This project was part of coursework for CS61C at the University of California, Berkeley.
Optimization of a formula for doing 2D convolution on images. 2D convolution is the main machinery behind most forms of image processing. From computer vision to feature detection, from processing MRI scans to making a selfie look like pencil sketches, 2D convolution is the main computation that is involved in a large portion of image processing algorithms. So it is greatly desired to have a very fast implementation of this computationally intensive operation, since image procesing relies heavily upon it.
This project demonstrates the use of intels intrinsics, parallelization, register caching, and other optimizations. This project also demonstrates working with a partner, and thorough understanding of the differences between heap, stack, and static memory.
Project details can be found here: