Chaos in Computer Performance

by Hugues Berry, Daniel Gracia Pérez, & Olivier Temam arXiv.org e-Print archive, 13 Jun 2005 Modern computer microprocessors are composed of hundreds of millions of transistors that interact through intricate protocols. Their performance during program execution may be highly variable and present aperiodic oscillations. In this paper, we apply current nonlinear time series analysis techniques to the performances of modern microprocessors during the execution of prototypical programs. While variability clearly stems from stochastic variations for several of them, we present pieces of evidence strongly supporting that performance dynamics during the execution of several other programs display low-dimensional deterministic chaos, with sensibility to initial conditions comparable to textbook models. Taken together, these results confirm that program executions on modern microprocessor architectures can be considered as complex systems and would benefit from analysis with modern tools of nonlinear and complexity science. Read more