Re: Gfortran vs Ifort when using MPI

[Tobias Burnus <burnus at net-b dot de>]

Tim Prince wrote:
> > I think I have used the right compiler options (-O3 -ffast-math
> > -funroll-all-loops), so I don't think that is the culprit?
Depending on the system, using the right -march makes a big difference. 
Try -march=native (not supported on all targets). (If the compiler 
thinks that there is no SSE, -march=native can help a lot; if I recall 
correctly, for x86 the settings are rather conservative.

> Effective auto-vectorization would give an advantage of maybe 2x 
> speed. If you want a gfortran where you don't need to turn that option 
> on explicitly, upgrade to the current release.
-ftree-vectorize (plus -fvect-cost-model) is the option, which is in 
newer gfortran's enabled by default.

Tobias