In Partial Fulfillment of the Requirements for the Degree of Master of Science
will defend his thesis
Enhanced Debugging of Data Races in Parallel Programs Using OpenMP
Multicore computers are everywhere, with manycore computers on the horizon. They can be found in many applications from gaming devices, embedded systems, desktop/laptop computers and supercomputing. The presence of this technology brings opportunities as well as challenges. One challenge is appropriately synchronizing and data scoping shared-memory variables to ensure program correctness. Programmers of shared-memory computers commonly generate data race errors that may not be detected using standard verification techniques, due to their non-deterministic nature.
Stand-alone static and dynamic tools are currently the most common approach to detecting data races, however, these techniques suffer from a lack of precision and excessive detection execution times. Combining the benefits of static and dynamic analysis in a complementary fashion, can improve the accuracy and practicality of data race debugging tools.
The thesis defense will begin with a discussion of synchronization in OpenMP programs, and some of the most common errors with respect to data races. The experiments performed on data race dynamic analysis tools will be presented. Finally, a complementary approach including static analysis and dynamic analysis is proposed to improve the data race detection experience.
Date: Tuesday, November 22, 2016
Time: 1:00 PM
Place: PGH 362
Advisor: Dr. Edgar Gabriel
Faculty, students, and the general public are invited.