The goal of this CUDA or OpenCL project is to implement a non-trivial parallel program. By default, that program would be the one described in Assignment 2a. However, especially if you are a grad student registered under the 699 number, you should pick your own project.
The project you pick can use either CUDA or OpenCL, but you must state which you intende to use. The project should require kernel(s) containing at least 32 statements (i.e., bigger than Assignment 0, but not necessarily as big as Assignment 1). You may use pre-written library code with appropriate citation, but that code will not count toward the minimum complexity requirement. Pick a topic that interests you... I'll help you find a good algorithm and mapping of it to a GPU.
The actual project will be due December 12, 2011. However, if you want to pick your own project, it must be approved by me first. The deadline for submitting a brief HTML description of your proposed project is Saturday, November19, 2011. I will get back to you with approval and/or changes by November 22, 2011.
Submit your HTML tar file here:
GPU Computing