Re: Multi-node-gpu NAMD

From: Vermaas, Josh (vermaasj_at_msu.edu)
Date: Wed Apr 14 2021 - 15:41:07 CDT

What is the error it spits out when you try adding more devices? I can’t quite wrap my head around how the +p and ++ppn arguments are supposed to work together. Are you trying to run 4 processors per node, even if you are asking for 28 processors?

-Josh

From: <owner-namd-l_at_ks.uiuc.edu> on behalf of "Ropon-Palacios G." <biodano.geo_at_gmail.com>
Reply-To: "namd-l_at_ks.uiuc.edu" <namd-l_at_ks.uiuc.edu>, "Ropon-Palacios G." <biodano.geo_at_gmail.com>
Date: Wednesday, April 14, 2021 at 2:33 PM
To: KS UIUC <namd-l_at_ks.uiuc.edu>
Subject: namd-l: Multi-node-gpu NAMD

Dear user,

I’m try run NAMD on 4 node each with 2 GPU Tesla K80

My PBS file

#PBS -N test_nand_GPU
#PBS -l nodes=n004+n005+n006:ppn=28:gpus=2


cd $PBS_O_WORKDIR

/opt/shared/NAMD_2.13-multinode-gpu/charmrun /opt/shared/NAMD_2.13-multinode-gpu/namd2 +idlepoll +p 12 +devices 0 ++ppn 4 run_fixed/md_run.namd | tee md2.log

When set “+devices 0,1” not run, only run when set to 0 .

How can correct it?

Help me.

Geo.


This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:11 CST