The University of Maryland is holding its first annual GPGPU programming competition! The challenge is to write a generic and efficient kernel for sparse matrix operations. The underlying goal of the competition is to help support future work and research by providing open and efficient solutions to common problems. As such, all source code will be released under a free software license at the close of the competition. Nvidia is helping to support this effort by providing some prizes for the winner of the competition. You can get all the information at http://scriptroute.cs.umd.edu/gpucompete.