Compiler Techniques for Efficient Communications in Circuit Switched Networks for Multiprocessor Systems

被引:5
作者
Shao, Shuyi [1 ]
Jones, Alex K. [2 ]
Melhem, Rami [1 ]
机构
[1] Univ Pittsburgh, Dept Comp Sci, Pittsburgh, PA 15260 USA
[2] Univ Pittsburgh, Dept Elect & Comp Engn, Pittsburgh, PA 15261 USA
基金
美国国家科学基金会;
关键词
Compiled communication; high-performance computing; multiprocessor systems; MPI; circuit-switching networks; communication patterns; INTERCONNECT; DESIGN;
D O I
10.1109/TPDS.2008.82
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
In this paper, we explore compiler techniques for achieving efficient communications on circuit switching interconnection networks. We propose a compilation framework for identifying communication patterns and compiling these patterns as network configuration directives. This has the potential of providing significant performance benefits when connections can be established in the network prior to the actual communications. The framework includes a flexible and powerful communication pattern representation scheme that captures the property of communication patterns and allows manipulation of these patterns. In this way, communication phases can be identified within the application. Additionally, we extend the classification of static and dynamic communications to include persistent communications. Persistent communications are a subclass of dynamic communications that remain unchanged for large segments of the application execution. An experimental compiler has been developed to implement the framework. This compiler is capable of detecting both static and persistent communications within an application. We show that for the NAS Parallel Benchmarks, 100 percent of the point-to-point communications can be classified as either static or persistent, and 100 percent of the collectives are either static or persistent with the exception of IS. Simulation-based performance analysis demonstrates the benefit of using our compiler techniques for achieving efficient communications in multiprocessor systems.
引用
收藏
页码:331 / 345
页数:15
相关论文
共 44 条
[1]  
ADIGA NR, 2002, P SUP SC
[2]  
[Anonymous], MPI MESS PASS INT ST
[3]  
[Anonymous], P 28 EPS C CONTR F A
[4]  
Bailey DavidH., 1995, NAS PARALLEL BENCHMA
[5]  
BALAJI P, 2007, P SUP SC
[6]  
BARKER KJ, 2005, P SUP SC
[7]   ULTRASTRUCTURE OF THE SKIN OF HUMAN ALBINOS [J].
BROODBAKKER, JTW ;
WESTERHOF, W ;
VANDORP, DB .
OPHTHALMIC PAEDIATRICS AND GENETICS, 1983, 2 (02) :95-107
[8]  
Cappello F., 1995, Proceedings. First IEEE Symposium on High-Performance Computer Architecture, P44, DOI 10.1109/HPCA.1995.386556
[9]   Partitioned optical passive star (POPS) multiprocessor interconnection networks with distributed control [J].
Chiarulli, DM ;
Levitan, SP ;
Melhem, RP ;
Teza, JP ;
Gravenstreter, G .
JOURNAL OF LIGHTWAVE TECHNOLOGY, 1996, 14 (07) :1601-1612
[10]   Design and nonlinear servo control of MEMS mirrors and their performance in a large port-count optical switch [J].
Chu, PB ;
Brener, I ;
Pu, C ;
Lee, SS ;
Dadap, JI ;
Park, S ;
Bergman, K ;
Bonadeo, NH ;
Chau, T ;
Chou, M ;
Doran, RA ;
Gibson, R ;
Harel, R ;
Johnson, JJ ;
Lee, CD ;
Peale, DR ;
Tang, B ;
Tong, DTK ;
Tsai, MJ ;
Wu, Q ;
Zhong, W ;
Goldstein, EL ;
Lin, LY ;
Walker, JA .
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, 2005, 14 (02) :261-273