fix: use multi-GPU RWKV strategy from visible CUDA devices#3882
Open
Chessing234 wants to merge 1 commit into
Open
fix: use multi-GPU RWKV strategy from visible CUDA devices#3882Chessing234 wants to merge 1 commit into
Chessing234 wants to merge 1 commit into
Conversation
Use multi-GPU RWKV strategy when more than one GPU is visible via --gpus / --num-gpus instead of hardcoding single-device cuda fp16. Co-authored-by: Cursor <cursoragent@cursor.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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #1248
Bug
Launching an RWKV model worker with
--num-gpus/--gpusstill loads withstrategy="cuda fp16", so RWKV ignores the requested GPU list and runs on a single device.Root cause
RwkvModel.__init__hardcodesstrategy="cuda fp16"instead of building a per-device strategy string for multiple visible GPUs.Why this fix is correct
When more than one CUDA device is visible (including after
CUDA_VISIBLE_DEVICESis set by the worker), buildcuda:0 fp16 -> cuda:1 fp16 -> ...as RWKV expects. Single-GPU behavior stayscuda fp16.Made with Cursor