Skip to content

feat: improve cuda shared memory tensor handling.#1222

Merged
RobbieLeung merged 1 commit intojd-opensource:mainfrom
RobbieLeung:feat/cuda_shm
Apr 14, 2026
Merged

feat: improve cuda shared memory tensor handling.#1222
RobbieLeung merged 1 commit intojd-opensource:mainfrom
RobbieLeung:feat/cuda_shm

Conversation

@RobbieLeung
Copy link
Copy Markdown
Collaborator

No description provided.

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request introduces a structured serialization and deserialization mechanism for forward input data, utilizing a layout that separates metadata descriptors from a tensor arena to improve memory management and support CUDA zero-copy. The review feedback identifies several opportunities to better align with the project's style guide, specifically regarding naming conventions for struct members (using trailing underscores), preferring emplace_back for vector operations, and optimizing loops by hoisting conditionals or using function pointer casts instead of verbose lambdas.

Comment thread xllm/core/runtime/forward_shared_memory_manager.cpp
Comment thread xllm/core/runtime/forward_shared_memory_manager.cpp Outdated
Comment thread xllm/core/runtime/forward_shared_memory_manager.cpp Outdated
Comment thread xllm/core/runtime/forward_shared_memory_manager.cpp Outdated
@RobbieLeung RobbieLeung merged commit a7a52bb into jd-opensource:main Apr 14, 2026
7 of 23 checks passed
@RobbieLeung RobbieLeung deleted the feat/cuda_shm branch April 14, 2026 07:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants