Abstract: In this tutorial we'll cover the features OpenMP has incorporated since version 4 to leverage heterogeneous architectures. In particular, the target construct and the various methods for transferring work and data to computational accelerators.
Bio: Tom Scogland is a computer scientist in the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. His primary research centers around the design and development of runtime systems and programming models for scientific applications dealing with heterogeneous and many-core architectures, heterogeneous memory systems, and scheduling in heterogeneous systems. He is an active developer on the flux resource manager and RAJA library as well as member of the OpenMP language subcommittee, WG21 C++ committee and participates in numerous other research and development efforts. Tom holds a B.S. in Computer Science from Purdue University, and an M.S. and Ph.D. in Computer Science from Virginia Tech.