This document compares the performance of three iterative methods - Gauss-Seidel, Point Successive Over-Relaxation, and Generalized Minimal Residual - for solving large sparse linear systems arising from computations of incompressible Navier-Stokes equations. The methods are applied to solve the lid-driven cavity benchmark problem. It is found that as the mesh size increases, GMRES converges faster and requires less CPU time than the other two methods.