RE: AMBER: yet another SANDER bug?

From: Ross Walker <>
Date: Sat, 27 Jan 2007 20:18:24 -0800

> most of you should not be affected by this one, but anyway,
> here it goes:
> --- sander.f 2006-04-03 19:35:55.000000000 -0400
> +++ sander.f 2007-01-27 21:49:31.000000000 -0500
> .@ -1004,7 +1004,7 @@
> if( ntb>0 .and. ifbox==1 .and. ew_type==0 .and. mpoltype==0 ) &
> call deallocate_m1m2m3()
> call AMOEBA_deallocate
> -
> + close (6)
> return
> end subroutine sander
> [shows up in multisander runs (some compilers); sander.MPI
> dies with the message 'UNIT 6: error on open' (or something
> like that)]

Which compilers?

Also this fix is incomplete. Firstly only the master mpi thread called
amopen so I believe this should be:

if (master) close(6)

Secondly multisander when it completes calls mexit which will also close
this file which means there is a second bug being introduced by this patch
in that the code will try to close an already closed file. I am suprised
that a compiler that refuses to open an already open file also does not
refuse to close and already closed file.

There should be a cleaner way to do this. Probably something along the lines
of an edit in mexit.f to be:

   if (output_unit > 0 .and. status/=0) then
   end if

Maybe I'll put a bugfix together for this if I get a chance but I'd like to
know what compilers give a problem first though.

ps. please start posting under your real name or don't bother.

All the best

|\oss Walker

| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- |
| | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Jan 28 2007 - 06:08:14 PST
Custom Search