From: cit-vax!harry.ugcs.caltech.edu!andrey@elroy.Jpl.Nasa.Gov (Andre Yew)
Subject: Re: VR: Low-End 3D -- Big in U.K., Why Not the U.S.?
Date: Tue, 15 Oct 1991 18:00:05 GMT
Organization: California Institute of Technology, Pasadena



rodent@netcom.com (Ben Discoe) writes:

>Our brightest minds get snatched up by Institutes of Technology which set
>them down at $500,000 workstations and let them make "pretty pictures"
>requiring countless MIPS and often hours or days.  No one in the U.S.
>considers polygon graphics a software problem - they assume that it
>will be done in hardware (a la Silicon Graphics) meaning that personal
>computer users are out of luck.

	I beg to differ.  One, SGI machines achieve graphics performance that
are way beyond the hopes of any PC and have lots of software available already.
For people who want to get work done and not reinvent the wheel, they are
fine machines to work on.  Physicists, chemists, and other researchers don't
have time and/or the skill to write their own graphics code.  Two, here, at
an "Institute of Technology", James Blinn uses his own code on PC-clones and
a TIGA board to do his animations.  The system is functional, but Blinn put
lots of years into programming it to get it to where it is.  In class a week
or two ago, Blinn said something like, "I'm a graphics designer" implying
that after working out his software for years, the end result is that he
uses it to do real work, not see how many pixels he can blast to the screen
per second.

>> You might imagine that I was
>>shocked to see someone suggest using C to write their 3D routines.

>  Of course, it IS a good idea to develop in C first for ease of debugging,
>then re-code in assembly when everything is known to work.

	In case anyone's curious, Blinn programmed his system with a mix
of Fortran and assembly language.  He is trying to learn C++ though.

	Also, IMO, you shouldn't recode everything in assembly.  Profile your
code and optimize the part that gets used 85% of the time and not the parts
that take 5%.

>Ben Discoe, radical ecologist, computer scientist, geometer, amigoid

						Andre

--
Andre Yew                     andrey@through.ugcs.caltech.edu (131.215.131.169)