multicore and gputools – Quick Parallel Computing in R – December 4
December 4, 2012 @ 12:00 am
R Users Group Meeting – Tuesday, December 4, 3:30 – 5:00 PM Modern Languages Building, Room 2001A Bobby Yuen, PhD student in Statistics
In this talk I will cover two methods of parallel processing in R. The first is using a multiple core CPU with the package multicore. This simple package is easy to use and works well for embarrassingly parrallel jobs like replications in simulations. The second method I will talk about is parallel computing on the GPU (Graphical Processing Unit) through the package gputools. This method requires a CUDA enabled NVIDIA GPU. Processing on the GPU can drastically cut computation time for some basic linear algebra routines.