Performance Portable Applications for Hardware Accelerators: Lessons Learned from SPEC ACCEL

被引：6
作者：
Juckeland, Guido [1 ]
Grund, Alexander [1 ]
Nagel, Wolfgang E. [1 ]
机构：
[1] Tech Univ Dresden, Ctr Informat Serv & High Performance Comp ZIH, D-01062 Dresden, Germany
来源：
2015 IEEE 29TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS | 2015年
关键词：
SPEC ACCEL; OpenCL; OpenACC; OpenMP; 4.0; target directive; AMD GPU; NVIDIA GPU; Intel Xeon Phi;
D O I：
10.1109/IPDPSW.2015.26
中图分类号：
TP3 [计算技术、计算机技术];
学科分类号：
0812 ;
摘要：
The popular and diverse hardware accelerator ecosystem makes apples-to-apples comparisons between platforms rather difficult. SPEC ACCEL tries to offer a yardstick to compare different accelerator hardware and software ecosystems. This paper uses this SPEC benchmark to compare an AMD GPU, an NVIDIA GPU and an Intel Xeon Phi with respect to performance and energy consumption. It also provides observations on the performance portability between the different platforms. Since the SPEC ACCEL OpenACC suite can yet not be run on a Xeon Phi, that suite was ported to OpenMP 4.0 target directives to enable a comparison. The challenges and solutions of this porting of 15 applications are described as well.
引用
收藏
页码：689 / 698
页数：10
相关论文
共 19 条
[1]  
Amdahl G. M., 1967, P APR 18 20 1967 SPR, P483, DOI [10.1145/1465482.1465560, DOI 10.1145/1465482.1465560]
[2]  
[Anonymous], HIGH PERFORMANCE PAR
[3]  
[Anonymous], 2020, OPENACCAPPLIC PROGR
[4]  
[Anonymous], 2013, Tech. Rep.
[5]  
[Anonymous], 2010, P INT S WORKL CHAR
[6]  
[Anonymous], 2012, INT WORKSHOP OPENMP
[7]  
Bailey D., 1994, NAS PARALLEL BENCHMA
[8]  
Che SA, 2009, I S WORKL CHAR PROC, P44, DOI 10.1109/IISWC.2009.5306797
[9]   From CUDA to OpenCL: Towards a performance-portable solution for multi-platform GPU programming [J].
Du, Peng ;
Weber, Rick ;
Luszczek, Piotr ;
Tomov, Stanimire ;
Peterson, Gregory ;
Dongarra, Jack .
PARALLEL COMPUTING, 2012, 38 (08) :391-407
[10]  
Jaaskelainen P, 2014, INT J PARALLEL PROG, P1
12