会议11_针对 InfiniBand 上的高性能 MPI 库优化的全对全连接建立.pdf

《会议11_针对 InfiniBand 上的高性能 MPI 库优化的全对全连接建立.pdf》由会员分享，可在线阅读，更多相关《会议11_针对 InfiniBand 上的高性能 MPI 库优化的全对全连接建立.pdf（22页珍藏版）》请在三个皮匠报告上搜索。

1、OPTIMIZED ALL-TO-ALL CONNECTION ESTABLISHMENT FOR HIGH-PERFORMANCE MPI LIBRARIES OVER INFINIBAND2024 OFA Virtual WorkshopShulei Xu,Mustafa Abduljabbar,Dhabaleswar K.(DK)PandaNetwork Based Computing LaboratoryThe Ohio State Universityhttp:/nowlab.cse.ohio-state.edu/Follow us onhttps:/ 2024 Virtual OF

2、A Workshop2Network Based Computing Laboratory Introduction&Background Motivation&Challenges Design&Optimization Performance Evaluation ConclusionPresentation Outline2024 Virtual OFA Workshop3Network Based Computing LaboratoryWhat is Featured in Modern HPC Cluster Architectures?Multi-core/many-core t

3、echnologiesRemote Direct Memory Access(RDMA)-enabled networking(InfiniBand and RoCE)Solid State Drives(SSDs),Non-Volatile Random-Access Memory(NVRAM),NVMe-SSDAccelerators(NVIDIA GPGPUs and AMD GPUs)Accelerators/Coprocessors high compute density,high performance/watt1 TFlop DP on a chip High Performa

4、nce Interconnects-InfiniBandMulti-core ProcessorsSSD,NVMe-SSD,NVRAMK-ComputerSunway TaihuLightSummitSierra2024 Virtual OFA Workshop4Network Based Computing LaboratoryOverview of the MVAPICH ProjectHigh Performance open-source MPI Library Support for multiple interconnectsInfiniBand,Omni-Path,Etherne

5、t/iWARP,RDMA over Converged Ethernet(RoCE),AWS EFA,OPX,Broadcom RoCE,Intel Ethernet,Rockport Networks,Slingshot 10/11Support for multiple platformsx86,OpenPOWER,ARM,Xeon-Phi,GPGPUs(NVIDIA and AMD)Started in 2001,first open-source version demonstrated at SC 02Supports the latest MPI-3.1 standardhttp:

6、/mvapich.cse.ohio-state.edu Additional optimized versions for different systems/environments:MVAPICH2-X(Advanced MPI+PGAS),since 2011MVAPICH2-GDR with support for NVIDIA(since 2014)and AMD(since 2020)GPUsMVAPICH2-MIC with support for Intel Xeon-Phi,since 2014MVAPICH2-Virt with virtualization support