[gmx-users] grompp

[Nathan Moore]

I'm working with a ppc32 system with the intention of finding the best set
of optimizations for the architecture and xlC compiler. (eg comparing
"-O3" to "-O3 -qhot", etc)

In testing performance, is it safe to compile only different versions of
the mdrun executable, while using the same grompp to generate the .tpr run
files?

Does anyone have --disable-float, (ie double-precision) benchmark data for
Villin (or any of the other GROMACS benchmarks)?  I'm very intertested in
making this comparison.


regards,

NT Moore