[k8s] sky local up speed up for GPUs
#3664
Merged
Conversation
This file contains 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
Michaelvll
approved these changes
Jun 18, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This PR speeds up
sky local upon GPU machines to ~2 min 30s, down from ~12 min on current master.We remove the GPU labelling step, which is no longer required after #3493 since we can use nvidia's official labels.
We also remove the image pre-pull steps, since kubernetes (kind) pulling the image directly during launch is faster than
docker pull->kind load, which takes ~2x the time.Tested on T4 and L4 GCP instances with
sky local upandsky launch --gpus T4/L4:1 --cloud kubernetes -- nvidia-smi.