uri@arnor (07/19/89)
Hi guys! Some time ago Intel introduced iAPX-432 chip, which had a lot of goodies on board - object-oriented chip, I would say. Now the question. Is it being used by ANYBODY anywhere? It seems to me the chip never made the market. Can anybody explain why? I'm really curious. Maybe five years ago there was no need, but now it should fit perfectly into almost any application demand (well, I don't know its performance, but can it be that bad)? Feel free to respond by e-mail. Regarsd, Uri. (uunet!bywater!arnor!uri) ---------------------- <Standard Disclaimer> (but of course!) +++++++++++++++++++++++++++++++++++++++++++ "Don't do to others, what you don't want to be done to yourself". So - no flames, please.
farber@linc.cis.upenn.edu (David Farber) (07/20/89)
The iapx 432 was an experiemntal system which while leading the way for a letter system, was not commercially successful. However its decendant, the Biin gemini is an excellent system with many of the features of the 432 done better based on the experiences with the 432. I would strongly suggest looking at the Biin Gemini system as an excellent example of a modern object based fault tolerant design Dave David Farber; Prof. of CIS and EE, U of Penn, Philadelphia, PA 19104-6389 Tele: 215-898-9508(off); 215-274-8292 (home); FAX: 215-274-8192; Cellular: 302-740- 1198 "The fundamental principle of science, the definition almost, is this: the sole test of the validity of any idea is experiment." -- R. P. Feynman
jac@paul.rutgers.edu (Jonathan A. Chandross) (07/20/89)
uri@arnor > Some time ago Intel introduced iAPX-432 chip, which had a > lot of goodies on board - object-oriented chip, I would > say. More like terminal creaping featurism. > Now the question. Is it being used by ANYBODY anywhere? > It seems to me the chip never made the market. Can > anybody explain why? I'm really curious. Maybe five > years ago there was no need, but now it should fit > perfectly into almost any application demand (well, > I don't know its performance, but can it be that > bad)? This chip used copious quantites of microcode to emulate the virtual Ada machine. This wasn't enough to kill it on its own, but the fact that it was a 2 chip architecture (data path on 1 chip, uCode on the other), and that it ran about the speed of a Z-80 had a lot to do with it. Intel killed it off very quietly, just like the bubble memories. (Killing off the bubble memory division was a great pity since Intel made the most beautiful bubble memory controllers and interface chips I have ever seen. The corresponding Western Electric bubble chips required a whole board of nasty analog electronics to make them work, whereas you only needed 5 chips for the Intel approach.) I seem to recall a paper by Ed McCreight (sp?) from Xerox on the Dragon Processor (ASPLOS-II, 1987). They made a similar mistake with a multichip CPU. Jonathan A. Chandross Internet: jac@paul.rutgers.edu UUCP: rutgers!paul.rutgers.edu!jac
thatcher@claris.com (Jon Thatcher) (07/21/89)
jac@paul.rutgers.edu (Jonathan A. Chandross) writes: > Intel killed it [iAPX-432] off very quietly, just like the bubble memories. >(Killing off the bubble memory division was a great pity since Intel >made the most beautiful bubble memory controllers and interface chips >I have ever seen. The corresponding Western Electric bubble chips >required a whole board of nasty analog electronics to make them work, >whereas you only needed 5 chips for the Intel approach.) Intel ALMOST killed off their bubble memories. They sold that division to a company called MemTech (?), which rehired many ex-Intel engineers and scientists. They're still developing and selling bubble memories. --Jon (thatcher@claris.com)