Brent Leback, PGI Applications & Service Manager for NVIDIA (which recently acquired PGI), will present “deep-dive” PGI compiler classes on OpenACC and CUDA Fortran on Aug. 19 and 20, at the NCAR Mesa Lab in Boulder. Both classes will be in the Main Seminar Room.
The OpenACC class is scheduled for 9 a.m. to 4 p.m. Tuesday, Aug. 19. Topics will include an introduction to accelerated computing, parallel and loop directives, kernels constructs, data directives and lifetimes, optimization strategies, CUDA-like interoperability, and others.
The CUDA Fortran class is scheduled for 9 a.m. to 4 p.m. Wednesday, Aug. 20. Topics will include an introduction to CUDA Fortran (CUF), CUF kernels, compilation, profiling, debugging, optimization strategies, OpenACC interoperability, managed data, and more.
Leback has worked in various positions over the past 30 years in HPC customer support, math library development, applications engineering, and consulting at QTC, Axian, PGI, STMicroelectronics, and now NVIDIA. Until recently he managed the compiler group at PGI.