Computer Science 365 - Performance Evaluation

Computer systems are complex structures of multiple hardware and software components, with each component affecting the performance of every computation. To measure and improve performance requires a full understanding of how a system's components interact and the unexpected ways that changes in one component may affect others. In this course we will develop methods for measuring computational performance that account for these complex interactions. We will optimize computations first by empirically identifying the greatest opportunities for improvement and then by evaluating those optimizations through proper measurement and analysis.

Topics will include compiler optimizations, memory hierarchy performance, probe effects, cache contention, parallel bottlenecks and scaling, and performance analysis techniques.

Requisite COSC 171 and 211. Spring semester. Professor Kaplan.