feat(nodewright-customizations): add gb200 eks support#699
Merged
Conversation
📝 WalkthroughWalkthroughThis change updates the Nodewright tuning package definitions to reference new package images and versions, upgrading Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes 🚥 Pre-merge checks | ✅ 4✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
recipes/overlays/gb200-eks-training.yaml (1)
67-70:⚠️ Potential issue | 🟡 MinorRefresh obsolete no-op explanation in the training overlay.
Line 67–70 says tuning is unavailable, but Line 74 now consumes
tuning.yaml. Please update the comment so it reflects the new GB200 EKS behavior.🛠️ Proposed comment fix
- # GB200 uses nodewright no-op: the H100 tuning packages (nvidia-setup, - # nvidia-tuned) are not compatible with GB200's ARM64 host CPU and - # Blackwell GPU architecture. Replace with tuning.yaml once GB200-specific - # packages are available. + # GB200 on EKS uses nodewright tuning. + # This overlay points to tuning.yaml to apply GB200-compatible setup/tuning + # packages for training workloads.Also applies to: 74-74
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@recipes/overlays/gb200-eks-training.yaml` around lines 67 - 70, Update the explanatory comment around the GB200 EKS training overlay to reflect that GB200 now consumes tuning.yaml instead of using a nodewright no-op; replace the current lines mentioning "nodewright no-op" and that tuning packages are incompatible with GB200 with a concise note that GB200 uses a GB200-specific tuning.yaml (or placeholder until GB200-specific packages are available) and remove the outdated statement about H100/ARM64 incompatibility, referencing "tuning.yaml" and "GB200" so readers know where the actual tuning is now applied.recipes/overlays/gb200-eks-inference.yaml (1)
61-64:⚠️ Potential issue | 🟡 MinorUpdate stale no-op commentary to match actual tuning configuration.
Line 61–64 still says GB200 uses no-op and tuning is unavailable, but Line 68 now points to
tuning.yaml. Please update this comment to avoid misleading future changes.🛠️ Proposed comment fix
- # GB200 uses nodewright no-op: the H100 tuning packages (nvidia-setup, - # nvidia-tuned) are not compatible with GB200's ARM64 host CPU and - # Blackwell GPU architecture. Replace with tuning.yaml once GB200-specific - # packages are available. + # GB200 on EKS uses nodewright tuning. + # This overlay points to tuning.yaml to apply GB200-compatible setup/tuning + # packages for inference workloads.Also applies to: 68-68
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@recipes/overlays/gb200-eks-inference.yaml` around lines 61 - 64, Update the stale comment in recipes/overlays/gb200-eks-inference.yaml: replace the existing "GB200 uses nodewright no-op..." text with a concise note that GB200 now references tuning.yaml for GB200-specific tuning (instead of claiming tuning is unavailable), mention that H100 packages are incompatible with GB200's ARM64/Blackwell but a GB200-specific tuning.yaml is provided, and update the comment near the reference to tuning.yaml so it accurately describes the current configuration.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Outside diff comments:
In `@recipes/overlays/gb200-eks-inference.yaml`:
- Around line 61-64: Update the stale comment in
recipes/overlays/gb200-eks-inference.yaml: replace the existing "GB200 uses
nodewright no-op..." text with a concise note that GB200 now references
tuning.yaml for GB200-specific tuning (instead of claiming tuning is
unavailable), mention that H100 packages are incompatible with GB200's
ARM64/Blackwell but a GB200-specific tuning.yaml is provided, and update the
comment near the reference to tuning.yaml so it accurately describes the current
configuration.
In `@recipes/overlays/gb200-eks-training.yaml`:
- Around line 67-70: Update the explanatory comment around the GB200 EKS
training overlay to reflect that GB200 now consumes tuning.yaml instead of using
a nodewright no-op; replace the current lines mentioning "nodewright no-op" and
that tuning packages are incompatible with GB200 with a concise note that GB200
uses a GB200-specific tuning.yaml (or placeholder until GB200-specific packages
are available) and remove the outdated statement about H100/ARM64
incompatibility, referencing "tuning.yaml" and "GB200" so readers know where the
actual tuning is now applied.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Enterprise
Run ID: 6ae5d415-69f6-4bac-be1c-5fe8bfd27a96
📒 Files selected for processing (3)
recipes/components/nodewright-customizations/manifests/tuning.yamlrecipes/overlays/gb200-eks-inference.yamlrecipes/overlays/gb200-eks-training.yaml
Contributor
Coverage Report ✅
Coverage BadgeNo Go source files changed in this PR. |
mchmarny
approved these changes
Apr 27, 2026
lockwobr
pushed a commit
that referenced
this pull request
Apr 28, 2026
Co-authored-by: Mark Chmarny <mchmarny@users.noreply.github.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Summary
Update nvidia-tuned and nvidia-setup for bug fixes related to gb200 on eks
Fixes: #656
Type of Change
Component(s) Affected
cmd/aicr,pkg/cli)cmd/aicrd,pkg/api,pkg/server)pkg/recipe)pkg/bundler,pkg/component/*)pkg/collector,pkg/snapshotter)pkg/validator)pkg/errors,pkg/k8s)docs/,examples/)Testing
Ran CUJ1 on 2 node eks gb200 cluster
< Evidence and final testing to come >
Risk Assessment
Checklist
make testwith-race)make lint)git commit -S) — GPG signing info