AMBER Archive (2008)

Subject: Re: AMBER: Installing OpenMPI

From: Hopkins, Robert (hopkins_at_uhcl.edu)
Date: Wed Jan 23 2008 - 18:40:22 CST


Dear Dr. Pietra and Dr. Case,

Thank you both for your comments -- I'm relieved to know that with LAM
and a glass of wine in hand, I should be able to circumvent some of the
problems of OpenMPI :-)

Cheers!
Bob Hopkins
 

-----Original Message-----
From: owner-amber_at_scripps.edu [mailto:owner-amber_at_scripps.edu] On Behalf
Of Francesco Pietra
Sent: Wednesday, January 23, 2008 12:19 PM
To: amber_at_scripps.edu
Subject: [Subject Filtered] Re: AMBER: Installing OpenMPI 7%

cygwin. Although I don't know it, I always found curious that someone
gets
involved in simulating an OS that is free as Linux. Normally, one should
expect
limitations in simulations. With Linux, getting parallelization of Amber
upon
OpenMPI is plain doing. I understand the inverse path, simulating
Windows
because certain non-computational programs have been written at the time
of
non-graphical DOS and continued with Windows. A case in point: I
simulate
Windows with wine in order to use Asksam, which has no counterpart in
unix OSs
and, being useful even without fields, is the only database that I found
useful
during the study of literature, better than a book notes. But, is it
anything
in cygwin that justifies its existence? As cygwin exists, there must be
such a
thing. Cheers, francesco pietra

--- "David A. Case" <case_at_scripps.edu> wrote:

> On Tue, Jan 22, 2008, Hopkins, Robert wrote:
>
> > I used the g95 Fortran compiler (I'm operating in a Cygwin
> > environment on a Windows machine). As a result, I went ahead and
tried
> > to install OpenMPI
>
> I've never had any luck using openMPI with cygwin. The only MPI that
works
> for me is LAM, and that compiles out of the box. As an alternative,
you
> can get precompiled libraries for MPICH2 and link those in.
>
> If someone on the list knows how to make openMPI work with cygwin, I
hope
> they
> will contribute their knowledge. LAM is pretty good, but suffers
(again, for
> me) the problem that the daemon can get confused if you try to run
lots of
> MPI
> jobs one right after another (e.g. as in running the test cases). But
it
> seems to work fine for "real" runs.
>
> ...dac
>
>
-----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

 
________________________________________________________________________
____________
Looking for last minute shopping deals?
Find them fast with Yahoo! Search.
http://tools.search.yahoo.com/newsearch/category.php?category=shopping
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu