约束(计算机辅助设计)
拓扑(电路)
计算机科学
薄脆饼
网络拓扑
逻辑拓扑
分布式计算
计算机网络
材料科学
工程类
电气工程
纳米技术
机械工程
作者
Qize Yang,Taiquan Wei,Sihan Guan,Chengran Li,Haoran Shang,Jinyi Deng,Huizheng Wang,Chao Li,Lei Wang,Yan Zhang,Shouyi Yin,Yang Hu
出处
期刊:
日期:2025-06-20
卷期号:: 49-64
被引量:20
标识
DOI:10.1145/3695053.3731045
摘要
As cluster scales for LLM training expand, waferscale chips, characterized by the high integration density and bandwidth, emerge as a promising approach to enhancing training performance.The role of Network on Wafer (NoW) is becoming increasingly significant, which puts an emphasis on two facts: physical and logical topology.However, existing networks fail to co-design both aspects.Additionally, physical topology typically focuses on optimizing communication or computation separately, neglecting opportunities to improve overall training performance.In this paper, we propose a physical design (PD) constraint-aware joint optimization strategy, developing mesh-switch physical topology and a dual-granularity logical topology.Mesh-switch leverages the high integration density of mesh and the efficient communication performance of fat tree, optimizing the allocation of on-chip
科研通智能强力驱动
Strongly Powered by AbleSci AI