> I hope you will extend your support to more platforms, beside Gnu/Linux.
> (OpenSolaris, BSDs etc)

It is too early to say anything about this. Testing software on clusters
of these would be a whole lot of effort - so anybody trying to use VizStack
on such systems will have to be involved.
 
> And why (oh why) you are stranded yourself only on Nvidia chips shore ?
>
> That is problem that seems to be intentionally put in inside underlying
> components.
> If you use OpenCL, you could use all comutational power and Including
> Nvidia and AMD (cheaper per speed) GPUs and other processing CPUs available.

That's because the developers use nVidia cards. nVidia drivers provide a 
consistent 
interface to all their GPUs, and we use that. The API provided by VizStack 
tries 
to be vendor-agnostic, and relies only on concepts rather than actual 
implementations 
(e.g. "GPU Combiner" rather than "SLI"). 

The tool that generates the X configuration files is where most of the nvidia 
specific logic resides. 

All said, I do not believe that VizStack should be limited to nVidia chips only.
If anybody is interested to add support for other cards (ATI, for example), I'd 
welcome that - and aid it too ! 

However, note that adding such support is a significant amount of work - 
VizStack
has a much more capable core than what it shows on the outside; we are hoping to
add support for more applications over time.

Regards
-- Shree

-----Original Message-----
From: Nikola M. [mailto:[email protected]] 
Sent: Thursday, May 20, 2010 2:56 PM
To: VirtualGL Users
Subject: Re: [VirtualGL-Users] Announcing VizStack

On 05/20/10 06:18 AM, Kumar, Shree wrote:
> I am pleased to announce that a new release of VizStack 
> (http://vizstack.sourceforge.net/) 
> is now available. VizStack can be installed on GPU clusters running Linux. 
> Each node can 
> have 1 or more GPUs; only nVidia GPUs work are supported for now.
>   
I hope you will extend your support to more platforms, beside Gnu/Linux.
(OpenSolaris, BSDs etc)

And why (oh why) you are stranded yourself only on Nvidia chips shore ?

That is problem that seems to be intentionally put in inside underlying
components.
If you use OpenCL, you could use all comutational power and Including
Nvidia and AMD (cheaper per speed) GPUs and other processing CPUs available.


------------------------------------------------------------------------------

_______________________________________________
VirtualGL-Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/virtualgl-users

------------------------------------------------------------------------------

_______________________________________________
VirtualGL-Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/virtualgl-users

Reply via email to