Colloquium: David McAllister

 

Title: CUDA: C for Data Parallel Problems

 

Abstract: I will provide a gentle introduction to GPU programming using Nvidia’s CUDA platform. CUDA offers a simple way to express data parallelism in C or C++. CUDA code executes on Nvidia GPUs in machines ranging from thin and light laptops to Titan, the world’s fastest supercomputer. I will survey a few application domains and give a taste of how to code with CUDA for optimal performance.

Leave a Reply