Single-node DSR1 FP8 sglang smoke on CoreWeave B300; prune matrix to dsr1-b300-cw + kimi multinode

[JordanNanos]

What

Wires the CoreWeave B300 cluster (runner cluster:b300-cw) into the single-node benchmark pipeline and reduces the benchmark matrix to a focused test surface.

Added

• runners/launch_b300-cw.sh — single-node launcher for the CoreWeave B300 fleet. Cloned from launch_b200-cw.sh (the proven CoreWeave template): partition b300, gpu:b300 gres, node-local /tmp squash import under flock (avoids root-squash NFS enroot failures).
• configs/runners.yaml — hardware: cluster:b300-cw (8 GPUs/node).
• configs/nvidia-master.yaml — dsr1-fp8-b300-cw-sglang (DeepSeek-R1-0528 FP8, sglang, runner: cluster:b300-cw), cloned from dsr1-fp8-b300-sglang with the runner label repinned to the CoreWeave fleet.
• perf-changelog.yaml — appended the dsr1-fp8-b300-cw-sglang entry to trigger the sweep.

Removed

Every other single-node benchmark and every non-kimi multinode benchmark — configs + orphaned scripts/recipes — across both master configs.

Survivors:

• Single-node: dsr1-fp8-b300-cw-sglang
• Multinode: all kimik2.5 configs — nvidia dynamo-trt/dynamo-vllm (gb200/gb300) + amd mi355x-vllm-disagg

Notes

• perf-changelog.yaml is append-only and existing entries are immutable (enforced by validate_perf_changelog.py / process_changelog.py), so prior entries are left intact rather than pruned — deleting them would fail the check-changelog gate. Only-added-entries are validated against master, so the removed configs' historical entries are harmless.
• Validated locally: generate_sweep_configs.py full-sweep succeeds on both pruned masters and on the new config; validate_perf_changelog.py accepts the diff.

Test

Apply sweep-enabled to run the single-node DSR1 FP8 sglang smoke on cluster:b300-cw (trimmed to the lowest concurrency per parallelism config).

🤖 Generated with Claude Code

[github-actions]

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

[github-actions]

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

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=28746749170
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=28746749170