RE: AMBER: Test Failure during amber9 serial compilation

From: Ross Walker <ross.rosswalker.co.uk>
Date: Tue, 25 Apr 2006 12:30:19 -0700

Hi Thomas,

> I have compiled and tested a version of amber9 using gcc 4.0.2 on an
> Athlon Suse Linux system. While most of the tests in the test suite
> PASSED, apart from rounding problems one FAILURE seemed to be severe:

By gcc 4.0.2 I assume you mean gfortran 4.0.2.

I have tried this on my machine (a P4 2.8GHz 32 bit machine) with gfortran
version 4.0.2 20051130 (Red Hat 4.0.2-14.EL4) and it compiles okay and
passes the pb_pgb test fine but on my machine dies when it gets to the
pure_QM_MD test run.nma where it gives the error:

At line 1961 of file _mdread.f
Fortran runtime error: End of record
  ./Run.nma: Program error

This corresponds to the following code.

   do 5 i=1,mxresat
      if (i >= ix(ir+i02)) ir=ir+1
      write(resat(i),'(a4,1x,a4,i4)') ih(m04+i-1), &
            ih(ir+m02-1),ir
      ! ---null terminator:
      resat(i)(14:14) = char(0)
   5 continue

Putting this in debugger shows it dies with the error message above the
first time through this loop... Who knows why? Checking in the debugger
shows that the two ih arrays are fine and contain valid data. What is even
weirder is that this works for Run.nma_md but not for Run.nma - these are
identical except one does MD and the other minimisation. The prmtop and crd
files are identical and the contents of the ih arrays here are identical but
one works fine and the other bombs...

Who knows why. All in all I think it is just a flakey compiler. If anyone
can see anything wrong with the code above please let me know. In the
meantime I would just steer away from gfortran.

> Can someone suggest to me what might be the cause for this problem?

My suggestion of what is the cause of the problem is gfortran. To date it
has been remarkably buggy and I don't think is mature enough yet to be
considered a serious candidate for compiling AMBER. Perhaps later versions
might fix things.

If you want a free compiler that will compiler and run AMBER 9 correctly I
would get g95 (http://www.g95.org). Download the binary distribution of it
and you are good to go.

All the best
Ross

/\
\/
|\oss Walker

| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
| http://www.rosswalker.co.uk | 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 amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Apr 26 2006 - 06:07:11 PDT
Custom Search