A codeless introduction to GPU parallelism
Day: Monday, September 23Time: 2:10 PM - 3:00 PM
Place: Snedecor Hall 2113
This talk will explain GPU parallelism without the use of computer code. After reviewing the paradigm in general, I will explain how to parallelize some example algorithms, including pairwise summation, matrix multiplication, the classic K-means clustering algorithm, and a Markov chain Monte Carlo. The ideas presented will lay the conceptual foundation for the next three talks, which discuss how to implement GPU parallelism with CUDA C.
Materials
SlidesVideo