A Framework for Performance Modeling and Prediction

• Published in: ACM/IEEE SC 2002 Conference (SC'02) p. 21
• Main Authors: Snavely, A., Carrington, L., Wolter, N., Labarta, J., Badia, R., Purkayastha, A.
• Format: Conference Proceeding
• Language: English; Japanese
• Published: IEEE 2002
• Subjects: Application software; Computational modeling; Computer applications; Concurrent computing; Delay; High performance computing; Parallel processing; Predictive models; Supercomputers; Systems engineering and theory
• ISBN: 076951524X; 9780769515243
• ISSN: 1063-9535
• DOI: 10.1109/SC.2002.10004

Cycle-accurate simulation is far too slow for modeling the expected performance of full parallel applications on large HPC systems. And just running an application on a system and observing wallclock time tells you nothing about why the application performs as it does (and is anyway impossible on yet-to-be-built systems). Here we present a framework for performance modeling and prediction that is faster than cycle-accurate simulation, more informative than simple benchmarking, and is shown useful for performance investigations in several dimensions.