Hi,
There is also an exclusive mode one can set using 'nvidia-smi'
-c, --compute-mode= Set MODE for compute applications:
0/DEFAULT, 1/EXCLUSIVE_THREAD,
2/PROHIBITED, 3/EXCLUSIVE_PROCESS
Best,
Andrew
On Mar 1, 2013, at 6:43 AM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
> Hi,
>
> I think you may be talking about setting the CUDA_VISIBLE_DEVICES
> environment variable. For example, if you have two CUDA GPUs on your
> system, they will by default be known as devices 0 and 1. To ensure
> your next job runs on the 2nd device, you would (in bash) type:
>
> export CUDA_VISIBLE_DEVICES=1
>
> More details are on this page: http://ambermd.org/gpus/
>
> -Dan
>
> On Fri, Mar 1, 2013 at 6:42 AM, Baldoni Hector Armando
> <hbaldoni.unsl.edu.ar> wrote:
>> Dear Amber users,
>>
>> I want to know if anyone remembers which CUDA flag should be setup to
>> prevent GPU overload, avoiding a second job execution in the same GPU.
>>
>> Many thank you.
>>
>> Hector.
>>
>> --
>> ---------------------------------------
>> Dr. Hector A. Baldoni
>>
>> Area de Quimica General e Inorganica (FQByF)
>> Instituto de Matematica Aplicada San Luis (IMASL)
>> Universidad Nacional de San Luis (UNSL)
>> Chacabuco 917
>> 5700 San Luis - Argentina
>> hbaldoni.unsl.edu.ar
>> ---------------------------------------
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
> --
> -------------------------
> Daniel R. Roe, PhD
> Department of Medicinal Chemistry
> University of Utah
> 30 South 2000 East, Room 201
> Salt Lake City, UT 84112-5820
> http://home.chpc.utah.edu/~cheatham/
> (801) 587-9652
> (801) 585-9119 (Fax)
>
> _______________________________________________
> 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 Fri Mar 01 2013 - 07:00:04 PST
