NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/arch/9397

Text File | 1992-09-13 | 3.4 KB | 71 lines

Newsgroups: comp.arch
Path: sparky!uunet!spool.mu.edu!agate!dog.ee.lbl.gov!news!nosc!crash!akix!aki
From: [e-mail] (Aki Atoji)
Subject: Re: Computer architects forced to change careers. NOT!
Organization: CxS Technologies
Date: Thu, 10 Sep 1992 21:19:43 GMT
Message-ID: <[e-mail]>
In-Reply-To: [e-mail]'s message of 9 Sep 92 21:33:19 GMT
References: <[e-mail]>
    <[e-mail]>
Sender: [e-mail] (Aki Atoji)
Lines: 57

In article <[e-mail]> [e-mail] (Bohdan Tashchuk) writes:

>>  The concept of reward for a job well done and punishment for an abject
>>  failure doesn't exist in many high-technology companies.
>>
>>  The 432 should be studied in all engineering schools as an example of
>>  how NOT to design and implement a computer architecture.
>>
>>  When screwups like this happen, a LOT of people get hurt in the
>>  resulting layoffs that endear the company to Wall Street.
>>
>>  The concepts behind RISC have helped bring some sanity back into the field.

I think the above carries the confusion between technical achievement
and market acceptance/financial sucess.

Having worked with different CPU's of varying successes in market
place (i.e.  NS32K, x86, 29K and 960), I would have to say that the so
called 'technological superiority' of an architecture is often *not*
the only single important item for more design wins.

There really are numerous factors in getting design wins and thus more
market acceptance, such as pricing, availability, marketing, product
positioning in market place, tools (compilers, emulators, etc.), how
many freebies (i.e. compilers, eval board, eval chips) the
salescreature will give to you, etc. etc (free lunches would count
too, hint, hint :).

I'm not familiar with 432.  It just could have been a 'killer' chip.
On the other hand, according to one of the articles in comp.sys.intel
(I only browsed through it), 432 having been developed to exploit (or
'exercise') the HMOS process (including tools needed to do the chip
design), it may have served it's purpose very well, becoming the
spring board for the next generations of design.  Some designs can
purely be done for research.  Some designs can be done for research,
with some hopes for market success.  Some designs might have been done
for market acceptance, but the company might be pretending it was done
for research if it didn't sell well.  I somehow don't think 432 was
the last one.

Some people still swear by 32K being the cleanest architecture ever.
It sure was easy to memorize the instruction set, but I don't think I
would have much employment opportunity for knowing this chip.

When 29K first came out, with it's MMU and three bus design, I'm sure
AMD's intent behind the design of the architecture was not to be one
of the most dominant RISC designs in the *embedded* market, which it is
now.  But then again, the number of 29K shipped is likely far less than
the 68K derivatives shipped (someone have actual numbers?).

However, these numbers are simply dwarfed by the number of 8051
derivatives shipped world wide.
--
---------------------------------------------------------------------------
Aki Atoji              Unix, X, Networking and Embedded Realtime Consulting
           [e-mail]                      crash!akix![e-mail]
---------------------------------------------------------------------------