Single-CPU benchmark on a 2-socket dual-core Intel Woodcrest using the
SSE3 units.
Remarks:
- mod2d  : Incorrect answers for N = 30 and 75 <= N <= 500.
           However, when run with optimisation level -O3 (so, without
           the SSE units, the answers are all correct and also not
           very much slower than with the SSE units: compare
           mod2d.log and mod2d-O3.log
