Acceleration for MPI Derived Datatypes Using an Enhancer of Memory and Network
IWIA '08 Proceedings of the 2008 International Workshop on Innovative Architecture for Future Generation High-Performance Processors and Systems（2010）
摘要
This paper presents a support function for MPI derived datatypes on DIMMnet-3 network interface with multi-banked extended memory, which is under development. Semi-hardwired derived datatype communication based on RDMA with hardwired gather and scatter is proposed. This mechanism and MPI using it are implemented and validated on DIMMnet-2 which is a former prototype operating on DDR DIMM slot. The performance of gather or scatter transfer of 8byte elements with large interval by using vector commands of DIMMnet-2 is 6.8 compared with software on a host. Proprietary benchmark of MPI derived datatype communication for transferring a submatrix corresponding to a narrow HALO area is executed. Observed bandwidth on DIMMnet-2 is far higher than that for similar condition with VAPI based MPI implementation on Infniband, even though poorer CPU and motherboard are used.
更多查看译文
关键词
network interface,DIMMnet-3 network interface,RDMA,halo,application program interfaces,DDR DIMM slot,HALO,parallel programming,network interfaces,DIMMnet-2,memory enhancer,vector access functions,network enhancer,hardwired scatter,semihardwired derived datatype communication,vapi based mpi implementation,mpi derived datatype,scatter/gather,MPI derived datatypes,8 byte element,derived datatype,VAPI,memory architecture,ddr dimm slot,MPI,dimmnet-3,sgi altix uv,message passing,mpi,vector scatter-gather function,infniband,datatype communication,multibanked extended memory
AI 理解论文
溯源树
样例
生成溯源树，研究论文发展脉络
Chat Paper
正在生成论文摘要