[Paraview] Can pvserver run without X server?

Weiguang Guan guanw at rhpcs.mcmaster.ca
Wed Mar 14 12:52:30 EST 2007


Thanks for the instant responses.

I've already built Paraview with OSMesa and run it with offscreen 
rendering option on a cluster without X.

However, I want Paraview do the rendering using GPU rather CPU as 
rendering with GPU is about 10-20 times faster than with CPU. If DRI has 
been adapted to provide OpenGL acceleration without an X Server running 
(ref http://en.wikipedia.org/wiki/Direct_Rendering_Infrastructure), then 
it is possible (theoretically at least) to avoid X server and at the same 
time enjoy video hardware acceleration. Am I right?

Weiguang

On Wed, 14 Mar 2007, Moreland, Kenneth wrote:

> You should be able to use the OSMesa library to open a software-emulated
> OpenGL context without connecting to an X host.  If you compile with the
> OSMesa library and then use the offscreen command line option on the
> client, then you can render and composite on the server without an X
> host there.
>
> We have been doing that for years with ParaView 2.X.  If it does not
> work with ParaView 3, that would be a bug.
>
> -Ken
>
>> -----Original Message-----
>> From: paraview-bounces+kmorel=sandia.gov at paraview.org
> [mailto:paraview-
>> bounces+kmorel=sandia.gov at paraview.org] On Behalf Of Kent Eschenberg
>> Sent: Wednesday, March 14, 2007 11:33 AM
>> To: Weiguang Guan
>> Cc: ParaView
>> Subject: Re: [Paraview] Can pvserver run without X server?
>>
>> Weiguang Guan wrote:
>>> Has anyone run Paraview in client/server mode where the pvserver run
> on
>>> a computer (or cluster) without X server running? It seems to me
> that
>>> DRI can let an application access video hardware directly without
> going
>>> through X server.
>>
>> I think the answer is no. When composite is off the client does the
>> rendering and the graphics card has to be at the client. When
> composite
>> is on the server does the rendering but it only works through OpenGL
> or
>> through Mesa which emulates OpenGL.
>>
>> I think both must reach the graphics card through the X server. In
> other
>> words, the X server "owns" the graphics card and manages all access to
> it.
>>
>> The last system I know of that would do what you wanted was from SGI
>> using their GL language, the precursor to OpenGL.
>>
>> Kent
>> Pittsburgh Supercomputer Center
>> _______________________________________________
>> ParaView mailing list
>> ParaView at paraview.org
>> http://www.paraview.org/mailman/listinfo/paraview
>
>


More information about the ParaView mailing list