Refactor(GPU): switch to per-stream synchronization in CUDA and ROCm backend#699

Closed

victorcamaraa wants to merge 4 commits intoJuliaParallel:masterfrom

victorcamaraa:fix-per-stream-sync

Commits on Apr 15, 2026

refactor: change CUDA and ROCm backends to per-stream syncronization
victorcamaraa
committed

Commits on Apr 16, 2026