Skip to content

[Feature]: Integrate AITER MLA V1 from Upstream #62

@tjtanaa

Description

@tjtanaa

🚀 The feature, motivation and pitch

After syncing with ROCm/vllm main, cherry pick the changes from upstream to support AITER MLA Engine V1.

Prepare the command to run

VLLM_ATTENTION_BACKEND=ROCM_AITER_MLA VLLM_USE_V1=1 VLLM_ROCM_USE_AITER=1 vllm serve /amdhome/models/DeepSeek-R1 --trust-remote-code -tp 8 --max-model-len 32768 --block-size 1 --no-enable-prefix-caching --max-num-batched-tokens 32768 --max-num-seqs 1024

Suggest a few configurations for the testing team to benchmark.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions