u-370 is not a recoded 68000 (was: IBM 5100)

From: Jim Battle <frustum_at_pacbell.net>
Date: Tue Aug 17 21:44:37 2004

Ken Seefried wrote:

> From: Christian Corti <cc_at_corti-net.de>
>> You can't change the microcode of the PALM processor. That would be like
>> changing the microcode of e.g. a MC68000.
> Actually....you can do that.
> The IBM {PC,XT,AT}/370 card(s) (ISA-bus card that put a "baby" IBM S/370
> on your desktop) had a CPU that was a re-microcoded 68000 (emulating
> some subset of the IBM S/370 mainframe instruction set). IBM licensed
> the architecture from Motorola. Probably got familiar with it building
> the IBM 9000 Instruments System.
> That, of course, doesn't mean you could do it on a PALM.
> Ken

I've heard that repeatedly, but I actually think it isn't true. I think
I know the real story.

One of the designers of the 68000 was a guy named Nick Tredennick. Ten
years or so ago I read a book that Tredennick wrote on his experience as
the architect of the micro-370. He kept extensive notes on the project
and details how the architecture evolved, the politics that came into
play, the near death experiences of the project, etc. His design was
influenced by his experience with the 68000 and took a lot of ideas from
it, but it simply was not a remicrocoded 68000.

Ah, the book is available on amazon (used). The title is
"Microprocessor Logic Design".

Nick Tredennick is a colorful guy. He always does a kind of roast at
the hotchips presentations every year at Stanford, and is always good
for quotes. He has always been a CISC booster and thought RISC was the
idea of lazy architects. In the back of his Microprocessor Logic Design
book he has a glossary, with more than a few funny definitions. For
example, under "Testability", the definition is something like "Circuits
that the test engineers make you put in that serve no useful purpose."
Or under "Methodology: stuck up way of saying Method. Methodology is
technically the study of methods." etc
Received on Tue Aug 17 2004 - 21:44:37 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:36:34 BST