Skip to content

[NV] Add MiniMax M3 B300 Dynamo vLLM recipes #1787

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

Closed
jasonlizhengjian wants to merge 7 commits into from
+1,511 −1
Closed

[NV] Add MiniMax M3 B300 Dynamo vLLM recipes #1787

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
de20fa8
Add MiniMax M3 B300 Dynamo vLLM recipes
jasonlizhengjian Jun 15, 2026
131639e
Update MiniMax M3 B300 changelog link
jasonlizhengjian Jun 15, 2026
a2d9824
Add MiniMax M3 B300 8k Dynamo vLLM recipes
jasonlizhengjian Jun 16, 2026
7dbec24
[NV][MiniMax-M3]point to sa-submission-q2-2026 branch instead (#1822)
biswapanda Jun 18, 2026
371117a
Merge branch 'main' into nv/jasonli/minimaxm3-fp8-b300-dynamo-vllm
xinli-sw Jun 18, 2026
8abe295
[NV][MiniMax-M3]fix UCX_ settings to fix nixl handshake failure (#1823)
biswapanda Jun 18, 2026
1ac5daa
Merge branch 'main' into nv/jasonli/minimaxm3-fp8-b300-dynamo-vllm
Oseltamivir Jun 20, 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
226 changes: 226 additions & 0 deletions .github/configs/nvidia-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12817,6 +12817,232 @@ minimaxm2.5-fp8-b300-dynamo-vllm:
ep: 4
dp-attn: true

minimaxm3-fp8-b300-dynamo-vllm:
image: vllm/vllm-openai:minimax-m3
model: MiniMaxAI/MiniMax-M3-MXFP8
model-prefix: minimaxm3
runner: b300
precision: fp8
framework: dynamo-vllm
multinode: true
disagg: true
scenarios:
fixed-seq-len:
- isl: 1024
osl: 1024
search-space:
- conc-list: [4, 16, 64, 128, 4096]
prefill:
num-worker: 1
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/1p1d-dep2-tep8-1k1k.yaml"
decode:
num-worker: 1
tp: 8
ep: 8
dp-attn: false
- conc-list: [1, 4, 8, 16]
prefill:
num-worker: 1
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/1p1d-dep2-tp8-marlin-1k1k.yaml"
decode:
num-worker: 1
tp: 8
ep: 1
dp-attn: false
- conc-list: [2048]
prefill:
num-worker: 1
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/1p2d-dep2-dep4-1k1k.yaml"
decode:
num-worker: 2
tp: 4
ep: 4
dp-attn: true
- conc-list: [512, 4096]
prefill:
num-worker: 2
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/2p1d-dep2-dep8-1k1k.yaml"
decode:
num-worker: 1
tp: 8
ep: 8
dp-attn: true
- conc-list: [32]
prefill:
num-worker: 2
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/2p1d-dep2-tep8-1k1k.yaml"
decode:
num-worker: 1
tp: 8
ep: 8
dp-attn: false
- conc-list: [16]
prefill:
num-worker: 2
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/2p2d-dep2-tep8-1k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: false
- conc-list: [4]
prefill:
num-worker: 3
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/1k1k/3p2d-dep2-tep8-1k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: false
- isl: 8192
osl: 1024
search-space:
- conc-list: [128]
prefill:
num-worker: 1
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/1p2d-dep2-dep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: true
- conc-list: [256, 512]
prefill:
num-worker: 2
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/2p2d-dep2-dep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: true
- conc-list: [16]
prefill:
num-worker: 2
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/2p2d-dep2-tep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: false
- conc-list: [512]
prefill:
num-worker: 3
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/3p2d-dep2-dep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: true
- conc-list: [32]
prefill:
num-worker: 3
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/3p2d-dep2-tep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: false
- conc-list: [4096]
prefill:
num-worker: 4
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/4p2d-dep2-dep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: true
- conc-list: [4096]
prefill:
num-worker: 4
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/4p3d-dep2-dep4-8k1k.yaml"
decode:
num-worker: 3
tp: 4
ep: 4
dp-attn: true
- conc-list: [4, 64]
prefill:
num-worker: 5
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/5p2d-dep2-tep8-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 8
dp-attn: false
- conc-list: [1, 4, 8, 16]
prefill:
num-worker: 5
tp: 2
ep: 2
dp-attn: true
additional-settings:
- "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp8/8k1k/5p2d-dep2-tp8-marlin-8k1k.yaml"
decode:
num-worker: 2
tp: 8
ep: 1
dp-attn: false

minimaxm2.5-fp8-gb300-dynamo-vllm:
image: vllm/vllm-openai:v0.20.1
model: MiniMaxAI/MiniMax-M2.5
Expand Down
79 changes: 79 additions & 0 deletions ...marks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp8/1k1k/1p1d-dep2-tep8-1k1k.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,79 @@
name: "minimax-m3-vllm-disagg-b300-1p1d-fp8-dep2-tep8-1k1k"

