Porting Optimized GPU Kernels to a Multi-core CPU: Computational Quantum Chemistry Application Example

Author

Dong Ye ; Titov, Andrii ; Kindratenko, V. ; Ufimtsev, I. ; Martinez, Thierry

Author_Institution

Nat. Center for Supercomput. Applic., Univ. of Illinois at Urbana-Champaign, Urbana, IL, USA

fYear

2011

fDate

19-21 July 2011

Firstpage

72

Lastpage

75

Abstract

We investigate techniques for optimizing a multi-core CPU code back ported from a highly optimized GPU kernel. We show that common sub-expression elimination and loop unrolling optimization techniques improve code performance on the GPU, but not on the CPU. On the other hand, register reuse and loop merging are effective on the CPU and in combination they improve performance of the ported code by 16%.

Keywords

computer graphic equipment; coprocessors; merging; multiprocessing systems; optimisation; highly optimized GPU kernel; loop merging; loop unrolling optimization techniques; multicore CPU code; porting; register reuse; sub-expression elimination; Central Processing Unit; Chemistry; Graphics processing unit; Instruction sets; Kernel; Optimization; Registers; GPU; OpenMP; common sub-expression elimination; loop merging; lop unrolling; register reuse;

fLanguage

English

Publisher

ieee

Conference_Titel

Application Accelerators in High-Performance Computing (SAAHPC), 2011 Symposium on

Conference_Location

Knoxville, TN

Print_ISBN

978-1-4577-0635-6

Electronic_ISBN

978-0-7695-4448-9

Type

conf

DOI

10.1109/SAAHPC.2011.8

Filename

6031568