Cygwin snapshot slow
Jim Dishaw
jim@dishaw.org
Sun Feb 6 12:16:00 GMT 2005
THOMAS Paul Richard <prthomas@drfccad.cea.fr> writes:
> Jim,
>
> You do not report on the version of Windows you are using nor the processor.
> These are important because I tried to get rid of the architecture
> dependence in this snapshot. I rebuilt gmp/mpfr using gcc-4.0 and
> -march=i486. Maybe this latter was a bit too low down the evolutionary
> ladder?
>
> That said, the LLNL Flops Kernel Test gives
>
> 20050131 : 6.1s (user) for compile and a median of 366.6Mflops
> 20041202 : 6.5s (user) for compile and a median of 363.5Mflops
>
> WIndows2000/CYGWIN_NT-5.2/PIV 2.5GHz
>
> Paul T
>
Windows XP/sp2
Pentium M 1.80GHz 512MB
CYGWIN_NT-5.1 1.5.12
Does the current CVS contain the patched namelist I/O?
-jd
More information about the Fortran
mailing list