MULTI-GPU DGEMM AND HIGH PERFORMANCE LINPACK ON HIGHLY ENERGY-EFFICIENT CLUSTERS

被引：4

作者：

Rohr, David ^{[1
]}

Bach, Matthias ^{[1
]}

Kretz, Matthias ^{[1
]}

Lindenstruth, Volker ^{[1
]}

机构：

[1] Goethe Univ Frankfurt, Frankfurt Inst Adv Studies, D-60438 Frankfurt, Germany

来源：

IEEE MICRO | 2011年 / 31卷 / 05期

关键词：

DGEMM; double-precision general matrix multiply; GPGPU; Green IT; heterogeneous (hybrid) systems; High Performance Linpack; HPL; multi-GPU; system architecture;

D O I：

10.1109/MM.2011.66

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

High Performance Linpack can maximize requirements throughout a computer system. An efficient multi-GPU double-precision general matrix multiply (DGEMM), together with adjustments to the HPL, is required to utilize a heterogeneous computer to its full extent. The authors present the resulting energy-efficiency measurements and suggest a cluster design that can utilize multiple GPUs. © 2011 IEEE.

引用

页码：18 / 26

页数：9

共 13 条

[1] [Anonymous], 1979, LINPACK Users' Guide
[2] Optimized HPL for AMD GPU and multi-core CPU usage
Bach, Matthias
Kretz, Matthias
Lindenstruth, Volker
Rohr, David
[J]. COMPUTER SCIENCE-RESEARCH AND DEVELOPMENT, 2011, 26 (3-4): : 153 - 164
[3] COPENBARGER A, ILLINOIS WINS GREENE
[4] DONGARRA JJ, 1990, ACM T MATH SOFTWARE, V16, P1, DOI 10.1145/77626.79170
[5] FENG WC, 2010, GREEN500 LIST NOVEMB
[6] FENG WC, 2005, CTWATCH Q, V1
[7] Kamil S, 2008, 2008 IEEE INTERNATIONAL SYMPOSIUM ON PARALLEL & DISTRIBUTED PROCESSING, VOLS 1-8, P870
[8] Accelerating Cosmological Data Analysis with FPGAs
Kindratenko, Volodymyr V.
Brunner, Robert J.
[J]. PROCEEDINGS OF THE 2009 17TH IEEE SYMPOSIUM ON FIELD PROGRAMMABLE CUSTOM COMPUTING MACHINES, 2009, : 11 - 18
[9] MATSUOKA S, 2010, TSUBAME E SCI J, V1, P16
[10] *MELL TECHN, 2010, CONNECTX SINGL DUAL

← 1 2 →