-
Notifications
You must be signed in to change notification settings - Fork 176
[Handoff to @Oseltamivir Claude /loop] [Klaud Cold] Update dsv4-fp8-h200-sglang (+mtp) SGLang image to v0.5.12-cu130 #1460
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.
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
Open
functionstackx
wants to merge
4
commits into
main
Choose a base branch
from
update-dsv4-fp8-h200-sglang-v0.5.12
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit
Hold shift + click to select a range
5169262
[Klaud Cold] Update dsv4-fp8-h200-sglang (+mtp) SGLang image to v0.5.…
functionstackx 2044102
chore: fill pr-link for #1460
functionstackx 1ff90ec
Merge remote-tracking branch 'origin/main' into update-dsv4-fp8-h200-…
Oseltamivir 6d2b2be
fix: raise dsv4 h200 sglang memory fraction
Oseltamivir File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🟡 The comment at .github/configs/nvidia-master.yaml:2891-2893 justifies the h200-dgxc-slurm runner pinning on the grounds that 'the deepseek-v4-hopper image needs the /ix mount layout that only launch_h200-dgxc-slurm.sh sets up' — but this PR removes the only references to that image, making the rationale stale and misleading. Please either update/remove the comment, or explicitly note the runner pinning is now retained for a different reason (and consider whether it can be relaxed, since the conditional in runners/launch_h200-dgxc-slurm.sh:308 that gates /ix vs /workspace on the deepseek-v4-hopper substring is now dead code).
Extended reasoning...
What the bug is
The PR bumps both
dsv4-fp8-h200-sglanganddsv4-fp8-h200-sglang-mtpfromlmsysorg/sglang:deepseek-v4-hopper@sha256:7f19c6dc...to the genericlmsysorg/sglang:v0.5.12-cu130. Sitting just above the first updated entry, at.github/configs/nvidia-master.yaml:2891-2893, is an explanatory comment:After this PR, no config in the repo references a
deepseek-v4-hopperimage, so the documented rationale no longer applies. The comment now describes a constraint that does not exist in the current state of the repo.Why this matters
A future maintainer reading this block will be told the runner pin exists because of a /ix-mount requirement tied to the deepseek-v4-hopper image. That is no longer true. They may either:
runner: h200-dgxcpin because they cannot trust the now-incorrect rationale.The companion runtime code at
runners/launch_h200-dgxc-slurm.sh:308(if [[ "$IMAGE" == *deepseek-v4-hopper* ]]; then CONTAINER_MOUNT_DIR=/ix; else CONTAINER_MOUNT_DIR=/workspace; fi) is also affected: after this PR the*deepseek-v4-hopper*substring will never match any image in the configs, so the /ix branch is dead. These two recipes will now be mounted at/workspacerather than/ixat runtime.Step-by-step proof
runner: h200-dgxc, referencing thedeepseek-v4-hopperimage's /ix-mount requirement.image:fields fromlmsysorg/sglang:deepseek-v4-hopper@sha256:7f19c6dc...tolmsysorg/sglang:v0.5.12-cu130.deepseek-v4-hopperin.github/configs/nvidia-master.yamlfinds only line 2892 — the comment text itself. No config still uses that image.launch_h200-dgxc-slurm.sh:308doesIMAGE=$IMAGE; if [[ "$IMAGE" == *deepseek-v4-hopper* ]]; then CONTAINER_MOUNT_DIR=/ix; else CONTAINER_MOUNT_DIR=/workspace; fi. WithIMAGE=lmsysorg/sglang:v0.5.12-cu130, the substring match fails and the else branch runs, so the mount point is now/workspacerather than/ix. The documented /ix behavior no longer occurs for these recipes.Impact
No functional impact on this sweep — the new generic image works fine with the
/workspacemount the else branch provides. This is a documentation-rot / dead-code issue introduced by this PR.How to fix
Pick one:
runner: h200-dgxccan be relaxed to a less restrictive pool now that the /ix constraint is gone).*deepseek-v4-hopper*branch inrunners/launch_h200-dgxc-slurm.sh:308as a drive-by cleanup.