Benchmark for Gauss-Seidel method
 

Hello to everybody. I need some benchmarks, in terms of number of iterations vs. number of nodes, for a Gauss-Seidel solver for 2D Laplace equation on a cartesian grid using double precision. The problem I would like them for is standard: x = 1 on boundary, initial guess is 0 and no production (hey, it' s Laplace ), but numbers relative to any standard problem will do the same. I know from theory that the behaviour should asimptotically approach linearity, and I did write a routine myself, whose results show the correct behaviour but, since I need this for a research project, I would like to have actual, absolutely error-proof numbers. If you know them, or if you can provide a good reference, or if you can address me to a routine which is surely bug-free, please let me know ASAP. Thanks in advance,

Andrea Panizza