Skip to content

[Klaud Cold] Update dsv4-fp4-b300-vllm vLLM image to v0.21.0 #1456

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

Merged
functionstackx merged 3 commits into from
May 18, 2026
Merged

[Klaud Cold] Update dsv4-fp4-b300-vllm vLLM image to v0.21.0 #1456

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
c08dcef
Update dsv4-fp4-b300-vllm vLLM image to v0.21.0
functionstackx May 17, 2026
b75ad8c
chore: fill pr-link for #1456
functionstackx May 17, 2026
6cee4a0
Merge branch 'main' into update-dsv4-fp4-b300-vllm-v0.21.0
functionstackx May 18, 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 .github/configs/nvidia-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2964,7 +2964,7 @@ dsv4-fp8-h200-sglang-mtp:
# field, so dp-attn=true is used as the existing vLLM script switch for DP4
# layouts on 4 allocated GPUs.
dsv4-fp4-b300-vllm:
image: vllm/vllm-openai:v0.20.0-cu130
image: vllm/vllm-openai:v0.21.0
model: deepseek-ai/DeepSeek-V4-Pro
model-prefix: dsv4
runner: b300
Expand Down
6 changes: 6 additions & 0 deletions perf-changelog.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2802,3 +2802,9 @@
description:
- "Update vLLM image from v0.20.0-cu130 (14d old) to v0.21.0"
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1457

- config-keys:
- dsv4-fp4-b300-vllm
description:
- "Update vLLM image from v0.20.0-cu130 (18d old) to v0.21.0"
Copy link
Copy Markdown
Contributor

@claude claude Bot May 17, 2026

Choose a reason for hiding this comment

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

🟡 The new perf-changelog entry contains the parenthetical "(18d old)" describing the prior image tag, which is anchored to the PR creation date and becomes meaningless when read later. All other entries in the file use the plain "Update X image from vA to vB" form — recommend dropping the "(18d old)" so this entry matches established style and remains useful as a permanent record.

Extended reasoning...

What's the issue

At perf-changelog.yaml:2636, the new entry reads:

- config-keys:
    - dsv4-fp4-b300-vllm
  description:
    - "Update vLLM image from v0.20.0-cu130 (18d old) to v0.21.0"
  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1456

The (18d old) parenthetical is an ephemeral relative-time annotation — it was true on the PR creation date (2026-05-17) but degrades immediately afterward. A reader looking at the changelog months from now has no implicit anchor for what "18d old" was measured from.

Why this conflicts with established style

Grepping for relative-time annotations across perf-changelog.yaml turns up exactly one match — the new entry from this PR. Every other entry, including the most recent neighbors, uses the bare form:

  • Line 2624: "Update vLLM image from v0.15.1 to v0.20.2"
  • Line 2630: "Update vLLM ROCm image from v0.18.0 to v0.21.0"

None of them include a freshness hint for the prior tag. The perf-changelog is a permanent record of recipe changes; ephemeral commentary anchored to a moving "now" doesn't fit that purpose.

Addressing the refutation

One verifier argued the absolute date can always be recovered from the pr-link field, so the staleness framing is overblown. That's true — a reader can click through to GitHub to recover the date. But that's not really the point: the entry is inconsistent with every other entry in the file, and "18d old" provides no information that the rest of the description (specific version strings) doesn't already convey. The cost to drop it is zero, and the result matches the file's established convention.

Impact

None functionally — this is purely a documentation/style issue. The changelog is human-readable metadata; it doesn't affect config parsing, image resolution, or sweep behavior. Flagging as nit accordingly.

Step-by-step proof

  1. Open perf-changelog.yaml at line 2636 — see "Update vLLM image from v0.20.0-cu130 (18d old) to v0.21.0".
  2. Look at the immediately preceding entry at line 2630 — "Update vLLM ROCm image from v0.18.0 to v0.21.0" (no relative-time hint).
  3. Look at line 2624 — "Update vLLM image from v0.15.1 to v0.20.2" (no relative-time hint).
  4. Grep the entire file for [0-9]+d (old|ago) — only the new entry matches.
  5. Imagine reading this changelog in 2027: "18d old" relative to what? Without an explicit anchor, the reader must check the linked PR's creation date — at which point the annotation conveys nothing the version strings don't already say.

Suggested fix

Drop the parenthetical so the entry reads:

- "Update vLLM image from v0.20.0-cu130 to v0.21.0"

pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1456
Loading