AMBER Archive (2004)

Subject: Re: AMBER: Problem amber8 compilation in Opteron(Linux) with PGI-workstation compiler

From: Jiten (jiten_at_postech.ac.kr)
Date: Mon Jul 12 2004 - 04:36:36 CDT

Hello Dr. Ross Walker,

As Mark Williamson is working on compiling without -big options in
configure - I wish to continue to work around with -big options. In fact I
got the similar massage without the -big options. Unfortunately I could not
find any k8-32 in the config.h fine generated with the -big option.

1. So first compilation with the original config.h
(./configure -opteron -verbose -big pgf90 ) then test run

make -i test > test.log

Here I nottice that many of the modules are passed and many failed. I
attached the modified test.log --> test-org.log which includes only the
failed or program error - also pinpoint where the error after each of
theprogram error/diffing.

The major few findings are depending on the programs that used for the test
example. details are in the attached test-org.log

ASSERTion 'ier == 0' failed in extra_pts.f at line 223.
ASSERTion 'ier==0' failed in ew_direct.f at line 409.
ASSERTion 'ier==0' failed in ew_direct.f at line 457.
ASSERTion 'ier == 0' failed in nmode.f at line 105.

2. I modify the extra_pts.f (in src/sander folder) file as instructed by
you then recompile and test run again

   max11 = natom+numextra
   max12 = 3*(nbonh+nbona+nbper)
   max13 = 3*(ntheth+ntheta+ngper)
   max14 = 3*(nphih+nphia+ndper)
   write(6,'(''max11 = '',i8, '' max12 = '',i8)') max11,max12
   write(6,'(''max13 = '',i8, '' max14 = '',i8)') max13,max14
   maxa = max(max11,max12,max13,max14)
   write(6,'(''maxa = '',i8, '' ier = '',i8)') maxa, ier
   allocate( s3(maxa), stat=ier )
   write(6,'(''ier = '',i8)') ier

make -i test

cd dmp; ./Run.dmp
  ./Run.dmp: Program error
make: [test.sander.no_lmod] Error 1 (ignored)
cd adenine; ./Run.adenine
  ./Run.adenine: Program error
make: [test.sander.no_lmod] Error 1 (ignored)
cd cytosine; ./Run.cytosine
  ./Run.cytosine: Program error
make: [test.sander.no_lmod] Error 1 (ignored)
cd nonper; ./Run.nonper 

--
--

Unfortunately there is no massage written on the screen as expected. I was
confused. But it is true - I did make claen and even removed the exe folder
before the recompilatoins with the modifications.

I hope this information will help you to fix  the problem.

With regards,

Jiten

----- Original Message ----- 
From: "Ross Walker" <ross_at_rosswalker.co.uk>
To: <amber_at_scripps.edu>
Sent: Monday, July 12, 2004 3:38 AM
Subject: RE: AMBER: Problem amber8 compilation in Opteron(Linux) with
PGI-workstation compiler

> Dear Jiten,
>
> > I would like to AMBER program developers that can I still use
> > AMBER modules
> > such as sander etc without any problem even if the test run was not
> > successful.
>
> What is the exact error in the tests? Are you getting "Program Error" for
> every test like in your previous email or are you getting the error Mark
> quoted for just dmp?
>
> Is the dmp test the only one that fails? Try commenting it out of the
> makefile and then running the tests again. This way you can build up a
list
> of failed tests. If all the other tests pass then you are probably safe to
> use Sander (although with caution) as long as you avoid using polarisable
> dipoles. If you plan on using polarisable dipoles (FF02EP) then you are
> likely to have problems. If several tests fail then please let me know
which
> ones and what the error message is in each case and I can try to
investigate
> it further.
>
> It might be worth testing the values of the various variables in
> extra_pts.f. Try adding the following lines around line 223 (those marked
> with ** are the original lines DO NOT modify these):
>
>     write(6,'(''max11 = '',i8, '' max12 = '',i8)') max11,max12
>     write(6,'(''max13 = '',i8, '' max14 = '',i8)') max13,max14
> **    maxa = max(max11,max12,max13,max14)
>     write(6,'(''maxa = '',i8, '' ier = '',i8)') maxa, ier
> **    allocate( s3(maxa), stat=ier )
>     write(6,'(''ier = '',i8)') ier
> **    REQUIRE( ier == 0 )
>
> Modify the code here and then recompile and run the dmp test. Here is what
I
> get on a 32bit P4. You 'SHOULD' get the same. Try it with and without the
> -big congfigure option. Unfortunately I don't have access to an Opteron
> system at the moment to test this myself.
>
> >cd $AMBERHOME/test/dmp/
> >./Run.dmp
>
> >less mdout.dmp
>
> Polarizable options:
>      indmeth =       1, maxiter =      20, irstdip =       0, scaldip =
> 0
>      diptau  =  11.00000, dipmass =   0.33000
> max11 =       13 max12 =       36
> max13 =       60 max14 =       36
> maxa =       60 ier =        0
> ier =        0
>
> --------------------------------------------------------------------------
--
> ----
>    3.  ATOMIC COORDINATES AND VELOCITIES
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | Department of Molecular Biology TPC15 |
> | The Scripps Research Institute |
> | Tel:- +1 858 784 8889 |
> | EMail:- ross_at_rosswalker.co.uk |
> | http://www.rosswalker.co.uk/ | PGP Key available on request |
>
>
>
> -----------------------------------------------------------------------
> 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