> > ac 5.8s 6.0s 24.7s 23.0s + 37.9s > > Hm, so that regresses even more. Weird. I'd like to have it to > investigate.
Sorry this was a typo, the first two lines should be: ac 5.8s 6.0s 24.7s 23.0s + 4.0s aermod 223.9s 227.3s 485.8s 467.6s + 233.1s > Can you send me links to this (air) and the previous > (ac) so I can investigate and maybe improve it further? All the sources are on http://www.lps.ens.fr/~dominiq/polyhedron/ under name.f90. There are also some variants with hand optimization: ac.v1, air.v1, chan.v1, indu.v2, indu.v3, indu.vm, test_fpu_v1, for those interested. Except for the indu.v*, the optimization has been minimal. Cheers Dominique