On May 6, 2010, at 11:58 AM, Alan wrote:
> Thanks Ben,
>
> BTW, I've heard here in the list that MPICH2 is the one to use with Intel,
> but then I didn't try this.
>
I was an openmpi fan at first, but the numbers we got on our 8- and 16-core SMP Linux systems encouraged me to give it a try on Mac OS X.
I'll see if I can find my old benchmarks and get back with you.
You will have to first compile mpich2 with wrappers for icc/ifort.
I have given instructions before on how to do this - so search the ambermd mailing list archive.
I decided to put my mpich2 in /opt/mpich2 to keep it separate from the system mpi utilities.
Double check that the mpich2 binaries come before /usr/local/bin or /usr/bin in your path.
But it seems he figured it all out.
> Cheers,
> Alan
>
> On Thu, May 6, 2010 at 18:26, Ben Roberts <roberts.qtp.ufl.edu> wrote:
>
>> Hi Alan,
>>
>> On 6/5/2010, at 10:18 a.m., Alan wrote:
>>
>>> Hi Ben,
>>>
>>> As far as I remember, you're trying to install AT1.4/Amber11 on a Mac
>> with
>>> Intel Compilers?
>>>
>>> Have you tried with GNU compilers first? Are you using Fink or Macports?
>>>
>>> In the linked email I didn't give some pertinent details for those
>>> installing Amber with Fink.
>>>
>>> For sure that for production I wouldn't use Intel compilers right now and
>> it
>>> doesn't seem you miss much with GNU.
>>
>> Thanks for the helpful response. A bit more Googling revealed the problem.
>> It seems that Intel compilers have certain issues with Open MPI. When that's
>> fixed is, I guess, up to the Intel programmers (or maybe the Open MPI guys).
>>
>> In the meantime, MPICH2 gives much better outcomes, so I've switched to
>> using that for the time being.
>>
>> I'll put something up at the Amber On Mac blog presently, too.
>>
>> Cheers,
>> Ben
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
>
>
> --
> Alan Wilter S. da Silva, D.Sc. - CCPN Research Associate
> Department of Biochemistry, University of Cambridge.
> 80 Tennis Court Road, Cambridge CB2 1GA, UK.
>>> http://www.bio.cam.ac.uk/~awd28<<
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu May 06 2010 - 10:30:03 PDT
