Recognition and Optimization of Loop-Carried Stream Reusing of Scientific Computing Applications on the Stream Processor

Authors:
Ying Zhang;Gen Li;Xuejun Yang
Affiliations:
Institute of Computer, National University of Defense Technology, 410073 Changsha, China;Institute of Computer, National University of Defense Technology, 410073 Changsha, China;Institute of Computer, National University of Defense Technology, 410073 Changsha, China
Venue:
ICCS '07 Proceedings of the 7th international conference on Computational Science, Part I: ICCS 2007
Year:
2007

Citing 8
Cited 0

Practical dependence testing

PLDI '91 Proceedings of the ACM SIGPLAN 1991 conference on Programming language design and implementation
A quasi-minimal residual variant of the Bi-CGSTAB algorithm for nonsymmetric systems

SIAM Journal on Scientific Computing
Hitting the memory wall: implications of the obvious

ACM SIGARCH Computer Architecture News
Memory bandwidth limitations of future microprocessors

ISCA '96 Proceedings of the 23rd annual international symposium on Computer architecture
Stream processor architecture

Stream processor architecture
A programming system for the imagine media processor

A programming system for the imagine media processor
Programmable Stream Processors

Computer
Merrimac: Supercomputing with Streams

Proceedings of the 2003 ACM/IEEE conference on Supercomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Compared with other stream applications, scientific stream programs are usually constrained by memory access. Loop-carried stream reusing means reusing streams across different iterations and it can improve the locality of SRF greatly. In the paper, we present algorisms to recognize loop-carried stream reusing and give the steps to utilize the optimization after analyzing characteristics of scientific computing applications. Then we perform several representative microbenchmarks and scientific stream programs with and without our optimization on Isim. Simulation results show that stream programs optimized by loop-carried stream reusing can improve the performance of memory-bound scientific stream programs greatly.