A PARALLELISM-BASED ANALYTIC APPROACH TO PERFORMANCE EVALUATION USING APPLICATION PROGRAMS

doi:10.1109/5.236190

A PARALLELISM-BASED ANALYTIC APPROACH TO PERFORMANCE EVALUATION USING APPLICATION PROGRAMS

1993-01-01

5

PRE

AI

DOI：10.1109/5.236190

原文求助

摘要

En 中文

In this paper we present a brief overview of performance evaluation and benchmarking. We demonstrate that traditional performance measurements recorded in these activities are really a direct measurement of the parallelism in software and hardware. A framework called the Path to Performance is developed that identifies the agents and activities that change the parallelism as it moves from problem to solution. We show where various current application benchmarks apply probes on the Path and which agents am being measured. We recast the traditional time-based performance measurements into parallelism-based performance measurements to show that understanding performance implies understanding the parallelism. A simulation tool and methodology are described for measuring and comparing the executed parallelism on a single CRAY Y-MP CPU. We apply our methodology to several of the Perfect Benchmarks to quantify their executed parallelism on this machine. Our results suggest that since some of the benchmarks have a similar mix of different levels of parallelism on the Y-MP, the benchmarking utility of these programs on the Y-MP may be questioned because they exercise the machine in the same way.

Keyword:

PERFORMANCE EVALUATION

BENCHMARKING

CRAY Y-MP

INSTRUCTION-LEVEL PARALLELISM

PARALLELISM MATRIX

SCIENTIFIC METHOD

PATH TO PERFORMANCE

A PARALLELISM-BASED ANALYTIC APPROACH TO PERFORMANCE EVALUATION USING APPLICATION PROGRAMS

摘要

AI总结

期刊

机构

引用论文