under the class docstring.
Avoid using reflection in Python when functionality can be easily achieved without it.
When using try-except blocks in Python, limit the except to the smallest set of errors possible.
When using try-except blocks to handle multiple possible variable types in Python, keep the body of the try as small as possible, using the else block to implement the logic.
Files:
tensorrt_llm/_torch/pyexecutor/guided_decoder.pytests/integration/defs/accuracy/test_llm_api_pytorch.py
**/*.{cpp,h,hpp,cc,cxx,cu,py}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the current year. This includes .cpp, .h, .cu, .py, and any other source files which are compiled or interpreted.
Files:
tensorrt_llm/_torch/pyexecutor/guided_decoder.pytests/integration/defs/accuracy/test_llm_api_pytorch.py
🧬 Code Graph Analysis (1)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (3)
tensorrt_llm/llmapi/llm_args.py (2)
CudaGraphConfig(106-163)NGramDecodingConfig(453-485)tests/unittest/_torch/test_beam_search.py (1)
llm(37-49)tensorrt_llm/llmapi/llm.py (1)
LLM(1079-1095)
⏰ 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 (3)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (3)
328-328: CUDA graph padding enabled — consider specifying capture sizesEnabling padding is good for throughput. To avoid relying on defaults, consider also setting batch_sizes or max_batch_size so capture shapes are deterministic for these tests.
Do you expect batch size >1 in these runs? If not, using batch_sizes=[1] (or an appropriate max_batch_size) will make captures explicit and predictable.
337-339: LGTM: Guided decoding wired with CUDA graph padding and chunked prefillPassing cuda_graph_config and enable_chunked_prefill=True aligns with the intended feature. disable_overlap_scheduler=True is set below, which is appropriate here.
356-358: LGTM: NGram guided decoding with CUDA graph padding + chunked prefillConfiguration is consistent; no issues spotted.
|
PR_Github #14736 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #14775 [ run ] triggered by Bot |
|
PR_Github #14775 [ run ] completed with state |
…dding and draft model chunked prefill (NVIDIA#6774)" This reverts commit 7c686ba.
syuoni
changed the title
3 participants
Summary by CodeRabbit
New Features
Bug Fixes
Tests
Description
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.