Skip to content

[devcontainer] Fix mount and GPU & Support rust dev - #2978

Merged
zhyncs merged 2 commits into
mainfrom
byhsu/devcontainer
Jan 19, 2025
Merged

zhyncs merged 2 commits into
mainfrom
byhsu/devcontainer

Conversation

@ByronHsu

Copy link
Copy Markdown
Collaborator

Motivation

  1. The current workspaceFolder in dev container is wrong. We should just remove it and use the default setting.

Reason: The local workspace is mounted to /workspace/sglang but vscode will open /sgl-workspace/sglang/, which is actually the pre-installed sglang from the base image instead of the local one/

  1. Install uv for faster dev installation and rust for rust development

  2. Add --gpus=all for the container to access GPUs

Modifications

Checklist

@ByronHsu
ByronHsu requested a review from zhyncs January 19, 2025 08:24
@ByronHsu

Copy link
Copy Markdown
Collaborator Author

cc @zhyncs

@zhyncs

zhyncs commented Jan 19, 2025

Copy link
Copy Markdown
Contributor

@ByronHsu ByronHsu changed the title Fix vscode devcontainer setup [devcontainer] Fix mount and GPU & Support rust dev Jan 19, 2025
@zhyncs
zhyncs merged commit 53cc91e into main Jan 19, 2025
@zhyncs
zhyncs deleted the byhsu/devcontainer branch January 19, 2025 08:34
timethink pushed a commit to timethink/sglang that referenced this pull request Mar 9, 2025
Ankur-singh added a commit to Ankur-singh/sglang that referenced this pull request Oct 8, 2026
Add an "Agentic aggregated + router (B300 NVFP4)" deployment to the
Qwen3.5 selector, generating worker and router commands for the five
variants qualified in InferenceX sgl-project#2978 (TP2, TP2/EP2, TP4 with MTP and
HiCache), and document the container, checkpoint, excluded harness
settings and caveats in section 3.3.

中文:在 Qwen3.5 选择器中新增“Agentic 聚合 + router(B300 NVFP4)”部署,
为 InferenceX sgl-project#2978 已验证的五种配置(TP2、TP2/EP2、TP4,含 MTP 与
HiCache)生成 worker 与 router 命令,并在 3.3 节说明容器、checkpoint、
已排除的测试框架设置及注意事项。

AI disclosure: prepared with Claude Code (Claude Sonnet 5.5); commands were
checked against executed server logs, but not run on B300 hardware.
nvpohanh added a commit to nvpohanh/sglang that referenced this pull request Oct 9, 2026
…fresh GLM-5.2, DeepSeek-V4, DeepSeek-V4.1

Mirror every NVIDIA sglang / dynamo-sglang AgentX lane on SemiAnalysis
InferenceX main (ceb50ba2) into the Agentic Long-Context Deployment
explorer added in sgl-project#42726, with these open InferenceX PRs overlaid: sgl-project#3816
(Qwen3.8 GB300 SGLang), sgl-project#3840 (Qwen3.5 FP8 GB300 Dynamo+SGLang), sgl-project#2978
(Qwen3.5 B300 Dynamo+SGLang), sgl-project#3817 (DSV4 B300 Dynamo+SGLang), sgl-project#3606
(DSV4 B300 SGLang), sgl-project#3701 (DSV4 GB200 Dynamo+SGLang), sgl-project#3782 (DSV4.1 GB300
SGLang), sgl-project#3744 (DSV4.1 B300 SGLang) and sgl-project#3836 (GLM-5.2 FP8 B300 SGLang).
Lanes of PRs without an authorized /use run are shown as in progress.

- Qwen3.5: new "3.3 Agentic Long-Context Deployment" subsection (the page
  keeps its legacy generator); the hand-written 5.2.3 AgentX HiCache recipe
  is removed.
- Qwen3.8-Flash-Next: new "Agentic Long-Context Deployment" section after
  the Playground with the Deployment panel's agenticLink footer. The static
  "2.3 GB300 long-context serving with HiCache and MTP" section added in
  sgl-project#43159 is removed (its InferenceX sgl-project#3816 configuration is now in the
  explorer), and the Deployment panel warning that linked to it now links
  to the explorer section.
- GLM-5.2 / DeepSeek-V4 / DeepSeek-V4.1: regenerated data; commands change
  only for the overlaid lanes and one derived HiCache variant.
- _agentx.jsx: notes for --mamba-scheduler-strategy and
  SGLANG_DISAGGREGATION_NIXL_BACKEND=LIBFABRIC, and wording for derived
  Dynamo variants whose image does not bundle ai-dynamo.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
nvpohanh added a commit to nvpohanh/sglang that referenced this pull request Oct 14, 2026
…ain (GLM-5.2 NVFP4 B300 agg TP4, Dynamo frontend counts)

Regenerate the Agentic Long-Context Deployment data of GLM-5.2, DeepSeek-V4,
DeepSeek-V4.1, Qwen3.5 and Qwen3.8-Flash-Next from InferenceX main (01e3b46a)
with only two open InferenceX PRs overlaid: sgl-project#3953 (GLM-5.2 NVFP4 B300
Dynamo+SGLang aggregated TP4, in progress at its head) and sgl-project#3927 (DeepSeek-V4.1
GB200 SGLang, pinned to its /use run).

- InferenceX sgl-project#3859 and sgl-project#3836 merged; their lanes are now pinned to the same
  runs on main.
- InferenceX sgl-project#3817, sgl-project#2978, sgl-project#3606 and sgl-project#3860 are no longer overlaid: their
  PR-only lanes (DeepSeek-V4 B300 Dynamo+SGLang, Qwen3.5 B300 Dynamo+SGLang
  pareto) are removed, and the lanes they revised go back to their measured
  main versions.
- The four GLM-5.2 GB200 Dynamo+SGLang lanes are pinned to the InferenceX
  sgl-project#3401 run the dashboard has served since 2026-10-07.
- Dynamo frontend replica counts (and nginx on single-node jobs) now match
  the nodes each SemiAnalysis job allocated.

Only the generated data files change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants