微信内可能无法直接打开本站。请点右上角 ··· → 在浏览器打开，或复制链接。
Cycle-Level Simulator for Distributed GPUs For AI Workloads (Purdue)
RSS 官方收录 · 可信分层展示
关键摘要
普渡大学发布GPU周期级模拟器，支持Ampere至Blackwell架构，H100验证达99%相关性
- 首款支持Ampere/Hopper/Blackwell三代GPU的周期级模拟框架
- 在H100物理芯片上验证，Pearson相关系数达99%
- 面向异步分布式GPU架构的AI工作负载仿真工具
AI 摘要 · 来源可核验
正文提要
Researchers from Purdue University published a technical paper titled “Architecting the Next Generation of Asynchronous, Distributed GPUs for the AI Era.”
Abstract Excerpt: The paper presents a “cycle-level simulation framework” for modern GPU generations including Ampere, Hopper, and Blackwell. It also reports validation against physical silicon, including a “99% Pearson correlation coefficient” on H100 GPU results.
Find the technical paper here. August 2026.
Pan, Junrui, Weili An, Cesar Avalos Baddouh, Ni Kang, Christin David Bose, Aaron Barnes, Ahmad Alawneh, Fangjia Shen, Yechen Liu, Anusuya Nallathambi, Atthin Chandrashekar, and Timothy G. Rogers. “Architecting the Next Generation of Asynchronous, Distributed GPUs for the AI Era.” arXiv, August 2026. https://doi.org/10.48550/arXiv.2608.22602.
The post Cycle-Level Simulator for Distributed GPUs For AI Workloads (Purdue) appeared first on Semiconductor Engineering.