[AMD] MINIMAX-M3 vLLM Agentic Support. - #631
Open
ajith-sirra-amd wants to merge 3 commits into
Open
Conversation
…x-M3.yaml File. Signed-off-by: ajith-sirra-amd <ajith.sirra@amd.com>
Contributor
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request adds a new 'KV Cache Offloading' section to the MiniMax-M3.yaml guide, detailing Simple CPU offload and LMCache configurations, along with installation steps and tuning notes. The review feedback suggests correcting the lmcache.mp.host parameter by removing the tcp:// prefix to avoid connection failures, and changing the "lazy_offload" value from a string to a boolean literal in the JSON configuration.
…x-M3.yaml File. Signed-off-by: ajith-sirra-amd <ajith.sirra@amd.com>
Contributor
Contributor
Author
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Adds a KV Cache Offloading guide section to the MiniMax-M3 recipe for AMD MI355X agentic workloads.
Covers two backends (both available on the KV Offload row in the command builder — no per-recipe opt-in needed):
SimpleCPUOffloadConnector, zero extra infrastructure, ~220 GiB host RAM per GPU rank.LMCacheMPConnectorwith a shared CPU-DRAM pool via a companion LMCache server process. ROCm/HIP build path included. Single-node strategies only.Includes MI355X-specific tuning from the validated InferenceX agentic benchmark:
--chunk-size 256(beats 32 at high concurrency / 1M context)--l1-read-ttl-seconds 7200(default 300 s TTL too short for TP8/conc32 agentic queues)--max-workers TP × 2,PYTHONHASHSEED=0,LMCACHE_BLOCKING_TIMEOUT_SECS=60Files changed
models/MiniMaxAI/MiniMax-M3.yaml— guide section only; KV offload scaffolding already in place from [Feature] Interactive KV Offload row: Simple / Mooncake / LMCache #628.