bugfix: fix the synchronization issue in distributed operators

flashinfer-ai / flashinfer

FlashInfer: Kernel Library for LLM Serving

https://flashinfer.ai

Apache License 2.0

768 stars 64 forks source link

Closed yzh119 closed 4 weeks ago

yzh119 commented 4 weeks ago

We missed some synchronizations in all reduce operators, which causes deadlock and incorrect result some times.