Skip to content

feat: add mlu host kv cache support.#2012

Draft
phantomlei3 wants to merge 2 commits into
xLLM-AI:mainfrom
phantomlei3:feat/host-kv-cache-mlu
Draft

feat: add mlu host kv cache support.#2012
phantomlei3 wants to merge 2 commits into
xLLM-AI:mainfrom
phantomlei3:feat/host-kv-cache-mlu

Conversation

@phantomlei3

Copy link
Copy Markdown
Collaborator

Description

Related Issues

Change Type

  • Bug fix
  • New feature
  • Performance improvement
  • Refactor
  • Documentation
  • Test
  • Build or CI

Pull Request Checklist

Thank you for contributing to xLLM. Before requesting review, please make sure the following items are complete.

PR Title and Commit Messages

  • The PR title and each commit message follow the xLLM commit format: <type>: <subject>.

Allowed types: feat, bugfix, docs, test, refactor, chore, style, revert, perf, model, build, release.
The subject should use clear English, start with a verb, include at least 4 words, and end with ..

Pre-commit Checks

  • I have installed pre-commit by running pip install pre-commit or an equivalent command.
  • I have installed the hooks with pre-commit install.
  • I have run pre-commit run --all-files and fixed any reported issues.

If you are unsure how to set up pre-commit, see the pre-commit documentation.

Self Review

  • I have self-reviewed the code according to .agents/skills/code-review/references/custom-code-style.md, especially code written or assisted by AI.
  • I have rebased this PR onto the latest main branch.

Build and Test Coverage

  • Tests have been added or updated as needed.
  • CUDA: python setup.py build test has passed on a CUDA machine.
  • NPU: python setup.py build test has passed on an NPU machine.
  • MLU: python setup.py build test has passed on an MLU machine.

Reviewer Notes

- add pinned host allocation and batched MLU cache transfers
- synchronize cache-consuming model layers and reject unsupported layouts
- cover allocation, transfer, synchronization, and capability gates with tests
- restore the DeepSeek-V4 host offload rejection in the engine
- require graph-off and automatic KV/indexer cache dtypes in the common path
- remove the MLU-only capability validator and replace its tests
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.

1 participant