[CP2K:1278] compiler influence question & intro

[Teodoro Laino]

Dear Ondrej,

On 6 Aug 2008, at 10:25, Ondrej Marsalek wrote:

> my first contact with cp2k was a miscompiled build created by pgi. i
> have made a working build of the whole library stack and cp2k with gcc
> and g95. i am not getting anywhere near pgi-compiled cp2k, unless
> someone convinces me that it is worth the extra effort needed to get
> it right. does anyone have a direct comparison of g95 and pgi cp2k
> builds on amd64 (or intel compiler on intel hw) as for overall speed?

PGI has serious problems. The only thing I can tell you is that an  
*extremely* good
version is the 6.2.5. People here have reported to have used also  
recent versions (like
7.1.4) but my personal experience is that starting from 7.0.0 up to  
7.2.3 I could not find
any compiler version that properly compiled cp2k (unless you don't  
compile few *peculiar*
files with -O0). Of course, based on my personal experience, you may  
easily find that
one version is working with an architecture while it's not working on  
another one..
so bugs introduced in the several compiler releases seem to be   
architecture dependent or at least
behave like..

Let's go to your question now: as long as you do DFT MD you won't see  
too much difference between
PGI and g95. g95 can be sensibly slow for the classical part but you  
should not see any
sensible difference on the DFT part.
If you really want to put some effort in having a fast executable  
then I would suggest you to go
for the intel compiler+mkl.. It won't be easy.. but at least at the  
end you will have something that both
for DFT or classical will be extremely faster than g95.

Cheers
Teo

>
> regards,
> ondrej marsalek
>
> >