Re: [AMBER] Amber16 -mpt-2.15-gcc/6.1.0-RHEL7.3- compilation error

From: Feng Pan <fpan3.ncsu.edu>
Date: Thu, 16 Feb 2017 14:12:04 -0500

Hi, Sudhakar

As I understand, you want to test to check Amber works well with SLURM
queue scritpts?

If so the basic SANDER or PMEMD runs need three files:
prmtop, inpcrd, mdin

It will be very good if you can take a quick look at the basic Tutorial 1
on official website (don't need to much knowledge)
http://ambermd.org/tutorials/basic/tutorial1/
and it also offers all the input files to download, so you can use them to
run a basic test.

Best
Feng



On Thu, Feb 16, 2017 at 12:23 PM, Sudhakar Lakkaraju <slakkaraju.asc.edu>
wrote:

> Hi Feng,
>
> Thank you for F90 code file, now Amber gets compiled with MPT, I did not
> have to roll back the minor versions of Amber16 to fix nfe module issues.
>
> -make.test serial and parallel works fine.
>
> I would like to test my Amber16 install using mpi and SLURM queue
> scripts, where can I find a sample input files to test Amber for
> parallel jobs?
>
> I'm a systems admin so my knowledge of Amber is very limited :)
>
> I really appreciate all the help.
>
> Regards
>
> Sudhakar
>
>
>
> On Feb/9/2017 10:09 AM, Feng Pan wrote:
> > Hi, Sudhakar
> >
> > It seems that the nfe code has problem with the MPT, since Openmpi and
> > mpich don't report these errors.
> > But it is kind of weird that it said
> > Common block member ‘mpi_bottom’ at (1) cannot be an EQUIVALENCE
> > object in the pure procedure ‘f6_12_value’
> > yet I don't use the variable mpi_bottom in nfe codes.
> >
> > I want to do tests on MPT but I don't have access to it and I didn't
> find a
> > way to install it on google. Could you give me a way to use it to do
> tests?
> > Thank you
> >
> > Best
> > Feng
> >
> > On Wed, Feb 8, 2017 at 12:56 PM, Sudhakar Lakkaraju <slakkaraju.asc.edu>
> > wrote:
> >
> >> AMBER16 gets compiled with OpenMPI 1.10.2 with gcc/6.1.0 on RHEL 7.3.
> >>
> >> MPT is an implementation of MPI on SGI ultraviolet machine.
> >>
> >> My gfortran --version
> >> GNU Fortran (GCC) 6.1.0
> >>
> >> Is it possible to turn off "nfe" capability?
> >>
> >>
> >>
> >> On Feb/8/2017 9:42 AM, David Case wrote:
> >>> On Wed, Feb 08, 2017, Sudhakar Lakkaraju wrote:
> >>>> I'm trying to compile Amber16 with MPT/2.15 and GCC/6.1.0 on RHEL
> 7.3.
> >>> I'm thinking that no one has ever tried this before, particularly on
> >>> the recently-added "nfe" capability, which was added to Amber last
> >>> July. Can you try an MPI implementation like mpich or openmpi (see
> >>> $AMBERHOME/AmberTools/src/configure_mpich or configure_openmpi)?
> >>>
> >>> Can you say where the MPT implementation came from? (Is it a standard
> >>> package on RHEL 7?) And, is your gfortran also at version 6.1.0?
> >>>
> >>> Note to other developers: this problem is in the pmemd version, not the
> >> sander
> >>> version.
> >>>
> >>>
> >>>> Error: Common block member ‘mpi_status_ignore’ at (1) cannot be an
> >>>> EQUIVALENCE object in the pure procedure ‘pythag’
> >>>> nfe_colvar.F90:9994:7:
> >>>>
> >>>> use nfe_lib_mod
> >>> ...dac
> >>>
> >>>
> >>> _______________________________________________
> >>> AMBER mailing list
> >>> AMBER.ambermd.org
> >>> http://lists.ambermd.org/mailman/listinfo/amber
> >>
> >> _______________________________________________
> >> AMBER mailing list
> >> AMBER.ambermd.org
> >> http://lists.ambermd.org/mailman/listinfo/amber
> >>
> >
> >
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
Feng Pan
Ph.D. Candidate
North Carolina State University
Department of Physics
Email:  fpan3.ncsu.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Feb 16 2017 - 11:30:02 PST
Custom Search