Back in 2001, researchers decided to try to use "the other" processor in their PC (the graphics card) for computation. These early, tedious efforts were promising and lead to a new type of computation called General Purpose GPU (GPGPU) Computing. NVIDIA vastly accelerated such efforts with the release of CUDA in 2007, that provided a much simpler interface to programmers wanting to use the GPU for general computation. Today, GPUs are being used to accelerate some of the most powerful supercomputers in the world (see top500.org). In this talk I'll describe the ideas behind this very different kind of computing, show some simple CUDA examples, and mention the open source alternative to CUDA (OpenCL). I'll give a bit more detail on our newly acquired NVIDIA K20, and briefly outline my GP-GPU Computing course for summer 2013.

About the speaker

Dr. Karavanic is an Associate Professor of Computer Science at Portland State, where she teaches courses in Operating Systems, performance measurement and modeling, and High Performance Computing. She was selected as an HPC Educator for the SC2012 Conference, where she offered a full-day tutorial on CUDA to faculty from around the U.S. Dr. Karavanic came to PSU in 2000 from Madison, Wisconsin, where she earned an M.S. and a Ph.D. in Computer Science.