[gmx-users] Multinode parallel GPU/CPU Gromacs performance

Szilárd Páll pall.szilard at gmail.com
Fri Jun 16 15:47:08 CEST 2017


Petar,

Please show full log files, it would be easier to give suggestions. One
thing that is clearly suboptimal is that you're using #ranks = #cores,
you'd probably get better performance with 4-6 ranks per node.

Cheers,

--
Szilárd

On Sun, Jun 11, 2017 at 11:17 AM, Petar Zuvela <petar.zuvela at nus.edu.sg>
wrote:

> Dear all,
>
> For a system with 32267 atoms (metalloprotein with bound ligand, using a
> parametrized bonded model, with Amber FF14SB; force field solvated in a
> cubic water box and with a few Na+ ions as counterions), run on 4 nodes
> with:
>
> 1. Intel(R) Xeon(R) CPU X5650  @ 2.67GHz (12 cores per node)
> 2. NVIDIA Tesla M2090 (2 per node)
>
> performance is about 73 ns/day.
>
> Performance on one node is about 25~30 ns/day. MD is run as such (
> LSB_DJOB_NUMPROC=48):
> mpirun -f ${LSB_DJOB_HOSTFILE} -np ${LSB_DJOB_NUMPROC} mdrun_mpi -nice 0
> -ntomp 1 -nb gpu -npme 24 -v -s ./Output/MD_01_200/MD_011.tpr
> -cpo ./Output/MD_01_20/MD_011.cpt -c ./Output/MD_01_200/MD_011.gro -e
> ./Output/MD_01_200/MD_011.edr -g./Output/MD_01_200/MD_011.log -x
> ./Output/MD_01_200/MD_011.xtc
>
> Gromacs version is: 5.1.1.; openMPI version is: 1.8.8; CUDA driver version
> is: 7.50; CUDA runtime version is: 6.50.
>
> Is this a reasonable performance for this setup or it could be improved
> considering the specifications?
>
> Thank you.
>
> Best regards,
> Petar Žuvela
>
> *Petar Žuvela**, Ph.D*
>
> Postdoctoral Research Fellow
> Room 0503, MD1 Building
> National University of Singapore
> 12 Science Drive 2
> 117549 Singapore
> Phone: +65 6516 5147 <+65%206516%205147>
> Mobile: +65 9039 9254 <+65%209039%209254>
>
> ------------------------------
>
> Important: This email is confidential and may be privileged. If you are
> not the intended recipient, please delete it and notify us immediately; you
> should not copy or use it for any purpose, nor disclose its contents to any
> other person. Thank you.
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at http://www.gromacs.org/
> Support/Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>


More information about the gromacs.org_gmx-users mailing list