Repository navigation
fix(apps): pin requested gpu devices instead of their whole pool - #624
Draft
justinwlin wants to merge 1 commit into
Draft
justinwlin wants to merge 1 commit into
justinwlin wants to merge 1 commit into
Conversation
Endpoints select GPUs by pool, so a device list like ["NVIDIA GeForce RTX 4090", "NVIDIA L4", "NVIDIA RTX A5000"] deployed as ADA_24,AMPERE_24 and workers could land on any device in those pools, including an RTX PRO 6000 MIG 1g.24gb slice the platform added to AMPERE_24. Exclude every other device in each pool selected through a device name with "-<device id>" tokens, as runpod-flash did. Membership comes from the platform's live GPU catalog, because the static pool table lags it (it has no MIG slices); the static table is the fallback when the catalog is unreachable. Pools named directly still allow all devices.
This branch has not been deployed
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
gpu=[...]with device names deploys onto whole pools, so workers can run on GPUs you didn't ask for. In a live test,["NVIDIA GeForce RTX 4090", "NVIDIA L4", "NVIDIA RTX A5000"]ran on anRTX PRO 6000 Blackwell Server Edition MIG 1g.24gbslice.Cause
gpu_ids_valuemaps each device to its pool and sends only the pool ids. The platform'sAMPERE_24pool also holds the RTX 3090 and the PRO 6000 MIG 24 GB slice. runpod-flash excluded the pool's other devices with-<device id>tokens, but only the ones in its static pool table. That table lacks the MIG slices, so Flash would land there too.Change
-<device id>. Pools named directly ("AMPERE_24") still allow all their devices.GET https://api.runpod.io/v2/catalog/gpus, thepoolfield), cached per process. Fall back to the static table when the catalog is unreachable.Verification
gpuIds = ADA_24,AMPERE_24,-NVIDIA GeForce RTX 3090,-NVIDIA RTX PRO 6000 Blackwell Server Edition MIG 1g.24gb. Four parallel jobs all reportedNVIDIA GeForce RTX 4090.tests/test_apps/conftest.pyfixture keeps other tests offline.pytest tests: 1483 passed, coverage 91.67%.