AMBER Archive (2000)

Subject: Re: I can't run sander6 with lam-mpi : is mdread.f wrong ??

From: Thomas A. Spraggins (tas_at_holmes.acc.virginia.edu)
Date: Wed Jun 28 2000 - 11:10:14 CDT


I've had various problems with namelist reads on both AIX and Alpha Linux
(the latter with the Compaq compiler). I remember this being a
problem with sander specifically under AIX; I can't remember where I
ran into this on the Alpha, though. There are two styles of namelist
input, usually called "old" and "new" (how helpful!).

Anyhow, you may want to see if there's any reference to "namelist" in the
PGI docs. You could, of course, reformat your input file, but that's
probably more work.

Hope this helps.

Tom Spraggins
tas_at_virginia.edu

On Wed, 28 Jun 2000, David Konerding wrote:

> Xavier Deupi writes:
> >Hi,
> >
> >we've compiled Amber6 on an AMD K7 600, running RedHat 6.0, using
> >Lam-MPI and PGI compilers.
> >
> >The compilation suceed, but when we send the calculation we get this error
> >message:
> >
> >PGFIO-F-239/namelist read/unit=5/entity name is not member of group.
> > File name = tap_gp.sander formatted, sequential access record = 3
> > In source file _mdread_.f, at line number 808
> >16789 (n0) exited with status 1
>
> Dear Xavier,
>
> I submitted one bugfix for PGI, see:
> http://www.amber.ucsf.edu/amber/bugfixes/6.0/bugfix.17
>
> It wouldn't surprise me if there are a few others. I assume you already
> checked whether it works without LAM? I've had nothing but trouble with LAM.
>
> Dave
>