Link to CALCM Home  

Understanding Application Performance

Tuesday April 1, 2008
Hamerschlag Hall D-210
4:00 pm

Marty Itzkowitz
Sun Microsystems

In this talk, we will first describe the importance of performance, and how to go about telling if there's a problem, and how to triage it. We then describe the Sun Studio Performance Tools, and show how they simplify the triaging problem.

We then go present two simple cases of low-hanging fruit in applications, and show how the tools help isolate each type of problem, and point to the place in the user's source code where the fixes are needed. We them explore CPU- and Memory-Performance problems, memory allocation issues, and multithreaded performance problems, in each case showing how the tools can help.

Finally, we discuss the applicability of the tools to measuring Java performance.

Marty Itzkowitz received an A.B. degree from Columbia College and a Ph.D. in Chemistry and Physics from CalTech. After a post-doctoral fellowship at UC-Berkeley, he worked on operating systems and distributed services at LBL. He was head of Operating Systems at Vitesse Electronics, and then worked on operating system performance and performance tools at Sun Microsystems and then at Silicon Graphics. He returned to Sun in 1998 as project lead for the Sun Studio Performance Tools. His interests include operating system design and performance, multiprocessor performance, performance tools and scientific visualization. He is an avid handball player and cook.

He has written and/or presented papers at various conferences, including: The Second International Conference on Distributed Computing Systems, Paris, 1981; AFUU Convention Unix 90, Paris, 1990; Supercomputing Debugging Workshop, Albuquerque, 1991; Workshop on Debugging and Performance Tuning for Parallel Computing Systems, Cape Cod, 1994; SuperComputing '96, Pittsburgh, 1996; WOMPAT 2000, San Diego 2000; SuperComputing '03, Phoenix, 2003; and the OpenMP BOF at SuperComputing '06, Tampa, 2006.


Department of Electrical and Computer EngineeringCarnegie Mellon UniversitySchool of Computer Science