首页|期刊导航|计算机科学技术学报（英文版）|VastPipe:A High-Throughput Inference System via Adaptive Space-Division Multiplexing for Diverse Accelerators
计算机科学技术学报（英文版）2025，Vol.40Issue(2)：444-463,20.DOI:10.1007/s11390-024-3773-5
VastPipe:A High-Throughput Inference System via Adaptive Space-Division Multiplexing for Diverse Accelerators
VastPipe:A High-Throughput Inference System via Adaptive Space-Division Multiplexing for Diverse Accelerators
摘要
关键词
cluster scheduling/resource management/reinforcement learning/DNN acceleratorKey words
cluster scheduling/resource management/reinforcement learning/DNN accelerator引用本文复制引用
马立贤,王乐平,邵恩,曹荣禹,谭光明..VastPipe:A High-Throughput Inference System via Adaptive Space-Division Multiplexing for Diverse Accelerators[J].计算机科学技术学报（英文版）,2025,40(2):444-463,20.基金项目
This work was supported by the National Key Research and Development Program of China under Grant No.2021YFB0300202,the National Natural Science Foundation of China under Grant Nos.62032023,T2125013,and 62102396,the Bei-jing Nova Program under Grant No.Z211100002121143,the Youth Innovation Promotion Association of Chinese Academy of Sci-ences under Grant No.2021099,the Innovation Funding of Institute of Computing Technology,Chinese Academy of Sciences under Grant No.E461030,and Tianjin Science and Technology Plan Project under Grant No.24ZXKJGX00060. （）
