AMBER Archive (2007)

Subject: Re: AMBER: Portland Group vs. Intel Compilers.

From: Andreas Svrcek-Seiler (svrci_at_tbi.univie.ac.at)
Date: Tue Oct 23 2007 - 15:24:29 CDT

Hi,
> Even through the Intel C and FORTRAN compilers are without cost
...they are not *really* without cost. Most amber users probably
don't qualify for "free unsupported use", at least if the usage conditions
have not changed significantly since I last read them
(see http://www.intel.com/cd/software/products/asmo-na/eng/219692.htm)
> there seems
> to be a lot more problems using them to compile AMBER than using Portland
> Group compilers.
...Why not just try the intel compilers (this is for free in any case)?
If they work and are cheaper use them, otherwise try someone else's.

Personally I've hardly ever had problems with the intel compilers,
less than with other brands I've tried. Besides, if you have a problem
and tell them, they fix it (this can take months, though, but the don't
forget anything).

> I am not trying to start a religious war but would
> appreciate comments on the pros and cons of each compiler.
...Pathscale, Intel and (presumably) the Portlang Group offer
trial downloads - try before you buy.

good luck,
Andreas

P.S.: If you try Pathscale make sure NLSPATH is unset when invoking the
compiler (they don't fix bugs even if told, this problem is old and simple
to fix, but still present in the last trial version I tested).
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu