Skip to content

[Klaud Cold] Update dsv4-fp4-b200-sglang-agentic-hicache-mtp SGLang image to v0.5.19-cu130 / 将 dsv4-fp4-b200-sglang-agentic-hicache-mtp 的 SGLang 镜像更新至 v0.5.19-cu130 #3012

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
Klaud-Cold wants to merge 3 commits into
base: main
Choose a base branch
from
+10 −1
Open

[Klaud Cold] Update dsv4-fp4-b200-sglang-agentic-hicache-mtp SGLang image to v0.5.19-cu130 / 将 dsv4-fp4-b200-sglang-agentic-hicache-mtp 的 SGLang 镜像更新至 v0.5.19-cu130 #3012

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
82ded89
config: update dsv4-fp4-b200-sglang-agentic-hicache-mtp SGLang image …
Klaud-Cold Sep 11, 2026
8209184
changelog: record dsv4-fp4-b200-sglang-agentic-hicache-mtp image upda…
Klaud-Cold Sep 11, 2026
da239a6
chore: merge main into the dsv4-fp4-b200-sglang-agentic-hicache-mtp i…
Klaud-Cold Sep 11, 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
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 @@ -929,7 +929,7 @@ dsv4-fp4-b200-sglang:
- { tp: 8, ep: 8, dp-attn: true, conc-start: 256, conc-end: 1024 }

dsv4-fp4-b200-sglang-agentic-hicache-mtp:
image: lmsysorg/sglang:nightly-dev-20260827-20621aa1
image: lmsysorg/sglang:v0.5.19-cu130
model: deepseek-ai/DeepSeek-V4-Pro-0813
model-prefix: dsv4
runner: cluster:b200-nscale
Expand Down
9 changes: 9 additions & 0 deletions perf-changelog.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7286,3 +7286,12 @@
description:
- "Use thinking-on golden synthetic AL 3.51 for five-token DSpark throughput; disable adaptive verification and retain real verification for evals"
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2974

- config-keys:
- dsv4-fp4-b200-sglang-agentic-hicache-mtp
scenario-type:
- agentic-coding
description:
- "Update the B200 DeepSeek-V4-Pro-0813 SGLang AgentX image from lmsysorg/sglang:nightly-dev-20260827-20621aa1 (2026-08-27 dev nightly, sglang commit 20621aa1) to the lmsysorg/sglang:v0.5.19-cu130 release (Docker Hub digest sha256:d6e7288627be8b02be88e4bba38e73f6d50e2826869f753c13a4c4385ab3eda9, release commit 0bcd8223, published 2026-09-05). The release branch contains every commit of the pinned nightly."
- "benchmarks/single_node/agentic/dsv4_fp4_b200_sglang_mtp.sh is unchanged: DSpark block size 6 with one speculative step and seven draft tokens, golden synthetic acceptance length 3.77, HiCache ratio 2.75 for TP8 and 8 for DP-attention, W4A4 MXFP4 MegaMoE plus the FP4 indexer on the DP-attention path, and sglang-router 0.3.2 cache-aware routing; the search space and cluster:b200-nscale runner are unchanged. Between the two images every server flag and SGLANG_* variable used by the script keeps its definition; --enable-w4a4-mxfp4-megamoe now selects DeepGEMM's mxf4xmxf4 MMA type directly instead of exporting DG_USE_FP4_ACTS/DG_USE_MXF4_KIND, DeepGEMM moves from 0.1.5.post2 to 0.1.7 and FlashInfer from 0.6.17 to 0.6.18."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/3012
Loading