Time: 17:00
Room: S4|10-1

Parallel Computing with CUDA

Modern graphics cards are not only powerful rendering systems but can also be used as massively parallel compute devices. Applications that are well suited for the cards' SIMD architecture can achieve speed ups of more than two orders of magnitude compared to single CPU implementations. In this talk, I will give an introduction to NVIDIA's CUDA technology which allows C-like programming of massively parallel applications on graphics hardware. I will also discuss general requirements for problems that are well suited for porting to CUDA. Finally, I will briefly describe one of our current CUDA-based projects which we perform in collaboration with the Bioinformatics group at TUD.