vllm icon indicating copy to clipboard operation
vllm copied to clipboard

Published 20 hours ago • verified publisher icon vllm-project

[Bug]: prefill/prefix FP8 triton kernel for opt-125m - an illegal memory access was encountered

Open jon-chuang opened this issue 1 year ago • 0 comments

Your current environment

As of merging https://github.com/vllm-project/vllm/pull/7208

🐛 Describe the bug

Illegal memory access for facebook/opt-125m

Specifically one of these errors:

RuntimeError: Triton Error [CUDA]: an illegal memory access was encountered
RuntimeError: CUDA error: an illegal memory access was encountered

jon-chuang avatar Aug 09 '24 20:08 jon-chuang