CUDA C: K-means and MCMC
Day: Monday, October 7Time: 2:10 PM - 3:00 PM
Place: Snedecor Hall 2113
At this point in the series, we're ready to parallelize some simple but interesting statistical algorithms. The first example in this talk is classic Lloyd's K-means in two dimensions, and the second is a Markov chain Monte Carlo from an example from Dr. Jarad Niemi's STAT 544 class from the spring semester of 2013. The MCMC will also be a preview of topics to come in future lectures, such as shared memory, the CURAND library, and the Thrust library. More importantly, both examples will showcase some best practices in CUDA C programming, including error checking with Valgrind and CUDA MEMCHECK.
Materials
SlidesVideo