[MITgcm-support] ifort optimization puzzle

Martin Losch mlosch at awi-bremerhaven.de
Wed Mar 9 05:11:12 EST 2005


Constantinos,

my full set of options is this:
ifort -132 -r8 -r4 -w95 -W0 -WB -O1 -align -tpp7 -xN -c find_rho.f
I tried to do it without the -xN flag, and yes, no vectorization and no 
problems. So the problem is indeed the vectorization. Sorry, for not 
reporting this detail (on /data4/mlosch/test1d.tgz you'll find all 
information necessary for reproducing my problem, including Makefiles 
and build option files, see my first email).

Martin

On Mar 9, 2005, at 10:56 AM, Constantinos Evangelinos wrote:

> On Wednesday 09 March 2005 02:40 am, Martin Losch wrote:
>
>> So, what do I do now? I have recompile find_rho with -O0 (the rest 
>> with
>> -O1) and everything looks OK, but maybe I can find out what type of
>> statement produces the problem and replace it by something more 
>> stable.
>> Does anyone know offhand the compiler directives for ifort that make
>> the compile vectorize or not vectorize specifice loops?
>
> Even -O3 will not do any vectorization of loops without a -axW/N flag. 
> So the
> issue is not vectorization but some rather innocuous (but in this case
> deadly) optimization that the compiler does.
>
> Constantinos
> -- 
> Dr. Constantinos Evangelinos
> Department of Earth, Atmospheric and Planetary Sciences
> Massachusetts Institute of Technology
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list