This document compares the speed of performing finite difference calculations using parallel computing toolbox functions on a GPU versus a CPU. It presents a script that runs timed loops of a 2D finite difference heat equation calculation on arrays of varying sizes, both on a GPU and CPU. The results are stored in a performance table showing the millions of operations per second for different loop and array sizes. The GPU is faster for larger problems, with certain GPU optimizations like using arrayfun and GPU arrays providing additional speedups over normal MATLAB syntax.