eric@snark.uu.net (Eric S. Raymond) (11/30/89)
In <1989Nov28.104128.8045@hellgate.utah.edu> Nick Michell wrote: > On a related subject, DARPA has funded a number of GaAs RISC chips, > which have been reported in various conferences and acedemic publications. > This research has, at least so far, resulted in no commercial spin-offs. Ah, yes. Commodity GaAs -- the electronics version of "jam yesterday, jam tomorrow, but never jam today". Don't hold your breath. Persons Who Know have told me that GaAs is a real revolving bitch to design for (as in "worse than bubble memory"). Also that the low density of GaAs is unlikely to improve much; there's some physical effect connected to its high electron mobility that makes packing gates very tightly a Bad Idea. Finally, the reagents used for production are intensely toxic and much more persistent in the environment than the garden-variety hydrofluoric acid used as silicon etchant. If you're the gambling type, bet your bux on ballistic-transistor technology or indium phosphide or even nanotechnology rod logic. But forget GaAs. It is almost certainly doomed to remain a niche technology funded by organizations that don't care how much of your money they spend for their fun. -- Eric S. Raymond = eric@snark.uu.net (mad mastermind of TMN-Netnews)
swarren@eugene.uucp (Steve Warren) (11/30/89)
In article <1Tcfjq#9jMTbv=eric@snark.uu.net> eric@snark.uu.net (Eric S. Raymond) writes: [...] >or indium phosphide or even nanotechnology rod logic. But forget GaAs. It is >almost certainly doomed to remain a niche technology funded by organizations [...] ;^) --Steve ------------------------------------------------------------------------- {uunet,sun}!convex!swarren; swarren@convex.COM
patrick@convexc.uucp (Patrick F. McGehearty) (11/30/89)
In article <1Tcfjq#9jMTbv=eric@snark.uu.net> eric@snark.uu.net (Eric S. Raymond) writes: >...deleted stuff about why GaAs is hard to work with... > >If you're the gambling type, bet your bux on ballistic-transistor technology >or indium phosphide or even nanotechnology rod logic. But forget GaAs. It is >almost certainly doomed to remain a niche technology funded by organizations >that don't care how much of your money they spend for their fun. >-- > Eric S. Raymond = eric@snark.uu.net (mad mastermind of TMN-Netnews) Funny, Convex is currently shipping machines with some GaAs parts in them. The parts are made by Vitesse, and are pin compatible with ECL parts that have similar characteristics, except the GaAs parts are cooler and slightly faster. I expect that designers find it no more difficult to work with than the ECL parts. We can built product with either type of part, and you can be sure that we don't stay in business by wasting money. Difficulties in manufacture and use of new technology have always been pointed to as reasons not to move into the future. GaAs has its weaknesses as well as strengths, but I expect many more GaAs based products to come to market years before the other exotic technologies that Eric Raymond mentions. One or more of them will also make it in later years, but commercial computer architects must use technology in immediate reach, not the generation after next. Of course, if one is too conservative, one is left at the starting gate. With the rapidly changing technology base, technology selection is more of an art than a science.
swarren@eugene.uucp (Steve Warren) (12/01/89)
In article <3509@convex.UUCP> swarren@convex.COM (Steve Warren) writes: =In article <1Tcfjq#9jMTbv=eric@snark.uu.net> eric@snark.uu.net (Eric S. Raymond) writes: = [...] =>or indium phosphide or even nanotechnology rod logic. But forget GaAs. It is =>almost certainly doomed to remain a niche technology funded by organizations = [...] = =;^) = =--Steve =------------------------------------------------------------------------- = {uunet,sun}!convex!swarren; swarren@convex.COM In case anyone is confused over this terse post, I was amused because it is public knowledge that Convex has machines currently in production that include GaAs gate arrays, as Patrick McGehearty also noted. --Steve ------------------------------------------------------------------------- {uunet,sun}!convex!swarren; swarren@convex.COM
hui@joplin.mpr.ca (Michael Hui) (12/01/89)
GaAs cannot be a dead end. It is used quite a bit in microwave circuits, and now you can get industry standard pin-out and function PALs from Gazelle in GaAs. In telecom, especially in TDM digital fiber transmission products, the serial<->parallel converters usually are done using GaAs chips. There are alternatives, of course, but currently those alternatives (in silicon) are _less_ mature than GaAs.