[Runtime] DeepSeek v4 pro multi node runtimes#631
Open
YouNeedCryDear wants to merge 1 commit into
Open
Conversation
YouNeedCryDear
requested review from
CatherineSue,
XinyueZhang369 and
slin1237
as code owners
github-actions
Bot
added
runtime
YouNeedCryDear
force-pushed
the
feat/deepseek-v4-pro-multi-node
branch
from
5862827 to
530087c
Compare
YouNeedCryDear
changed the title
| - --master-addr=$(LWS_LEADER_ADDRESS) | ||
| - --gpu-memory-utilization=0.95 | ||
| - --max-num-seqs=256 | ||
| - --max-num-batched-tokens=512 |
Collaborator
Author
There was a problem hiding this comment.
Yeah, sadly.
| - -cc.pass_config.fuse_allreduce_rms=False | ||
| - --master-addr=$(LWS_LEADER_ADDRESS) | ||
| - --gpu-memory-utilization=0.95 | ||
| - --max-num-seqs=256 |
Contributor
There was a problem hiding this comment.
Does decreasing this increase this improve the batched_tokens?
Collaborator
Author
There was a problem hiding this comment.
decreasing this help with the memory pressure.
shenoyvvarun
approved these changes
Jun 18, 2026
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What this PR does
Adds DeepSeek v4 Pro multi-node runtime configuration:
vllm-deepseek-v4-pro-multiClusterServingRuntimewith an SMG router and vLLM leader/worker engine configuration.config/runtimes/kustomization.yaml.InferenceServicesample fordeepseek-v4-pro-multi.Why we need it
Enables OME users to deploy DeepSeek v4 Pro on a two-node H100 topology with vLLM.
Fixes #
How to test
Not run locally; configuration-only PR submission.
Checklist
make testpasses locally