8月12日 杨峻:使GPU成为一流计算资源:共享和资源管理

时间:2019-08-04浏览:60设置


讲座题目:使GPU成为一流计算资源:共享和资源管理

主讲人:杨峻  教授

开始时间:2019-08-12 09:30:00  结束时间:2019-08-12 10:30:00

讲座地址:华东师范大学数学馆201

主办单位:计算机科学与技术学院

  

报告人简介:

       杨峻,匹兹堡大学电子与计算机工程系教授。研究方向为计算机体系结构,研究方向包括GPU设计、新兴的内存技术、互连网络、3D集成以及电源和功耗管理技术。获得了2008年的NSF CAREER award2010年的IEEE Micro   Top Picks奖、2013ISLPED2007ICCD最佳论文奖。被列入了HPCA名人堂。


报告内容:

GPUs have evolved from traditional graphics   accelerators into core compute engines for a broad class of general-purpose   applications. However, current commercial offerings fall short of the great   potential of GPUs largely because they cannot be managed as easily as the   CPU. The enormous amount of hardware resources are often greatly   underutilized. We developed new architecture features to enable fine-grained   sharing of GPUs, termed Simultaneous Multi-kernel (SMK), in a similar way the   CPU achieves sharing via simultaneous multithreading (SMT). With SMK,   different applications can co-exist in every streaming multiprocessor of a   GPU, in a fully controlled way. High resource utilization can be achieved by   exploiting heterogeneity of different application behaviors. Resource   apportion among sharers are developed for fairness, throughput, and   quality-of-services. We also envision that SMK can enable better   manageability of GPUs and new features such as more efficient synchronization   mechanisms within an application.

  


返回原图
/