[gmx-users] GROMACS with SGI native compliers

[Haruna Cofer]

Nathan Moore wrote:
> 
> Did you test the LYS/PME benchmark
> (http://www.gromacs.org/benchmarks/single.php)?  Usage of the Partial Mesh
> Ewald sum (PME) requires a bunch of Fourier transforms.

Yes, I did.  But maybe I ran it incorrectly, because when I profiled the
benchmark, I did not see any fftw routines being called.  This was with
GROMACS 3.2.1, and at the time, I saw 39% of the time being spent in
inl3130() from gmxlib/innerc.c, 14% of the time spent in ns5_core() from
mdlib/ns.c, and various smaller amounts in other routines.  How strange.

-- Haruna @ SGI