freeman@spar.UUCP (Jay Freeman) (06/13/85)
[] >I find the discussion at least a little informative. It's so hard to keep facts from leaking through when you're trying to flame. >I like seeing people who seem to know discuss the rational behind the design >for the 8086. I conjecture that the major motivation was to hit the market soonest with a microprocessor that was a healthy step up in performance over the "classic" eight-bit-data, sixteen-bit-address designs: It would appear that segments add address-space without requiring wider registers and data-paths, while "special-purpose" registers add at least some processing power without using too much silicon. And less silicon leads to lower cost and earlier production. Of course, the primary issue for this newsgroup has to do with the abstract merits of the various architectures, rather than the positions in time and in the marketplace of the various specific microprocessors. But the issue of how best to make use of resources -- in this case square microns of silicon -- is certainly an important one. And the fact that there can be a case made at all, that the 27000-transistor 8086 is in the same class as the 68000-transistor 68000, is an enormous compliment to the ingenuity of the 8086 design. >These notes are clearly marked by the subject line. When I get bored of this >discussion I'll stop reading them. Then we can start debating the merits of the 432. >I say: keep them here where we can keep an eye on them. Are you sure you want to know? :-) -- Jay Reynolds Freeman (Schlumberger Palo Alto Research)(canonical disclaimer)
jer@peora.UUCP (J. Eric Roskos) (06/17/85)
[The referenced article speculates on why the 8086 might have been designed
the way it was.]
This brings up an interesting point, one I've seen little data on. I've
heard it said that a lot of the reason why the 8086 was designed the
way it was was that, in the early days, there was an assembler for the 8086
that took 8085 assembler source and generated 8086 object code.
Yet the only evidence I've ever seen that this was actually the case is in
the listings provided with a popular word processor that show how to
customize it; these indeed use 8085 register names in place of the 8086
ones (although it clearly is generating code for an 8086).
Thus perhaps the properties of the 8086 are partly the result of this
attempt at upward compatibility. On the other hand, as far as I can tell
the 68000 designers did not make such an attempt, to make the 68000 upward
compatible with the 6800 (or 6809). (It's a good thing they didn't try
to make it compatible with the 6809! The 6809 was starting to take on some
of the irregularities of the 8086 in terms of the difficulty of generating
object code for it.)
--
Full-Name: J. Eric Roskos
UUCP: ..!{decvax,ucbvax,ihnp4}!vax135!petsd!peora!jer
US Mail: MS 795; Perkin-Elmer SDC;
2486 Sand Lake Road, Orlando, FL 32809-7642
"Gnyx gb gur fhayvtug, pnyyre..."henry@utzoo.UUCP (Henry Spencer) (06/18/85)
> Thus perhaps the properties of the 8086 are partly the result of this > attempt at upward compatibility [with the 8080/8085] ... What "partly"?!? The properties of the 8086 are almost entirely the result of trying to make the thing upward compatible with the 8085 (which was an upward-compatible upgrade of the 8080, which was an upward- compatible upgrade of the 8008). And the rest of them are the result of the decision -- for reasons of chip area and inertia -- to forget about a 32-bit architecture and go for 16 (whence the deranged MMU design, to try to squeeze a few more address bits into a 16-bit architecture). -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry