During fall semester 2012 I took a class entitled How to Write Fast Code. Among other techniques and technologies, we learned a lot about using GPUs for solving highly-parallel problems. Since the material is quite interesting, I presented an overview of the subject matter at a HackPittsburgh Friday night talk in January 2013. The slides don’t fully “stand on their own” as I always try to just show images and then explain them in person during the talk, but they may be interesting to look at regardless.
High Performance Computing with Graphics Processing Units (PDF, 2.2MB)