检索规则说明:AND代表“并且”;OR代表“或者”;NOT代表“不包含”;(注意必须大写,运算符两边需空一格)
检 索 范 例 :范例一: (K=图书馆学 OR K=情报学) AND A=范并思 范例二:J=计算机应用与软件 AND (U=C++ OR U=Basic) NOT M=Visual
作 者:徐德发[1] Xu Defa(Shanghai Supercomputer Center,Shanghai 201023,China)
机构地区:[1]上海超级计算中心,上海201023
出 处:《无线互联科技》2024年第9期111-116,共6页Wireless Internet Technology
摘 要:高性能计算(High Performance Computing,HPC)系统的通信接口一直在不断发展,然而对新一代的无限带宽(InfiniBand,IB)互联与RoCE互联在性能上的对比缺乏研究。文章从消息传递接口(Message Passing Interface,MPI)基准程序、HPC基准程序以及HPC应用程序等多个方面,提出了一套实验方法对IB互联和RoCE互联进行性能评估。结果表明,在拥塞条件下RoCE相较于IB有更好的网络性能;在MPI场景下,RoCE网络的通信性能略高于IB网络;在HPC基准程序测试中,RoCE方案相较于IB方案浮点性能略高但综合性能相当;在不同的终端应用场景下,RoCE方案和IB方案的性能则各有高低。The communication interface of high performance computing(HPC)systems is constantly evolving,but there is a lack of research on the performance comparison between the new generation of IB interconnect and RoCE interconnect.This paper presents a set of experimental methods for performance evaluation from MPI benchmark,HPC benchmark and HPC application.The conclusion shows that RoCE has better network performance than IB under congestion conditions,and the communication performance of RoCE network in MPI scenario is slightly higher than that of IB network.In HPC benchmark test,RoCE scheme has slightly higher floating point performance than IB scheme,but the overall performance is comparable.The performance of the RoCE and IB schemes varies according to the application scenarios.
关 键 词:RoCE INFINIBAND HPC 性能
分 类 号:TN711.1[电子电信—电路与系统]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在链接到云南高校图书馆文献保障联盟下载...
云南高校图书馆联盟文献共享服务平台 版权所有©
您的IP:216.73.216.49