AI · Technology · 2026-10-19 04:57
This paper proposes a novel perspective using API-level memory models to optimize multi-GPU communication efficiency, aiming to improve portability and performance in distributed computing systems.
Auto-generated summary for reference only. RSS