Skip to content

[Klaud Cold] Update dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg vLLM image to v0.29.0 / 将 dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg 的 vLLM 镜像更新至 v0.29.0 #3033

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
Klaud-Cold wants to merge 2 commits into from
Closed

[Klaud Cold] Update dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg vLLM image to v0.29.0 / 将 dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg 的 vLLM 镜像更新至 v0.29.0 #3033

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
393a0ca
feat: update dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg vLLM image t…
Klaud-Cold Sep 12, 2026
2b1f638
fix: lower GB200 DSV4 AgentX MTP agg gpu-memory-utilization to 0.90 f…
Klaud-Cold Sep 12, 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
10 changes: 7 additions & 3 deletions .../multi_node/srt-slurm-recipes/vllm/deepseek-v4/agentic/agg-gb200-tp8-c4-mtp2-agentic.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,14 +6,14 @@ name: "svf-vllm-agg-gb200-tp8-c4-mtp2-agentic"

model:
path: "deepseek-v4-pro"
container: "vllm/vllm-openai:nightly-dev-arm64-cu13.0.1-426e59f"
container: "vllm/vllm-openai:v0.29.0"
precision: "fp4"

identity:
model:
repo: "deepseek-ai/DeepSeek-V4-Pro"
container:
image: "vllm/vllm-openai:nightly-dev-arm64-cu13.0.1-426e59f"
image: "vllm/vllm-openai:v0.29.0"
frameworks:
dynamo: "1.2.1"

Expand Down Expand Up @@ -120,7 +120,11 @@ backend:
block-size: 256
compilation-config: '{"cudagraph_mode":"FULL_DECODE_ONLY","cudagraph_capture_sizes":[4,8,12,16,20,24,28,32,36,40,44,48,52,56,60,64],"mode":0}'
speculative-config: '{"method":"mtp","num_speculative_tokens":3}'
gpu-memory-utilization: 0.94
# vLLM v0.29.0 profiles only 3.8 GiB of peak activation and hands the
# rest to the KV cache; keep headroom for the lazily allocated sparse
# indexer logits workspace (VLLM_SPARSE_INDEXER_MAX_LOGITS_MB) that
# OOMed at 0.94 on the first long AgentX prefill.
gpu-memory-utilization: 0.90
stream-interval: 10
no-disable-hybrid-kv-cache-manager: true
tokenizer-mode: "deepseek_v4"
Expand Down
10 changes: 7 additions & 3 deletions .../multi_node/srt-slurm-recipes/vllm/deepseek-v4/agentic/agg-gb200-tp8-c8-mtp2-agentic.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,14 +6,14 @@ name: "svf-vllm-agg-gb200-tp8-c8-mtp2-agentic"

model:
path: "deepseek-v4-pro"
container: "vllm/vllm-openai:nightly-dev-arm64-cu13.0.1-426e59f"
container: "vllm/vllm-openai:v0.29.0"
precision: "fp4"

identity:
model:
repo: "deepseek-ai/DeepSeek-V4-Pro"
container:
image: "vllm/vllm-openai:nightly-dev-arm64-cu13.0.1-426e59f"
image: "vllm/vllm-openai:v0.29.0"
frameworks:
dynamo: "1.2.1"

Expand Down Expand Up @@ -119,7 +119,11 @@ backend:
block-size: 256
compilation-config: '{"cudagraph_mode":"FULL_DECODE_ONLY","cudagraph_capture_sizes":[4,8,12,16,20,24,28,32,36,40,44,48,52,56,60,64,68,72,76,80,84,88,92,96,100,104,108,112,116,120,124,128],"mode":0}'
speculative-config: '{"method":"mtp","num_speculative_tokens":3}'
gpu-memory-utilization: 0.94
# vLLM v0.29.0 profiles only 3.8 GiB of peak activation and hands the
# rest to the KV cache; keep headroom for the lazily allocated sparse
# indexer logits workspace (VLLM_SPARSE_INDEXER_MAX_LOGITS_MB) that
# OOMed at 0.94 on the first long AgentX prefill.
gpu-memory-utilization: 0.90
stream-interval: 10
no-disable-hybrid-kv-cache-manager: true
tokenizer-mode: "deepseek_v4"
Expand Down
2 changes: 1 addition & 1 deletion configs/nvidia-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8497,7 +8497,7 @@ dsv4-fp4-gb300-dynamo-vllm-agentic-mtp-disagg:
# mtp2 keys/recipe files so the existing GB200 AgentX MTP recipes above are
# untouched.
dsv4-fp4-gb200-dynamo-vllm-agentic-mtp2-agg:
image: vllm/vllm-openai:nightly-dev-arm64-cu13.0.1-426e59f
image: vllm/vllm-openai:v0.29.0
model: deepseek-ai/DeepSeek-V4-Pro
model-prefix: dsv4
runner: cluster:gb200-nv
Expand Down