Skip to content

[DSv4] Re-run GB300 dynamo-trt configs / 重跑 GB300 dynamo-trt 配置#2193

Open
Oseltamivir wants to merge 1 commit into
mainfrom
dsv4-gb300-trt-rerun
Open

[DSv4] Re-run GB300 dynamo-trt configs / 重跑 GB300 dynamo-trt 配置#2193
Oseltamivir wants to merge 1 commit into
mainfrom
dsv4-gb300-trt-rerun

Conversation

@Oseltamivir

Copy link
Copy Markdown
Collaborator

Summary

Split out of the closed #2189 — GB300 only, one PR per SKU.

The only GB300 dsv4 (DeepSeek-V4-Pro) 1k1k submissions come from a single 2026-06-15 run. Re-run dsv4-fp4-gb300-dynamo-trt(+mtp) — no config change, changelog-only trigger.

Validation: validate_perf_changelog.py --base-ref main passes.

中文说明

从已关闭的 #2189 拆分而来——仅 GB300,按 SKU 拆分为独立 PR。

GB300 dsv4(DeepSeek-V4-Pro)1k1k 目前仅有 2026-06-15 单日数据。重跑 dsv4-fp4-gb300-dynamo-trt(+mtp)——配置不变,仅通过 changelog 触发。

验证:validate_perf_changelog.py --base-ref main 通过。

The only GB300 dsv4 1k1k submissions are from a single 2026-06-15 run.
Re-run dsv4-fp4-gb300-dynamo-trt(+mtp) (no config change).

中文:重跑 GB300 dsv4 dynamo-trt 配置(配置不变)——目前 GB300 dsv4 1k1k
仅有 2026-06-15 单日数据。
@github-actions

Copy link
Copy Markdown
Contributor

Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase As a PR reviewer and CODEOWNER, I have reviewed this and have.

For PR verification, add the full-sweep-fail-fast label (strongly recommended) to this PR — the benchmark sweep only runs on labeled PRs. Use full-sweep-enabled only if you need matrix jobs to keep running past a failure.

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs


感谢你的贡献!请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md,然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人,你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板,包括保留英文语句 As a PR reviewer and CODEOWNER, I have reviewed this and have

如需进行 PR 验证,请为此 PR 添加 full-sweep-fail-fast 标签(强烈推荐)— 基准测试 sweep 仅在带有标签的 PR 上运行。仅当需要矩阵任务在失败后继续运行时才使用 full-sweep-enabled

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动(flake),重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档

1 similar comment
@github-actions

Copy link
Copy Markdown
Contributor

Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase As a PR reviewer and CODEOWNER, I have reviewed this and have.

For PR verification, add the full-sweep-fail-fast label (strongly recommended) to this PR — the benchmark sweep only runs on labeled PRs. Use full-sweep-enabled only if you need matrix jobs to keep running past a failure.

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs


感谢你的贡献!请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md,然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人,你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板,包括保留英文语句 As a PR reviewer and CODEOWNER, I have reviewed this and have

如需进行 PR 验证,请为此 PR 添加 full-sweep-fail-fast 标签(强烈推荐)— 基准测试 sweep 仅在带有标签的 PR 上运行。仅当需要矩阵任务在失败后继续运行时才使用 full-sweep-enabled

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动(flake),重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档

@Oseltamivir Oseltamivir force-pushed the dsv4-gb300-trt-rerun branch from 8a0c149 to 81c3f3f Compare July 14, 2026 07:00

@claude claude Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

LGTM — changelog-only re-run trigger, no config changes.

Extended reasoning...

Overview

This PR appends a single entry to perf-changelog.yaml to trigger a re-run of the dsv4-fp4-gb300-dynamo-trt and dsv4-fp4-gb300-dynamo-trt-mtp GB300 configs. No source, script, or master-config changes are included — it is purely a changelog-triggered sweep re-run, consistent with the PR description (split from #2189).

Security risks

None. The diff only adds a YAML changelog entry; no code, credentials, or CI configuration is touched.

Level of scrutiny

Low. This follows the established changelog-only re-run pattern used elsewhere in this file, and the referenced config keys (dsv4-fp4-gb300-dynamo-trt, dsv4-fp4-gb300-dynamo-trt-mtp) already exist in configs/nvidia-master.yaml and were previously introduced by an earlier changelog entry, so this is not introducing new, unvalidated recipes.

Other factors

The new entry matches the format of all preceding entries in the file (config-keys/description/pr-link), and the author notes validate_perf_changelog.py --base-ref main passes. No outstanding reviewer comments to address.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: No status

Development

Successfully merging this pull request may close these issues.

1 participant