model:
path: "MiniMaxAI/MiniMax-M3-MXFP8"
container: "vllm/vllm-openai:minimax-m3"
precision: "fp8"

resources:
gpu_type: "b300"
gpus_per_node: 8
prefill_nodes: 1
decode_nodes: 1
prefill_workers: 1
decode_workers: 1
gpus_per_prefill: 2
gpus_per_decode: 8

dynamo:
install: true
version: 1.3.0.dev20260614

frontend:
type: dynamo
enable_multiple_frontends: false

backend:
type: vllm
connector: null

prefill_environment:
UCX_TLS: "cuda_copy,rc"
VLLM_FLOAT32_MATMUL_PRECISION: "high"

decode_environment:
UCX_TLS: "cuda_copy,rc"
VLLM_FLOAT32_MATMUL_PRECISION: "high"

vllm_config:
prefill:
tensor-parallel-size: 1
data-parallel-size: 2
data-parallel-rpc-port: 13345
enable-expert-parallel: true
trust-remote-code: true
no-enable-prefix-caching: true
kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
block-size: 128
gpu-memory-utilization: 0.90
max-model-len: 2304
language-model-only: true
stream-interval: 32
max-cudagraph-capture-size: 2048
max-num-batched-tokens: 2048

decode:
tensor-parallel-size: 8
enable-expert-parallel: true
trust-remote-code: true
no-enable-prefix-caching: true
kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
block-size: 128
gpu-memory-utilization: 0.90
max-model-len: 2304
language-model-only: true
stream-interval: 32
max-num-seqs: 4096
max-num-batched-tokens: 16384
max-cudagraph-capture-size: 8196

@cursor cursor Bot Jun 16, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wrong cudagraph capture size

Medium Severity

All six new MiniMax M3 B300 decode blocks set max-cudagraph-capture-size to 8196, while prefill uses 2048 and decode sets max-num-seqs to 4096. That value is not used elsewhere in the repo and sits four above the usual power-of-two 8192 paired with 4096-sequence decode configs, so CUDA graph capture may not align with intended batch sizes.

Additional Locations (2)
Fix in Cursor Fix in Web

Reviewed by Cursor Bugbot for commit 54b829a. Configure here.


health_check:
max_attempts: 360
interval_seconds: 10

benchmark:
type: "sa-bench"
isl: 1024
osl: 1024
concurrencies: "4x16x64x128x4096"
req_rate: "inf"
80 changes: 80 additions & 0 deletions ...multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp8/1k1k/1p1d-dep2-tp8-marlin-1k1k.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,80 @@
name: "minimax-m3-vllm-disagg-b300-1p1d-fp8-dep2-tp8-marlin-1k1k"

model:
path: "MiniMaxAI/MiniMax-M3-MXFP8"
container: "vllm/vllm-openai:minimax-m3"
precision: "fp8"

resources:
gpu_type: "b300"
gpus_per_node: 8
prefill_nodes: 1
decode_nodes: 1
prefill_workers: 1
decode_workers: 1
gpus_per_prefill: 2
gpus_per_decode: 8

dynamo:
install: true
version: 1.3.0.dev20260614

frontend:
type: dynamo
enable_multiple_frontends: false

backend:
type: vllm
connector: null

prefill_environment:
UCX_TLS: "cuda_copy,rc"
VLLM_FLOAT32_MATMUL_PRECISION: "high"

decode_environment:
UCX_TLS: "cuda_copy,rc"
VLLM_FLOAT32_MATMUL_PRECISION: "high"

vllm_config:
prefill:
tensor-parallel-size: 1
data-parallel-size: 2
data-parallel-rpc-port: 13345
enable-expert-parallel: true
trust-remote-code: true
no-enable-prefix-caching: true
kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
block-size: 128
gpu-memory-utilization: 0.90
max-model-len: 2304
language-model-only: true
stream-interval: 32
max-cudagraph-capture-size: 2048
max-num-batched-tokens: 2048

decode:
tensor-parallel-size: 8
enable-expert-parallel: false
moe-backend: marlin
trust-remote-code: true
no-enable-prefix-caching: true
kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
block-size: 128
gpu-memory-utilization: 0.90
max-model-len: 2304
language-model-only: true
stream-interval: 32
max-num-seqs: 4096
max-num-batched-tokens: 16384
max-cudagraph-capture-size: 8196

health_check:
max_attempts: 360
interval_seconds: 10

benchmark:
type: "sa-bench"
isl: 1024
osl: 1024
concurrencies: "1x4x8x16"
req_rate: "inf"
Loading
Loading