I am trying to run SANDER in parallel in cluster. In Gromacs, I used to run
dplace -c 0-59 mdrun -v -s em.tpr -c em.pdb -nt 60 (here I was defining CPU numbers 0-60, and -nt was assigning the number of nodes).
For Sander in Amber, I tried
mpirun -np 10 sander -O -i min.in -o min.out -p test-solv.prmtop -c test-solv.inpcrd -r min.rst -ref test.inpcrd
I too also have tried
mpirun -np 10 pmemd -O -i heat.in -o heat.out -p test-solv.prmtop -c min.rst -r heat.rst -x heat.mdcrd -ref min.rst -inf heat.mdinfo &
However, it is running only in one cpu. I too have tried using -nt 10 or -t 10 etc. This is firing errors.
In this regard, can somebody help me in finding the correct command that can assign 10 CPU to calculate MD using Sander. I have referred mpirun from Amber FAQ and group discussions, but these are not working in my cluster.
Dear Bikash,
are you sure that you have enabled mpi support in compiling AMBER?
If yes you should have also the MPI version of sander called sander.MPI or pmemd.MPI.
Now you are trying to run in parallel the single-processor version of sander!
Let me know
F.
Respected Dr. Federico,
I have not installed the MPI version. Previously I was using Gromacs, where I can use parallel only by typing -nt with mdrun. So I was thinking, I could do the same in amber. I will uninstall, and will reinstall the MPI version as suggested.
Thanking You
Sincerely
Bikash
Hi Bikash
You do'nt have to uninstall your serila version of amber, just go to the manual. You have to install parallel, when you have installed serial version.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology