cmcmanis@stpeter.Eng.Sun.COM (Chuck McManis) (06/27/91)
I saw a demonstration of some real time ray tracing on a massively parallel machine (I believe it was a connection machine with something like 4096 processors) anyway, ray tracing is well enough known that it is probably possible to build a dedicated ray tracing vector unit that costs less than these massively parallel boxes. -- --Chuck McManis Sun Microsystems uucp: {anywhere}!sun!cmcmanis BIX: <none> Internet: cmcmanis@Eng.Sun.COM These opinions are my own and no one elses, but you knew that didn't you. "I tell you this parrot is bleeding deceased!"
jet@karazm.math.uh.edu (J Eric Townsend) (06/27/91)
In article <15882@exodus.Eng.Sun.COM> cmcmanis@stpeter.Eng.Sun.COM (Chuck McManis) writes: >I saw a demonstration of some real time ray tracing on a massively parallel >machine (I believe it was a connection machine with something like 4096 >processors) anyway, ray tracing is well enough known that it is probably >possible to build a dedicated ray tracing vector unit that costs less >than these massively parallel boxes. Vector raytracing is tough. The problem is that there are all these conditionals mixed in your code. I read a how-to paper on vectorizing ray tracing a while back, I think it was (quick grep of ~/bib) oops, I can't find it. Take a look at the Ray Tracing abstracts, there's a copy for ftp from karazm.math.uh.edu. At any rate, an RS/6000 model 320 with 32Mb of ram is slightly slower than a SparcStation-2 w/24 Mb of ram on my ray tracer. Why? Because it never has a chance to cache and use multiple instructions. (If I do exhaustive ray tracing, it's even worse. :-) -- J. Eric Townsend - jet@uh.edu - bitnet: jet@UHOU - vox: (713) 749-2126 Skate UNIX! (curb fault: skater dumped) PowerGlove mailing list: glove-list-request@karazm.math.uh.edu