Add Kimi K2.5 disagg STP and MTP recipes for GB200 NVfp4 (ISL8K_OSL1K…#24
Merged
nlevin-ui merged 6 commits intoNVIDIA:sa-submission-q2-2026from Apr 10, 2026
Conversation
… and ISL1K_OSL1K) Add optimized disaggregated inference recipes for Kimi K2.5 model with NVfp4 precision on GB200 GPUs. Includes both STP and MTP configurations for ISL8K_OSL1K and ISL1K_OSL1K workloads covering concurrency points from 5 to 2253, with Eagle speculative decoding for MTP variants.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## sa-submission-q2-2026 #24 +/- ##
========================================================
Coverage ? 58.54%
========================================================
Files ? 47
Lines ? 4048
Branches ? 0
========================================================
Hits ? 2370
Misses ? 1678
Partials ? 0 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
yeswanthk-26
force-pushed
the
yeswanthk/disagg-kimi2.5-gb200-Nvfp4
branch
from
e788dcf to
61f73c2
Compare
…and env cleanup - Update container to tensorrtllm-runtime-1.1.0-dev.2.sqsh - Point model path to shared /mnt/lustre01/models/kimi-k2.5-nvfp4 - Update Eagle model mount path for MTP configs - Remove HF_HOME (defaults to ~/.cache/huggingface) - Fix concurrency separator from space to 'x' for sa-bench compatibility - Enable multiple frontends for ctx1dep4_gen1dep32_batch64
yeswanthk-26
force-pushed
the
yeswanthk/disagg-kimi2.5-gb200-Nvfp4
branch
2 times, most recently
from
b355149 to
493b0ee
Compare
nlevin-ui
reviewed
Apr 10, 2026
| # concurrency: 666 | ||
|
|
||
| model: | ||
| path: "/mnt/lustre01/models/kimi-k2.5-nvfp4" |
Collaborator
There was a problem hiding this comment.
The paths and containers should not be cluster specific. The srtslurm.yaml should map the generic names to the specific cluster names.
path: "kimi-k2.5-nvfp4"
container: "vllm/vllm-openai:v0.18.0-cu130"
as seen here: https://github.com/NVIDIA/srt-slurm/blob/sa-submission-q2-2026/recipes/vllm/kimi-k2.5/1k1k/disagg-gb200-1p1d-dep4-dep16.yaml#L4C1-L5C46
Replace cluster-specific paths with generic alias names that are resolved via srtslurm.yaml model_paths and containers mappings, as per upstream convention.
Add model_paths alias resolution for extra_mount host paths in config.py, enabling MTP recipes to use generic name "kimi-k2.5-eagle3" instead of cluster-specific path for the Eagle speculative decoding model.
nlevin-ui
approved these changes
Apr 10, 2026
Per review feedback, update model paths to HuggingFace format (nvidia/Kimi-K2.5-NVFP4) and container to full NVCR registry path (nvcr.io/nvidia/ai-dynamo/tensorrtllm-runtime:1.1.0-dev.2) so recipes are portable and work without pre-built sqsh files.
yeswanthk-26
force-pushed
the
yeswanthk/disagg-kimi2.5-gb200-Nvfp4
branch
from
502d746 to
670289c
Compare
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.
Add optimized disaggregated inference recipes for Kimi K2.5 model with NVfp4 precision on GB200 GPUs. Includes both STP and MTP configurations for ISL8K_OSL1K and ISL1K_OSL1K workloads covering concurrency points from 5 to 2253, with Eagle speculative decoding for MTP variants.