[Paraview] Sharing GPUs Amongst Processes

[Rakesh Hammond]

If I understand your question correctly and if PV can start its own xserver
process, then if you boot your compute nodes into level 3, it will not start
your xserver.

What I am not sure about is if PV can start its own xserver?

For example, in VisIt this is possible, and have done it.  We now have
machines with multiple GPUs (2 per node), so hoping that 2 of the four cores
on the machine can be used while the other 2 remains idle (for now).

I am happy to share our findings once we have it working (machine only got
installed yesterday :)

Rakesh
On 4 March 2010 07:44, Biddiscombe, John A. <biddisco at cscs.ch> wrote:

> Ricardo,
>
> Thanks for the useful info. We may opt to start an X server at bootup and
> then allow anyone connected to use the display. We wanted to avoid the
> xserver because on our existing system, slurm scripts are used to allocate a
> node, the script starts an X server for the user to run GL apps.
> Unfortunately on a newer system with many more cores, this leads to wasted
> resources as one user allocates the GPU and one core, the other 5 are
> inaccessible ....
>
> JB
>
> > -----Original Message-----
> > From: paraview-bounces at paraview.org [mailto:
> paraview-bounces at paraview.org]
> > On Behalf Of Ricardo Reis
> > Sent: 03 March 2010 23:24
> > To: paraview at paraview.org
> > Subject: Re: [Paraview] Sharing GPUs Amongst Processes
> >
> > On Wed, 24 Feb 2010, Biddiscombe, John A. wrote:
> >
> > > Question : Supposing you have no Xserver running but you wish to take
> > advantage of hardware acceleration. Is it possible to create an offscreen
> > buffer for each pvserver without the X server. We'd like to share one GPU
> > between 4 cores, but without starting X.
> >
> > maybe not what you want and maybe of some usefullness...
> >
> > In the xorg.conf you can put
> >
> > Section "Device"
> >      Identifier     "Device0"
> >      Driver         "nvidia"
> >      BusID          "PCI:07:00:00"
> >      Option         "UseDisplayDevice" "none"
> >      VendorName     "NVIDIA Corporation"
> > EndSection
> >
> > I always had to start an xserver (if you have more than 1 GPU you can add
> > device and screen sections has needed to xorg.conf). I've been using this
> to
> > render using Tesla cards.
> >
> > Off course this needs to start X...
> >
> > best,
> >
> >   Ricardo Reis
> >
> >   'Non Serviam'
> >
> >   PhD candidate @ Lasef
> >   Computational Fluid Dynamics, High Performance Computing, Turbulence
> >   http://www.lasef.ist.utl.pt
> >
> >   Cultural Instigator @ Rádio Zero
> >   http://www.radiozero.pt
> >
> >   Keep them Flying! Ajude a/help Aero Fénix!
> >
> >   http://www.aeronauta.com/aero.fenix
> >
> >   http://www.flickr.com/photos/rreis/
> >
> >                             < sent with alpine 2.00 >
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://www.paraview.org/mailman/listinfo/paraview
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20100304/c7c4a008/attachment.htm>