NVidia presented a great tutorial on CUDA programming at the HPCMP User’s Group conference today. The slides start with the basics of CUDA & GPGPU technology, and then spread out into some of their specialized CUDA libraries such as CUBLas and NVPP, and discuss several memory optimization strategies. Definitely worth looking over if you’re a CUDA developer.
UPDATE: Now the slides are available on SlideShare for you to review before downloading. View the 192slides, & download the 3MB PDF after the break.