Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!lll-winken!sun-barr!newstop!sun!amdcad!dgcad!dg-rtp!addiction!brownr
From: brownr@addiction.rtp.dg.com (Randy Brown)
Newsgroups: comp.graphics
Subject: Re: IBM R6000 with 3D 24bit graphics--real time video?
Keywords: R6000, real time molecular modeling, Phigs
Message-ID: <1991May29.142421.13712@dg-rtp.dg.com>
Date: 29 May 91 14:24:21 GMT
References: <1991May26.142921.19574@batcomputer.tn.cornell.edu>
Sender: usenet@dg-rtp.dg.com (Usenet Administration)
Organization: DG-RTP Graphics Group
Lines: 29

In article <1991May26@batcomputer.tn.cornell.edu> cohen@quartz.ciw.edu writes:
>
>We are thinking of buying a 320H for solid modeling, in particular
>molecular modeling using Phigs.  The IBM specs for the 320 with
>24bit graphics are 10K Gouraud shaded polygons/sec.  Is this fast
>enough for real time video?  Does anyone have experience with such
>an application on an R6000?  Is it possible to save the frames
>and play them back at video speeds if that raw rate isn't fast enough?

Real-time at 30 frames/second gives you 300 polys/frame at best with a 10K
rated system.  Molecular modeling of any reasonable scale would have many
molecules, main chains, and assorted parts.  The UNC-CH graphics group
has been successfully performing molecular modeling tasks with a 30K poly/sec
machine, but does not use polys for spheres, which your machine probably
would.  How many polys per sphere will you use?  256, 128, 64?  You start
to run out of polygons pretty quickly at that rate.

So, I would expect that you will need to save frames, and perform single-frame
recording onto a VTR, unless you have a frame store like an Abekas to dump
them to tape with (which is a large investment for starting out).

>  Ron Cohen
>  Geophysical Laboratory
>  Carnegie Institution of Washington
>  Washington, D.C.

Good luck with it!

Randy
no new .sig yet