[None][feat] Enable CUDA graph support for KvConnectorWorker API #8275
Conversation
In model_engine.py, during the warmup phase, there is an option for graph capturing if graph capturing is enabled. This commit adds a lot of logging needed to trace the calls executing in different parts of TRTLLM, this can be removed later. Ability to insert an event at the end of the capture and then hold that event on replay is proven. Signed-off-by: Kyle McGill <[email protected]>
This update records an event at the end of the forward pass regardless of whether the forward pass was done in graph_capture graph_replay, or eager_execution mode. This will allow KVMB to register cuda events with TRTLLM and then await on them in different streams Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
nv-kmcgill53
changed the title
📝 WalkthroughWalkthroughIntroduces a forward-pass CUDA event callback mechanism. Adds an inner ForwardPassCallback class and registration API to KvCacheConnectorWorker. Integrates callback registration and invocation in ModelEngine/PyTorchModelEngine after each forward. Wires registration in py_executor_creator and removes a CUDA graph guard to allow this path. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor Client
participant ExecCreator as PyExecutorCreator
participant Engine as PyTorchModelEngine
participant CUDA as CUDA Stream/Event
participant KV as KvCacheConnectorWorker.ForwardPassCallback
Client->>ExecCreator: create_engine(worker=KvCacheConnectorWorker?)
alt Worker provided
ExecCreator->>KV: worker.register_forward_pass_callback()
KV-->>ExecCreator: ForwardPassCallback
ExecCreator->>Engine: register_forward_pass_callback(callback)
end
loop Each forward
Client->>Engine: forward(inputs)
Engine->>CUDA: run forward (eager or CUDA graph)
Engine->>CUDA: event.record() at end of forward
alt Callback registered
Engine->>KV: callback()
end
Engine-->>Client: outputs
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 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: 1
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py(2 hunks)tensorrt_llm/_torch/pyexecutor/model_engine.py(4 hunks)tensorrt_llm/_torch/pyexecutor/py_executor_creator.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/pyexecutor/py_executor_creator.pytensorrt_llm/_torch/pyexecutor/kv_cache_connector.pytensorrt_llm/_torch/pyexecutor/model_engine.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/pyexecutor/py_executor_creator.pytensorrt_llm/_torch/pyexecutor/kv_cache_connector.pytensorrt_llm/_torch/pyexecutor/model_engine.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/pyexecutor/py_executor_creator.pytensorrt_llm/_torch/pyexecutor/kv_cache_connector.pytensorrt_llm/_torch/pyexecutor/model_engine.py
🧬 Code graph analysis (3)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py (2)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
register_forward_pass_callback(474-476)tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py (1)
register_forward_pass_callback(110-123)
tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py (1)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
register_forward_pass_callback(474-476)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py (4)
KvCacheConnectorWorker(82-187)register_forward_pass_callback(110-123)callback(89-93)ForwardPassCallback(84-93)
svc-trtllm-gh-bot
added
the
Community want to contribute
Signed-off-by: Kyle McGill <[email protected]>
This commit changes the API expected by the model engine to call a user provided callable instead of a predefined class. The result is a more generic API which others can program for. Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
Signed-off-by: Kyle McGill <[email protected]>
pcastonguay
changed the title
|
/bot run --disable-fail-fast |
|
PR_Github #21479 [ run ] triggered by Bot |
|
PR_Github #21479 [ run ] completed with state |
Signed-off-by: Kyle McGill <[email protected]>
|
I'm good to merge this when you all are. Thanks for the help! |
Signed-off-by: Kyle McGill <[email protected]>
|
/bot run --disable-fail-fast --add-multi-gpu-test |
|
/bot kill |
|
PR_Github #21605 [ kill ] triggered by Bot |
|
PR_Github #21605 [ kill ] completed with state |
Signed-off-by: Kyle McGill <[email protected]>
|
/bot run --disable-fail-fast --add-multi-gpu-test |
|
PR_Github #21636 [ run ] triggered by Bot |
|
PR_Github #21636 [ run ] completed with state |
|
/bot run --disable-fail-fast --add-multi-gpu-test |
|
PR_Github #21719 [ run ] triggered by Bot. Commit: |
|
PR_Github #21719 [ run ] completed with state |
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]>
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: yufeiwu-nv <[email protected]>
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]>
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]>
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]>
…DIA#8275) Signed-off-by: Kyle McGill <[email protected]> Signed-off-by: Kyle McGill <[email protected]>
7 participants
Summary by CodeRabbit
Description
This PR enables NVIDIA Dynamo's KV Block Manager to register a forward pass cuda event and associated callback through the connector worker api.
Upon finishing the forward pass, TRTLLM now optionally records a user supplied cuda event on the same stream as the forward pass. Immediately thereafter, the user's provided callback is invoked. which (optionally) waits on the event.
In conjunction with: ai-dynamo/dynamo#3297
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.