Hazards of Multi-threaded and Multi-core Software Development
Data races are risky because they are easy to introduce and difficult to find. A widespread misconception is that some forms of data races are harmless. However, modern optimizing compilers generate code that can cause incorrect execution when data races exist, even for those that are thought to be benign. This 40-minute technical presentation will describe the causes and risks of concurrency defects and demonstrate the surprising ways in which programmers can inadvertently introduce such bugs into their code. Ways of finding and eliminating such defects automatically with static analysis will be shown.
Paul Anderson, VP of Engineering, GrammaTech
Paul Anderson is VP of Engineering at GrammaTech. He has managed the product-development and support teams for over 10 years. His articles have been featured in Dr. Dobbs, Embedded Systems Design, and other popular magazines. Paul completed his Ph.D. at City University London.
Ben Ylvisaker, Senior Scientist, GrammaTech
Ben Ylvisaker is the Principal Investigator for a DARPA-sponsored research project on static analysis for multi-core architectures. He is also the primary developer of GrammaTech's concurrency analyses. Ben completed his Ph.D. at the University of Washington, where he focused on concurrent programming language semantics.