parallel 2013 » Agenda »

// Data Race Detection with GDB

Data races are one of the most expensive and complex problems in parallel software development. Intel offers an efficient and flexible data race detector that comes with the Intel C++/Fortran compilers and is tightly integrated into GNU debugger as a plugin. It provides numerous features to fine-tune the data race analysis. We demonstrate how to use the data race detector, its key features and best known methods. Our focus is not only on small applications but also on real-world and big-sized applications for which traditional whole program analysis might not be feasible.

// Referent

// Dr. Walfred Tedeschi

is a Software Engineer at Intel, located in Ulm/Germany. He is currently working on GDB enabling for new hardware features. With background in physics he worked about 6 years in developing multithread applications for signal and image processing integrated on medical systems.