I'm not happy in running an ortodox fortran 77 (fair complex) code on a bi-processed Xeon 2.4 machine. I did a benchmark among others PC I have around. Comparing A and B
[ A: Intel Xeon CPU 2.40GHz & 1GB Mem
B: Intel Pentium 4 CPU 2.66GHz & 256MB Mem ]
I got good results from system benchmark. Even so, I got sad results for my fortran code implementation (both machines running IFC 7.1). The Pentium machine overwhelms the dual Xeon !
Things I have done (whithout success):
1- Installed new ifc 8.0;
2- turned hyperthreading off on setup interface;
3- keep same flags on both compilations (-O3);
4- added -parallel flag on the 2-Xeon machine;
5- run mpirun, (without anyother threading directive).
I'm not familiar with OpenMP or MPI, and don't have anyone around with enough time/skills to rewrite the code.
I would appreciate any clue on this issue,