Two Projects from the GPU Programming class I TA-ed for.

The first is a matrix multiply using shared memory.

The second is an example of the sobel algorithm which is used in edge detection for computer vision applications. A naive nested version is included then commented out with a faster and much simpler algoirith below it. This algorithm could be further improved by using sequencial memory accesses from global memory. I helped to create the run.bash script in the Sobel folder which was used in a competition for the class.