FW: [Paraview] Parallel ParaView- file formats and Chromium

Wylie, Brian bnwylie at sandia.gov
Tue, 23 Mar 2004 07:47:09 -0700


-----Original Message-----
From: Joel Welling [mailto:welling at stat.cmu.edu] 
Sent: Sunday, March 21, 2004 12:14 AM
To: Wylie, Brian
Cc: welling at psc.edu
Subject: Re: [Paraview] Parallel ParaView- file formats and Chromium



>Hi Brian;

>> Why would you want to use Chromium for compositing? ParaView has two
>> very good compositing methods. In fact if you compile with IceT and 
>> start up the client with --render-module=DeskTopRenderModule you will 
>> get the fastest software compositor in the world. :)  (I'm sure the 
>> flames will be extra hot on that statement....but it's Friday and I'm 
>> feeling cheeky)

> Well, fair enough- I don't mind if ParaView does the compositing.  I 
> do have
> to do the rendering via Chromium, though, because that's the only way to
get 
> the polygons to the graphics cards.  In principle there's no reason the 
> workers can't draw and read back via Chromium, then do the compositing.

There must have been a mis-communication somewhere. ParaView does hardware
rendering.

We've gotten 1 billion tris/sec with Ice-T on a 128 node cluster (that was
several years ago). We run ParaView on extremely large datasets on a daily
basis, works great and less filling. Also, the easiest way to run ParaView
is client server. You run the client on your desktop and do parallel
hardware renderering on the server (cluster of PCs).

I think a phone call might clear things up. Give me a ring 505-844-2238.

	Brian Wylie
	Sandia National Laboratories
	MS 0822 - Org 9227 - Building 880/A1-J
     (505) 844-2238   FAX (505) 845-0833