GPU Architecture and Programming. 3 credits

Catalog description

This course introduces the major architectural building blocks of contemporary Graphics Processing Units (GPUs). Lectures cover the GPU massive parallelism concept and techniques for developing optimum-performance programs targeting GPU platforms. The course introduces the widely-used Compute-Unified Device Architecture (CUDA) GPU programming language and covers methodologies for using CUDA to implement parallel sorting, reduction, numeric iterations, and fundamental graphics operations, such as ray tracing, in a manner that progressively increases the sophistication and performance of the developed programs. Students implement hands-on laboratory exercises on a high-performance UNIX cluster. Offered by Electrical & Comp. Engineering . Limited to two attempts.