Hi Thomas,
This is unlikely to be released as a patch - it will be in the next major
version of AMBER. In the meantime just do the following:
cd job1
export CUDA_VISIBLE_DEVICES=0
$AMBERHOME/bin/pmemd.cuda -O .. .. .. &
cd ../job2
export CUDA_VISIBLE_DEVICES=1
$AMBERHOME/bin/pmemd.cuda -O .. .. .. &
wait
This will utilized both GPUs, one for each job and the job submission
script will not quit until BOTH jobs have completed. This works well if
the two runs are expected to take similar times to complete.
All the best
Ross
On 4/10/13 3:10 AM, "Thomas Evangelidis" <tevang3.gmail.com> wrote:
>Dear Amber developers,
>
>Is there going to be any patch soon to allow support for extra point
>calculations with pmemd.cuda.MPI? I am running several simulations with
>TIP4P-Ew water model using pmemd.cuda and receive complaints by the
>supercomputer administrators for not utilizing both GPUs per node.
>
>Thomas
>
>
>--
>
>======================================================================
>
>Thomas Evangelidis
>
>PhD student
>University of Athens
>Faculty of Pharmacy
>Department of Pharmaceutical Chemistry
>Panepistimioupoli-Zografou
>157 71 Athens
>GREECE
>
>email: tevang.pharm.uoa.gr
>
> tevang3.gmail.com
>
>
>website: https://sites.google.com/site/thomasevangelidishomepage/
>_______________________________________________
>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
Received on Wed Apr 10 2013 - 07:00:03 PDT
