Skip to content

test the GB300 cluster after the node patch #1961

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
richardhuo-nv wants to merge 6 commits into
base: main
Choose a base branch
from
Open

test the GB300 cluster after the node patch #1961

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
88feffc
fix
richardhuo-nv Jun 30, 2026
ec81194
Merge branch 'main' into rihuo/test-gb300-cluster
richardhuo-nv Jun 30, 2026
b7c65f3
resolve conflict
richardhuo-nv Jun 30, 2026
c292fcc
fix newline
richardhuo-nv Jun 30, 2026
56f730d
fix newline
richardhuo-nv Jun 30, 2026
d339fce
resolve conflict
richardhuo-nv Jun 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions .github/configs/nvidia-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12362,14 +12362,14 @@ minimaxm3-fp8-gb300-dynamo-vllm:
osl: 1024
search-space:
# 1p1d DEP2+DEP8, 3n: conc 256
- conc-list: [256]
- conc-list: [256, 512]
prefill:
num-worker: 1
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-dep8-8k1k.yaml"
- "CONFIG_FILE=recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-dep8-8k1k-healthcheck.yaml"
decode:
num-worker: 1
tp: 8
Expand Down
4 changes: 2 additions & 2 deletions ...3-gb300-fp8/8k1k/1p1d-dep2-dep8-8k1k.yaml → ...8k1k/1p1d-dep2-dep8-8k1k-healthcheck.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ dynamo:
version: 1.3.0.dev20260614

health_check:
max_attempts: 720
max_attempts: 840
interval_seconds: 10

sbatch_directives:
Expand Down Expand Up @@ -103,5 +103,5 @@ benchmark:
type: "sa-bench"
isl: 8192
osl: 1024
concurrencies: "256"
concurrencies: "256x512"
req_rate: "inf"
6 changes: 6 additions & 0 deletions perf-changelog.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4343,3 +4343,9 @@
- "Use nvidia/MiniMax-M3-NVFP4 from /scratch/models/MiniMax-M3-NVFP4 with vllm/vllm-openai:vllm-minimax-m3-perf-x86_64-13.0.1-8b00f41, which includes vllm-project/vllm PR #46380; no runtime patch needed"
- "Reuse the existing MXFP8 B300 topology and concurrency matrix across 15 srt-slurm recipes, while dropping the FP8-only Marlin override from TP4 decode"
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1931

- config-keys:
- minimaxm3-fp8-gb300-dynamo-vllm
description:
- "Rename 1p1d-dep2-dep8-8k1k.yaml to 1p1d-dep2-dep8-8k1k-healthcheck.yaml and update the nvidia-master.yaml CONFIG_FILE pointer accordingly; increase health_check max_attempts from 720 to 840 (7200s → 8400s) to accommodate longer GB300 model load times."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1961
Loading