mwtilden@watmath.waterloo.edu (M.W.Tilden, Hardware) (06/07/90)
Third of an ongoing discussion about the Cyberdesk design. By the way, please post questions rather than clutter my mailbox if you want everyone to know your thots. Thanx. ---- erich writes: >By the way, I very much see your point in your latest letter to me. As to >the cost breakdown, I did see it on the last page. Minus the graphics >hardware, >I agree, it isn't bad. But as in all cases, the graphics hardware is the >problem with the price if you use standardized hardware (especially IRISes, >they are expensive!). Have you thought of the idea of working with custom >graphics hardware? There are relatively inexpensive, yet powerful graphics >coprocessors and such that could be integrated into central workstation >hardware (of course, you would need REALLY powerful, not just powerful). The >graphics system could be built in a modular, object-oriented fashion that could >make its implementation upon that platform would be doable. This is, of course, the big question. In my specs, I label the uP and associated graphics processors as SEP (Someone Elses Problem) because I realized that it was a topic which I was not expert in. Of course I read the latest electronic mags about the latest graphics devices but they all sound like Guinness World Record statistics to me ("RISC Graphic Engine Leaps Grand Canyon while Computing Googleplex Polygons a Second"). I wouldn't know the best engine to buy, and I'm damn sure that if it's available, it's probably obsolete anyway. If this thing takes off, hopefully it'll eventually dictate standards rather than just adapt to them. Personal IRIS's will only set you back $18,000 each (Cdn) and already feature seperate graphic engines with high-speed image exchange. SPARC 3's and 4's feature full Z-buffers, high speed and are the sexiest and smallest boards I've seen to date (about the same price). Both have scads of available software, run UNIX and their companies *could* be bullied into supporting such a venture as this. Thinking realistically about the first incarnation of the Cyberdesk, however, I had planned the CHUV screen to be composed of three or four adjacent NTSC style video screens. This is because I did not want to limit the versatility of the CHUV to just high-res computer screens. At the flick of a button, I wanted to be able to watch (multiple) TV images from a video-editting console to explore the potentials of high-speed video slice-and-dice. If you've ever tryed to put together a video of any length at all, you would realize how useful such a tool could be. The other reason was that the first projector TV's I had envisioned using (modifying) were the beautiful 86,000 pixel Sony 3.7" color TV's now available. Dismantling the one I have (verrry carefully. $800 of micro-box), I realized that they could be used as-is by just removing the diffussion plastic from behind the 2mm thick LCD screen and replacing the florecent bulb with a projector bulb at a safe distance. Cooling would be necessary, but I'm pretty sure it would work. Three or 4 of them could be easily mounted into the CHUV headrest (with full electronics including CATV receiver) with plenty of room left for the lense assemblies. Another reason is that I realized that an initial prototype using a IBM 386 and three fast video-boards would be a cheap way of testing the abilities of the Desk. Once the hardware was operational, the rest would be just software academics. Even if the uP was too slow for any significant computer demos, feeding syncronized video from several VCRs would give a good effect. IRISs also have a built in feature of generating real-time NTSC versions of their high-quality images, which is why I mentioned them as a potential uP. That's the plan. Future designers can attempt to mount those prodigious IRIS color monitors as projectors. I think, however, that the initial low cost and scope of the above implementation will be far more attractive to the average buyer. I know it'd be a damn site easier to build anyway. Is all. ---- Posted with permission. Is really all. -- Mark Tilden: _-_-_-__--__--_ /(glitch!) M.F.C.F Hardware Design Lab. -_-___ | \ /\/ U of Waterloo. Ont. Can, N2L-3G1 |__-_-_-| \/ (519) - 885 - 1211 ext.2454, "MY OPINIONS, YOU HEAR!? MINE! MINE! MINE! MINE! MINE! AH HAHAHAHAHAHAHAHAHA!!"