Abstract | ||
---|---|---|
The present paper describes an efficient communication optimization technique for Householder tridiagonalization called CAHTR and evaluates its parallel performance. CAHTR is intended to reduce the number of problems in collective communication, especially MPI Allreduce operations. We demonstrate the optimal version of CAHTR(3) compared with a naive implementation CAHTR(0). The CAHTR algorithms are evaluated on the K supercomputer system, and speed-up exceeds x1.4 for the case of N = 5000 and P = 1024. |
Year | DOI | Venue |
---|---|---|
2015 | 10.3233/978-1-61499-621-7-381 | PARALLEL COMPUTING: ON THE ROAD TO EXASCALE |
Keywords | Field | DocType |
communication avoidance, massively parallel eigenvalue solver, Householder tridiagonalization, MPI_Allreduce, parallel DSYMV | Computer science,Parallel computing | Conference |
Volume | ISSN | Citations |
27 | 0927-5452 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Toshiyuki Imamura | 1 | 95 | 22.21 |
Takeshi Fukaya | 2 | 13 | 7.61 |
Yusuke Hirota | 3 | 0 | 1.01 |
Susumu Yamada | 4 | 36 | 9.54 |
Masahiko Machida | 5 | 34 | 9.76 |