Manta Interactive Ray Tracer Development Mailing List

Text archives Help


[Manta] Re: Re: vtk manta news


Chronological Thread 
  • From: David E DeMarle < >
  • To:
  • Subject: [Manta] Re: Re: vtk manta news
  • Date: Fri, 7 Jan 2011 18:17:25 -0500
40 million unique triangles per node, then binary swap depth composited. Thus I was strictly  doing ray casting. As memory serves the immediate bottleneck to going higher was insufficient RAM for the dynbvh build, which took about 20 minutes. The bottleneck to higher frame rates was depth compositing as memory serves.

Doc about vtkManta is here:

I'm supposed to write a white paper with some Intel folks on the scalability study, but I doubt that I'll ever get to that. I will post the code I used on github however. It is essentially ParaView 3.8 with the addition of a simple binary swap compositer, some batch driving scripts and some extra timing utilities.

David E DeMarle
Kitware, Inc.
R&D Engineer
28 Corporate Drive
Clifton Park, NY 12065-8662
Phone: 518-371-3971 x109


On Fri, Jan 7, 2011 at 5:09 PM, Thiago Ize < "> > wrote:
That's great news!

Is that just replicating the data across all nodes or is it doing compositing?  Any numbers (or documentation!) on the efficiency?

Thiago


David E DeMarle wrote:
Howdy,

I thought I should announce that we are going to put Manta in the official ParaView binaries for Mac and Linux.

In related news, last November I did a scaling test and rendered 10 billion polygons at just over 10 frames per second with pvManta. This was on an Intel cluster with 256 nodes and 16 processors per node.

Thanks everybody!

David E DeMarle
Kitware, Inc.
R&D Engineer
28 Corporate Drive
Clifton Park, NY 12065-8662
Phone: 518-371-3971 x109


Archive powered by MHonArc 2.6.16.

Top of page