This course is a graduate level seminar style course covering performance measurement, diagnosis, and improvement of applications running in current and near-future high end systems. We will cover the fundamentals of computer performance measurement, then move on to an in-depth study of performance in environments such as servers and clusters, that include multiple single or multi-core processors. Specific topics include measurement tools and techniques, measurement error, Operating System interference, multithreading, and correct diagnosis and representation of runtime behavior.
The course readings are research papers and there is no textbook for the course.
In addition to the readings, you will be responsible for presenting one paper to the class, and for completing a project. In place of a final exam, students will present the results of their projects in an extra class session during exam week. Your grade will be determined as a combination of your project, your class presentation, and your weekly class participation.
For class Wed Jan. 16: "gprof: a call graph execution profiler" Susan L. Graham, Peter B. Kessler, Marshall K. McKusick, ACM SIGPLAN Notices Volume 39, Issue 4 (April 2004) Best of PLDI 1979-1999. ua