refactor cuda graph runner #6770
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: junq <[email protected]>
Signed-off-by: junq <[email protected]>
Signed-off-by: junq <[email protected]>
Signed-off-by: junq <[email protected]>
Signed-off-by: junq <[email protected]>
Contributor
|
Caution Review failedThe pull request is closed. 📝 WalkthroughWalkthroughIntroduces a new CUDA graph execution engine for PyTorch LLMs, integrates it into the existing model engine, updates warmup/forward/cleanup paths to delegate CUDA-graph handling to the engine, removes prior bespoke graph logic, and updates a third-party submodule pointer. Changes
Sequence Diagram(s)sequenceDiagram
participant Client
participant PyTorchModelEngine as ModelEngine
participant CUDAGraphModelEngine as CudaGraphEngine
participant CUDA as CUDA Graph
Client->>ModelEngine: forward(batch, inputs)
ModelEngine->>CudaGraphEngine: pad_batch(scheduled_requests)
CudaGraphEngine-->>ModelEngine: padded batch (context)
ModelEngine->>CudaGraphEngine: execute(batch, inputs, forward_fn)
alt first run or missing graph
CudaGraphEngine->>CUDA: capture(warmup + forward)
CUDA-->>CudaGraphEngine: graph handle + outputs ref
else replay
CudaGraphEngine->>CUDA: replay()
CUDA-->>CudaGraphEngine: outputs
end
CudaGraphEngine-->>ModelEngine: graph_output or None
alt got graph_output
ModelEngine-->>Client: graph_output
else fallback
ModelEngine->>ModelEngine: eager forward_fn()
ModelEngine-->>Client: eager output
end
sequenceDiagram
participant Control as Control Flow
participant ModelEngine
participant CudaGraphEngine
Control->>ModelEngine: warmup()
ModelEngine->>CudaGraphEngine: execute(warmup batch, inputs, forward_fn)
CudaGraphEngine->>CudaGraphEngine: capture graph for batch size
CudaGraphEngine-->>ModelEngine: output (ignored/validated)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~35 minutes Possibly related PRs
Suggested reviewers
Note 🔌 MCP (Model Context Protocol) integration is now available in Early Access!Pro users can now connect to remote MCP servers under the Integrations page to get reviews and chat conversations that understand additional development context. 📜 Recent review detailsConfiguration used: .coderabbit.yaml 📒 Files selected for processing (3)
✨ 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
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary by CodeRabbit
New Features
Refactor
Chores