Towards Fully-fledged GPU Multitasking via Proactive Memory Scheduling

arXiv:2512.24637 · cs.OS · Submitted 2025-12-31 · Read on arXiv

cs.OS

Submitted: 2025-12-31

Updated: 2026-11-06

Code: https://github.com/ggmlorg/llama.cpp

Terminology

Sources

Related papers