Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!jarthur!nntp-server.caltech.edu!mustang!data.nas.nasa.gov!wk209!yamo
From: yamo@nas.nasa.gov (Michael Yamasaki)
Newsgroups: comp.graphics
Subject: Re: Rendering performance
Message-ID: <1991May8.002428.21810@nas.nasa.gov>
Date: 8 May 91 00:24:28 GMT
References: <1991Apr29.162712.1905@canon.co.uk> <1991May3.190648.13574@ucunix.san.uc.edu> <1991May3.224900.12807@dsd.es.com> <1991May6.114747.12681@canon.co.uk> <1991May6.234927.17724@dsd.es.com> <1991May7.090203.18257@canon.co.uk> <1991May7.094641.18738@canon.co.uk>
Sender: news@nas.nasa.gov
Organization: NAS Program, NASA Ames Research Center, Moffett Field, CA
Lines: 42

Greetings. 

In article <1991May7.094641.18738@canon.co.uk> ads@canon.co.uk (Adam Billyard) writes:
>
>There must be someone who reads comp.graphics that has actually
>implemented a 3D modelling and rendering system.  What sort of
>performance do you get? Is it minutes or milleseconds to render 20k
>polygon scenes?
>

Okay. I'll bite. I'm doing visualization of unsteady fluid flow... CFD stuff.
I have a 4D/320VGX.  I threw some timers in my code and this is what I got.

The grid surface is a 64 x 64 array of verticies in 3D. 
GL calls to qstrip. Count each "quad" as 2 triangles.
Gouraud shaded, not lit. Z-buffered. Timed for 100 timesteps. 
Triangle size varies from a few pixels to thousands of pixels.

# of grid surfaces     tris total   tris/second     frames/second  tris/frame

         1              793,800       157,188           19.8          7938
         2            1,586,700       188,444           11.9         15876
         3            1,984,500       156,630            7.9         23814
         3            2,381,400       141,497            5.9         31752

Great. (so all together now) So What?

If this helps you to understand that a VGX is the appropriate thing for
your particular application, I'm surprised.  It's real world but hardly
repeatable.  It doesn't use the million other things the VGX can do to
make it look wizzy (on the todo list). 

I think its worth the money the government payed to put it on my desk. But
when I need real number crunching I "go to" (distributed processing)
the Crays that we have.

(I wouldn't buy either machine (VGX or Cray YMP) with my own money ;-)

                                       -Yamo-
                                 yamo@wk209.nas.nasa.gov
                                yamo@amelia.nas.nasa.gov
                     {ncar, decwrl, hplabs, uunet}!ames!amelia!yamo