Nemnich%mit-multics@sri-unix.UUCP (06/15/83)
From: Bruce Nemnich <Nemnich@mit-multics> The June issue of Byte contains the first article of a three-part series titled ``The 8086 -- An Architecture for the Future,'' written by someone from Intel (naturally). It seemed anachronistic after the issue (April, I think) which hightlighted the 680x0, 160xx, and iAPX x86 families. --bjn
freund@nsc.UUCP (06/19/83)
I suppose that even the dinosaurs in the La Brea Tar Pits claimed that all was well. After all, the stuff was supposed to be good for the complexion. Bob Freund
taylor@sdcsvax.UUCP (06/20/83)
Silly sort of comment, but the iAPX86 is the system based ON the 8086 chip, so saying that the iAPX86 is better (or worse) than the 8086 makes a very little amount of sense! I like Intel (tm) products....the iAPX432 system is for me!!! (as soon as they get it on 1 chip!!!) -- Dave Taylor
Nemnich%mit-multics@sri-unix.UUCP (06/21/83)
From: Bruce Nemnich <Nemnich@mit-multics> I did not mean to say the 8086 is better or worse than the iAPX86; indeed, that is silly. The April issue had an article on the 186 and the 286, the latter of which, to my understanding, greatly expands the 8086 architecture. The June article doesn't even mention the existance of the iAPX series, or any other 16-bit architecture. It seems to view the 8086 as a great new development, and it supports its claim by demonstrating how much better it is than the first- and second-generation 8-bit chips. In light of the MC68000 and NS16000 series, that seems anachronistic. I don't wish to put down Intel; the 8086 *was* a big step forward. It just seemed to me the 8086 article was a leftover from several years ago. --bjn
NSIN08%rlgk@BRL.ARPA (06/22/83)
From: Philip Gladstone (on GEC 4090 at Rutherford) <NSIN08%rlgk@BRL.ARPA> The snag with the iAPX432 is that it is MIND-BOGGLINGLY slow - and adding more processors does not really cure the problem as you run out of memory bandwidth. I guess bigger caches on the CPU (GDP?) chip would help. Philip Gladstone
guy@rlgvax.UUCP (06/23/83)
The 80186 and 80286 include some new instructions for procedure calls and the like; the 80286 includes an on-chip MMU and instructions to handle it. Other than that, there are no architectural differences between the 80x86 machines - still only 4 16-bit registers, still a preference for 16-bit pointers. Guy Harris RLG Corporation {seismo,mcnc,we13,brl-bmd,allegra}!rlgvax!guy