Hi, Chunli
For sequence job, the copies for each image will run in sequence but
for different images you still need parallel threads.
Here you define 8 images so you need 8 GPUs and run with pmemd.cuda.MPI
Feng
On Tue, Aug 1, 2017 at 4:07 PM, Chunli Yan <utchunliyan.gmail.com> wrote:
> Dear Pan,
>
> I am running the sequential jobs not. Attached were my input files (plus
> colvar file).
>
> The job was running on comet (San Diego Supercomputer Center
> <http://www.sdsc.edu/support/user_guides/comet.html>):
>
> #!/bin/bash
>
> #SBATCH --job-name="gpu_test"
>
> #SBATCH --output="gpu_test"
>
> #SBATCH --partition=gpu-shared
>
> #SBATCH --nodes=1
>
> #SBATCH --gres=gpu:p100:1
>
> #SBATCH --ntasks-per-node=7
>
> #SBATCH --export=ALL
>
> #SBATCH -t 48:00:00
>
>
> module load cuda/8.0
>
> module load intel/2013_sp1.2.144
>
> module load mvapich2_ib/2.1
>
>
> cd /oasis/scratch/comet/cyan/temp_project/string/dna1
>
> export AMBERHOME=/home/cyan/program/amber16/
>
>
>
> ibrun -n 1 $AMBERHOME/bin/pmemd.cuda -O -i md01.in -o remd.mdout.001 -c
> 0.rst -r remd.rst.001 -x remd.mdcrd.001 -inf remd.mdinfo.001 -p dna.prmtop
>
>
> Thanks,
>
>
> Best,
>
>
> *Chunli*
>
>
>
> On Tue, Aug 1, 2017 at 3:59 PM, Feng Pan <fpan3.ncsu.edu> wrote:
>
> > Hi, Chunli
> >
> > You should only use the pmemd.cuda.MPI to do string method since string
> > method requires
> > many images on the path.
> >
> > In the error, it seems some integer is divided by zero, could you send me
> > the mdin file to
> > check if any error in inputs?
> >
> > Best
> > Feng
> >
> > On Tue, Aug 1, 2017 at 3:00 PM, Chunli Yan <utchunliyan.gmail.com>
> wrote:
> >
> > > Hello,
> > >
> > > I want to test string methods using pmemd.cuda.
> > >
> > > Below is the steps how I compile amber16 with AmberTools16,
> > >
> > > 1. export AMBERHOME=`pwd`
> > > 2. ./update_amber --update (I can not pass update.4 and update.9 that
> > > relate to string methods)
> > > 3. ./update_amber --apply patch=../../nfe_advance.patch/
> > nfe_advance.patch
> > > 4. ./configure -cuda intel
> > > 5. make install
> > >
> > >
> > > But I always have problems to pass update.4 and update.9, it complains
> > > below:
> > >
> > >
> > > Applying AmberTools 16/update.9.gz
> > >
> > > /home/cyan/program/amber16/updateutils/patch.py:334: PatchingWarning:
> > > Permissions not found for every file in
> > > .patches/AmberTools16_Unapplied_Patches/update.9.
> > > All new files will be read-only!
> > >
> > > 'files will be read-only!') % self.name, PatchingWarning)
> > >
> > > Applying Amber 16/update.4.gz
> > >
> > > /home/cyan/program/amber16/updateutils/patch.py:334: PatchingWarning:
> > > Permissions not found for every file in
> > > .patches/Amber16_Unapplied_Patches/update.4.
> > > All new files will be read-only!
> > >
> > > 'files will be read-only!') % self.name, PatchingWarning)
> > >
> > >
> > > Once I finish the installing and ran string methods with pmemd.duda. I
> > got
> > > the following error:
> > >
> > >
> > > forrtl: severe (71): integer divide by zero
> > >
> > > Image PC Routine Line
> > Source
> > >
> > > libintlc.so.5 00002B862BA802C9 Unknown Unknown
> > Unknown
> > >
> > > libintlc.so.5 00002B862BA7EB9E Unknown Unknown
> > Unknown
> > >
> > > libifcore.so.5 00002B862A72908F Unknown Unknown
> > Unknown
> > >
> > > libifcore.so.5 00002B862A690D7F Unknown Unknown
> > Unknown
> > >
> > > libifcore.so.5 00002B862A6A2309 Unknown Unknown
> > Unknown
> > >
> > > libpthread.so.0 000000397460F7E0 Unknown Unknown
> > Unknown
> > >
> > > pmemd.cuda 000000000071B524 Unknown Unknown
> > Unknown
> > >
> > > pmemd.cuda 00000000006C4EF6 Unknown Unknown
> > Unknown
> > >
> > > pmemd.cuda 00000000004E5CD6 Unknown Unknown
> > Unknown
> > >
> > > pmemd.cuda 0000000000409F66 Unknown Unknown
> > Unknown
> > >
> > > libc.so.6 0000003973E1ED1D Unknown Unknown
> > Unknown
> > >
> > > pmemd.cuda 0000000000409DE1 Unknown Unknown
> > Unknown
> > >
> > > [comet-34-06.sdsc.edu:mpispawn_0][child_handler] MPI process (rank: 0,
> > > pid:
> > > 186206) exited with status 71
> > >
> > >
> > > Does anyone know what is wrong in it?
> > >
> > >
> > > Thanks,
> > >
> > >
> > > Best,
> > >
> > >
> > > *Chunli*
> > > _______________________________________________
> > > 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
> >
>
> _______________________________________________
> 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 Tue Aug 01 2017 - 14:00:02 PDT
