Evaluating Parallel Program and Architecture Scalability
Download:
by Xiaodong Zhang, Yong Yan, Keqiang Hej
http://www.cs.wm.edu/~zhang/../hpcs/WWW/HTML/publications/./papers/TR-94-02-01.pdf
Add To MetaCart
Abstract:
Latency measures the delay caused by communication between processors and memory modules over the network in a parallel system. Using intensive measurements and simulation, we show that network latency forms a major obstacle to improving parallel computing performance and scalability. We present an experimental metric, using network latency to measure and evaluate the scalability of parallel programs and architectures. This latency metric is an extension to the isoefficiency function lGrama et al.,
Citations
| 1206 | Introduction to Parallel Algorithms and Architectures: Arrays – Leighton - 1992 |
| 705 | SPLASH: Stanford Parallel Applications for Shared Memory – Singh, Weber, et al. - 1992 |
| 90 | Cache Invalidation Patterns in Shared-Memory Multiprocessors – Gupta, Weber - 1992 |
| 66 | Directory-based cache coherence in large-scale multiprocessors – Chaiken, Fields, et al. - 1990 |
| 50 | Scalability of parallel algorithm-machine combinations – SUN, ROVER - 1994 |
| 21 | Matrix computations on distributed memory multiprocessors", in [17 – Moler - 1989 |
| 8 | Isoefficiency Function: A Scalability Metric for Parallel Algorithms and Architectures – Grama, Gupta, et al. - 1993 |
| 4 | Execution behavior analysis and performance improvement in shared-memory architectures – Zhang, He, et al. - 1993 |
| 1 | The Cerberus user manual – Brooks, II1, et al. - 1991 |
| 1 | is currently employed as a system analyst at Cartotech – He |
