DACOS: Dependency-Aware Cross-Kernel Overlapping for Optimizing Short-Sequence Workloads in LLM Applications.
Zhaoyang Hao, Guangli Li, Fan Luo, Hao Qian, Xueying Wang, Jiacheng Zhao, Xiaobing Feng, Huimin Cui, Jingling Xue
Browse the full EuroPar paper archive.