Structured Evaluation of Computer Systems - Computer

Abstract

0018-9162/96/$5 O O Q 1996 IEEE ystem development involves deciding which configuration and characteristics will determine a successful design. Suppose you S must evaluate cache designs for a new computer system. The project manager might say something like “Determine the best cache for our new computer.” You know that you have to evaluate cach.e systems, but what does ‘‘best’’ really mean? High performance? Low hi:at? Low cost? Minimum space? Even if you learn that performance is irhe goal, what technical properties must your evaluation consider? Is it just cache size or cache organization? The replacement and refill strategy? The consistency strategy? Which architectures must you consider? What kind of workload will this computer system have to handle? A cache that offers high performance for computer-aided engineering tasks might perform poorly for text processing. Even if you perform the evaluation satisfactorily, six months later a customer could ask what a different cache parametrization might yield. Will your documentation reveal why you chose a particular range of architectures and workloads? Why you chose a set of criteria for evaluation? Will you be able to repeat the evaluation and produce the same reisults? Can you extend the evaluation to different sets of parameters to back up your results?

Extracted Key Phrases

5 Figures and Tables

Cite this paper

@inproceedings{Bockle2004StructuredEO, title={Structured Evaluation of Computer Systems - Computer}, author={Gunter Bockle and Hermann Hellwagner and Roland Lepold and Gerd Sandweg and Burghardt Schallenberger and Raimar Thudt and Stefan Wallstab}, year={2004} }