Book
Open access
Sep 2026
ReliefServe: Relieving GPU Pressure in Multi-Model Serving via Selective CPU Escape
The key is that migrating CPU-tolerant models to CPU during bursts can reduce overall GPU interference more than the local slowdown incurred, thus protecting performance-critical GPU models even with slower CPU execution.
Shi-Jie Peng, Yan-Ying Lin, Cheng-Zhi Lu et al.
· International Conference on... · 0 citations