Micro-benchmarks for cluster OpenMP implementations: memory consistency costs

Authors:
H. J. Wong;J. Cai;A. P. Rendell;P. Strazdins
Affiliations:
Australian National University, Canberra, ACT, Australia;Australian National University, Canberra, ACT, Australia;Australian National University, Canberra, ACT, Australia;Australian National University, Canberra, ACT, Australia
Venue:
IWOMP'08 Proceedings of the 4th international conference on OpenMP in a new era of parallelism
Year:
2008

Citing 11
Cited 1

A microbenchmark suite for OpenMP 2.0

ACM SIGARCH Computer Architecture News - Special Issue: PACT 2001 workshops
Shared Memory Consistency Models: A Tutorial

Computer
A Fully Compliant OpenMP Implementationon Software Distributed Shared Memory

HiPC '02 Proceedings of the 9th International Conference on High Performance Computing
ParADE: An OpenMP Programming Environment for SMP Cluster Systems

Proceedings of the 2003 ACM/IEEE conference on Supercomputing
Towards automatic translation of OpenMP to MPI

Proceedings of the 19th annual international conference on Supercomputing
Towards a more efficient implementation of OpenMP for clusters via translation to global arrays

Parallel Computing - OpenMp
Cluster-enabled OpenMP: An OpenMP compiler for the SCASH software distributed shared memory system

Scientific Programming
TreadMarks: distributed shared memory on standard workstations and operating systems

WTEC'94 Proceedings of the USENIX Winter 1994 Technical Conference on USENIX Winter 1994 Technical Conference
A case for low-complexity MP architectures

Proceedings of the 2007 ACM/IEEE conference on Supercomputing
The design of MPI based distributed shared memory systems to support OpenMP on clusters

CLUSTER '07 Proceedings of the 2007 IEEE International Conference on Cluster Computing
A Shared Memory Benchmark in OpenMP

ISHPC '02 Proceedings of the 4th International Symposium on High Performance Computing

Region-Based Prefetch Techniques for Software Distributed Shared Memory Systems

CCGRID '10 Proceedings of the 2010 10th IEEE/ACM International Conference on Cluster, Cloud and Grid Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

The OpenMP memory model allows for a temporary viewof shared memory that only needs to be made consistent when barrieror flush directives, including those that are implicit, are encountered.While this relaxed memory consistency model is key to developing clusterOpenMP implementations, it means that the memory performance ofany given implementation is greatly affected by which memory is used,when it is used, and by which threads. In this work we propose a microbenchmarkthat can be used to measure memory consistency costs andpresent results for its application to two contrasting cluster OpenMPimplementations, as well as comparing these results with data obtainedfrom a hardware supported OpenMP environment.