[Paraview] Sanity Check - Parallel GPU rendering

Philippe philippe.raven at gmail.com
Wed Apr 28 04:49:17 EDT 2010


On Wed, Apr 28, 2010 at 10:34 AM, Shree Kumar <shree.shree at gmail.com> wrote:

>
> On Wed, Apr 28, 2010 at 1:27 PM, Philippe <philippe.raven at gmail.com>wrote:
>
>>
>> There's another way to make *real* parallel GPU rendering.
>> If you want to use the power of some GPU from unrelated cards (non-SLI but
>> in the same computer) or from some graphical nodes, you can use VirtualGL
>> with Sort-Last or Sort-First rendering, who will be very usefull and
>> efficient to get very goods performances, and not just use 1 or 2 local GPU
>> to render your paraview instance.
>>
>>
> Could you explain this usage of VirtualGL a little more ?
>
> I am running ParaView render servers across multiple nodes, each of which
> may have one or more GPUs. Each server does local rendering on a GPU on the
> same node.  MPI takes care of job launching across multiple nodes. I am
> hoping this would give me the best possible performance, so someone please
> correct me if I am wrong :-)
>
> Thanks
> -- Shree
>
> Cheers.
>>
>> Philippe.
>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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
>>>
>>>
>>
>>  I think the Overview cat of the VirtualGL's documentation will me more
comprehensible than my english :
http://www.virtualgl.org/vgldoc/2_2evolving/#hd002

So this will able you to use as many as you want GPU for one rendering,
because mpi use only CPU, then I found a my test machine (core i7 920 with a
Quad-SLI of GTX295) that paraview got faster rendering if he's launched
with VirtualGL (rendering by GPU) than standalone (rendering by CPU). Maybe
my vision is not exact, but in fact, performances are very different ! Let's
give a try to VirtualGL Shree :)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20100428/44b053ba/attachment.htm>


More information about the ParaView mailing list