[NV] Refresh GB300 DeepSeek-V4-Pro AgentX with SGLang DSpark6 / 使用 SGLang DSpark6 更新 GB300 DeepSeek-V4-Pro AgentX - #2623
[NV] Refresh GB300 DeepSeek-V4-Pro AgentX with SGLang DSpark6 / 使用 SGLang DSpark6 更新 GB300 DeepSeek-V4-Pro AgentX#2623weireweire wants to merge 2 commits into
Conversation
b8b5b29 to
75c1056
Compare
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
Additional findings (outside current diff — PR may have been updated during review):
-
🟡
perf-changelog.yaml:5452-5461— AGENTS.md line 7 requires bilingual PR titles/descriptions (<English title> / <中文标题>plus a Chinese mirror section like## 中文说明), but PR #2623's title ("[NV] Refresh GB300 DeepSeek-V4-Pro AgentX with SGLang DSpark6") and body are English-only. Recently merged PRs (#2493, #2486) follow this bilingual convention — please add a Chinese title suffix and a## 中文说明section mirroring the summary.Extended reasoning...
What the rule is:
AGENTS.md(the single source of truth this repo'sCLAUDE.mdpoints to for all guidance) states verbatim in line 7: "PR and GitHub-issue titles & descriptions must be bilingual — include a Simplified Chinese version in addition to English. Title format:<English title> / <中文标题>. In the PR/issue body, follow the English content with its Chinese translation (e.g. a## 中文说明section mirroring the summary...)." This applies to every PR in the repository, not an optional style suggestion.Where it is violated: PR #2623's title is
[NV] Refresh GB300 DeepSeek-V4-Pro AgentX with SGLang DSpark6— no/ <中文标题>suffix. The description contains only## Summary,## Validation, and## Referencesections in English, with no## 中文说明(or equivalent) mirroring the summary.Why this is not a false positive: The convention is actively followed elsewhere in this same repo's recent history.
git logshows PR #2493 titled"[Klaud Cold] Enact the August 3, 2026 scenario and precision deprecations / 执行 2026 年 8 月 3 日场景与精度下线"and PR #2486 titled"[Klaud Cold] minimaxm3-fp8-gb300-dynamo-vllm-mtp: day-zero GB300 MXFP8 EAGLE3 MTP + FULL_DECODE_ONLY CG / 新增 GB300 MXFP8 EAGLE3 MTP 配方，解码启用 FULL_DECODE_ONLY 图模式"— both bilingual per the mandated format. This confirms the rule is a live, enforced practice, and PR #2623 is an outlier that skipped it.Step-by-step proof:
- Open
AGENTS.mdat the repo root, line 7 — read the bilingual title/description requirement. - Open PR #2623's title:
[NV] Refresh GB300 DeepSeek-V4-Pro AgentX with SGLang DSpark6. Search for/followed by CJK characters — none found. - Open PR #2623's description: three English-only
##sections (Summary,Validation,Reference). Search for中文or## 中文说明— none found. - Compare against
git log --oneline -5: #2493 and #2486 both show the<English> / <中文>pattern in their subject lines, confirming this is the repo's actual practice, not a dead-letter rule. - Conclusion: PR #2623 is out of compliance with an explicit, currently-enforced repository convention.
