Abstract | ||
---|---|---|
Towards gaining the performance improvement benefited from threaded MPI while supporting MPI standard well, in this paper, we propose a thread-based MPI program accelerator (MPIActor). MPIActor is a transparent middleware to assist general MPI libraries. People can choose to adopt or abandon MPIActor freely in compiling time for any MPI program (Currently only support C code). With the join of MPIActor, in each node, the MPI processes will be mapped as several threads of one process, and the intra-node point-to-point communication and collective communication will have been enhanced by take advantage of thread based mechanism. We have implemented the point-to-point communication module of our design and evaluated it on a real platform. Comparing with MVAPICH2, the experimental results of OSU PINGPONG benchmark show a significant performance improvement from 114% to 321% for transferring messages which size is between 4KB and 2MB. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1109/IWQoS.2010.5542703 | IWQoS |
Keywords | Field | DocType |
mpi program accelerator,mpiactor,threaded mpi,osu pingpong benchmark,mpi accelerator,multi-threading,mvapich2,mpi library,middleware,message passing,point-to-point communication,acceleration,moore s law,high performance computing,parallel programming,multi threading,multicore processing,point to point,point to point communication | Middleware,Multithreading,Computer architecture,Supercomputer,Computer science,Thread (computing),Point-to-point,Multi-core processor,Message passing,Performance improvement | Conference |
Volume | Issue | ISSN |
null | null | 1548-615X |
ISBN | Citations | PageRank |
978-1-4244-5987-2 | 0 | 0.34 |
References | Authors | |
3 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhiqiang Liu | 1 | 12 | 4.68 |
Junqiang Song | 2 | 185 | 26.86 |
Shaoliang Peng | 3 | 176 | 32.05 |