These codes for matrix multiplication are the ones that were used in the Single Processor Performance module. The profiling runs described below were done using the Microsoft Visual C/C++ compiler, version 6.0 and the Compaq Visual Fortran compiler, version 6.6. Performance runs were done on a single-processor Pentium 3 machine with a clock speed of 1 GHz.

A matrix size of 1024x1024 was appropriate for producing executables that run for at least a few seconds.