Impact and fix: This does not affect the functional correctness of the recipe/config changes in the diff — it is purely a process/documentation gap. The fix is mechanical: rename the PR title to append
/ <中文标题>(e.g.,/ 刷新 GB300 DeepSeek-V4-Pro AgentX 至 SGLang DSpark6) and add a## 中文说明section to the body mirroring the English summary bullets. Since every original source bug for this finding is a[quality]-prefixed convention/documentation issue with no correctness impact, it should be gradednit. - Open
c8261b2 to
e5935c1
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31924886695 |
e5935c1 to
033ef73
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31925668643 |
033ef73 to
ab65e7f
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31925892725 |
ab65e7f to
c68dad8
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31926958601 |
weireweire
changed the title
c68dad8 to
8a4a0a0
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31929326055 |
8a4a0a0 to
d9a6672
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31950260070 |
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31958640004 |
044f8dc to
91e9096
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31959753035 |
a3adb18 to
d94a943
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31967702713 |
d94a943 to
71b9873
Compare
Root cause: - The GB300 DeepSeek-V4-Pro AgentX matrix still used the older EAGLE/MTP serving path and an outdated SGLang nightly. - The previous draft targeted vLLM recipes instead of the existing SGLang topology ladder. Fix: - Replace the SGLang aggregate and disaggregated GB300 recipes with DSpark block size 6 on DeepSeek-V4-Pro-DSpark. - Use nightly-dev-cu13-20260816-4a6dc267, which already includes the required DSpark performance and MegaMoE support, without a runtime patch overlay. - Force thinking-on AL 3.82 with static ragged verification, retain MegaMoE on DEP workers, and set the 8P4D throughput endpoint to concurrency 1536. Validation: - Passed srtctl dry-run for all seven recipes. - Passed 224 matrix and schema tests. - Generated the eight targeted matrix jobs at concurrency 1, 4, 8, 32, 192, 400, 640, and 1536. 中文：将 GB300 DeepSeek-V4-Pro AgentX 的 SGLang 配置从 EAGLE/MTP 更新为 DSpark6，使用已包含所需上游修复的 20260816 nightly，不再注入补丁；固定 AL 3.82 与 static ragged verify，并将 8P4D 吞吐端点设为并发 1536。七份 recipe dry-run、224 项矩阵与 schema 测试以及八个目标矩阵任务生成均已通过。
71b9873 to
1539767
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31971449638 |
1539767 to
1ab7a23
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31993388848 |
1ab7a23 to
346c900
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31994228657 |
Root cause: - The sweep still referenced a nonstandard model/tokenizer path, included a low-value concurrency point, and allowed synthetic acceptance controls to leak into evaluation. - Aggregate TP checkpoint loading and long-lived AgentX replay connections could stall or expire during startup and multi-turn idle gaps. - Worker cache metrics were not exposed consistently across the recipe ladder. Fix: - Use the standard DeepSeek-V4-Pro-0813 model layout and probabilistic DSpark acceptance data, preserve real evaluation behavior, and keep the validated 1152-concurrency endpoint. - Enable cache reporting on every DSpark worker and checkpoint prefetch on pure-TP aggregate workers. - Use srt-slurm v1.0.40 and a 900-second nginx/AIPerf connection timeout for the 1152-concurrency recipe. - Retain the aggregate startup barrier and focused regression coverage. Validation: - Passed YAML and shell syntax validation, git diff checks, and the synthetic-acceptance launcher test. - Passed exact-key NVIDIA matrix expansion. - Passed srtctl dry-run for all six generated recipes. 中文：统一使用标准 DeepSeek-V4-Pro-0813 模型布局和 0813 DSpark 接收长度；所有 worker 开启 cache report，纯 TP 配置开启 checkpoint prefetch，并用 srt-slurm v1.0.40 与 900 秒连接保活修复高并发多轮回放连接超时。相关矩阵展开、六份 recipe dry-run、YAML/Shell 校验和 launcher 测试均已通过。
346c900 to
86a9eaa
Compare
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=31996425607 |
1 participant
Summary
deepseek-ai/DeepSeek-V4-Pro-DSparkandlmsysorg/sglang:nightly-dev-cu13-20260816-4a6dc267. This nightly already contains the required DSpark performance and MegaMoE changes, so no runtime patch is applied.SGLANG_RAGGED_VERIFY_MODE=staticand real DSpark verification without synthetic acceptance.Validation
srtctl dry-runpassed for all six recipes.Reference
中文说明
摘要
deepseek-ai/DeepSeek-V4-Pro-DSpark和lmsysorg/sglang:nightly-dev-cu13-20260816-4a6dc267。该 nightly 已包含所需的 DSpark 性能优化和 MegaMoE 改动，因此不再在运行时应用补丁。SGLANG_RAGGED_VERIFY_MODE=static，使用真实 DSpark 验证且不注入模拟接受长度。验证
srtctl dry-run。参考