Path: utzoo!attcan!uunet!zaphod.mps.ohio-state.edu!mips!winchester!mash
From: mash@mips.COM (John Mashey)
Newsgroups: comp.sys.m88k
Subject: Re: Tektronix shutdown & move away from 88k's??
Message-ID: <42310@mips.mips.COM>
Date: 23 Oct 90 02:22:26 GMT
References: <1536@ftc.framentec.fr> <1990Oct19.120218.9450@canterbury.ac.nz> <15497@hydra.gatech.EDU> <2176@lupine.NCD.COM>
Sender: news@mips.COM
Reply-To: mash@mips.COM (John Mashey)
Organization: MIPS Computer Systems, Inc.
Lines: 170

Well, I guess I'm forced to respond, although I'd been trying
to keep out of this one.  I'm trying to keep things factual, but some of
this kind of has to verge on marketing-in-self-defense.

In article <2176@lupine.NCD.COM> rfg@NCD.COM (Ron Guilmette) writes:
><
><I think this is pretty close to the mark.  While I would not call the
><88k doomed, any more than I would call, say, the Intergraph Clipper
><doomed, or the AMD29000, I think it has missed the chance to gain any
><sizeable market penetration.   I think you'll find a trend in the
><industry toward clustering around 2 or 3 architectures (as we have
><really always done: 8080/z80 and 6502, 80x86 and 680x0, etc.).
><Unless something really strange happens, you'll see the SPARC and
><MIPS chips float to the top of the heap.  The rest of the pack is
><left to niche markets or oblivion.
I assume Ken meant "to the top of the heap of the mult-vendor-used RISCs",
since all of the other chips compared to were things used in multiple
vendors' systems.  (Hence, begging the question of what will go on
with RS/6000, HP PA, both of which seem likely to survive OK.)

>I believe that rumors of the 88k's demise are greatly exagerated.  So
>what if one marginal player drops out of the workstation business.
>(Tek is still in the X-terminal business after all.)  BFD.
Of course it is not the demise.  I do suspect that it means there is
little likelihood of 88K presence in the higher end of the technical
workstation market, leaving as possibilities mid-range (maybe, due to
DG, Omron), and multi-user commercial (seems more likely: Moto, DG, Norsk).
Also, some Telecom or Automotive applications remain a traditional Moto
strength.  Since some PC vendors have said they'll use SPARC or MIPS,
that leaves Apple as the key design win that the 88K really needs,
as almost every other big vendor has picked their RISC, if any.

>Anyway, even if (over time) that became true, MIPS would *not* be the
>first chip I though of as being (necessarily) the next in line for the
>#2 spot behind Sparc.  (And keep in mind that we are only talking about
>RISC chips now... as far as other chips, the 68xxx and 80x86 still have
>lots of milage left in them, and when it comes to instruction sets in
>general, I imagine that the 370 and VAX instruction sets will be around
>awhile yet).
As noted elsewise, I don't expect any of these other instruction sets
to go away.  As to who's #1 & who's #2:
	a) There are workstations, ranging from low to high.  Although
	I don't believe most market forecasts (if you compare numbers,
	it's very hard to tell who's right & who's wrong, even when it
	is possible to get good numbers).
	Sun is ahead in units, and at the low-end,
	whereas MIPS-based systems are pretty strong in mid-range to
	high-end. ((DEC, SONY, NEC. MIPS, Sumitomo Electric); (DEC, SGI, E&S)).
	All of the numbers I've ever seen show the total values
	to be much closer than the units, given the different emphases.
	(This makes perfect sense, of course, given what DEC, and especially
	SGI, and E&S do, although the mix is changing somewhat, especially
	as folks like Sony get ramped up. It is clear that Sun is a year
	ahead in going for the lowest price-points (kudoes to Andy B,
	of course), but the dust has yet to settle on this.
	IBM is coming in, more at high end than low end so far.
	Note that last year, Sun was 97+ % of the SPARC systems market,
	and hence, fairly easy to track.

	b) There are also servers, and multi-user machines, especially for
	commercial market.  Here, HP PA
	does pretty well, as does MIPS.  There is some Sun stuff here, but
	not much, and what makes it really confusing is that the Sun stuff
	sometimes seems to get counted twice, depending on which
	market analysis company you talk to.  ICL does some here with SPARC.
	Note that a lot of the MIPS
	stuff, surprisingly, is already in the commerical market, and in
	fact, if you count commercial applications, I think the MIPS
	software catalog has more commercial applications that Sun's.
	(Whether bean counts are everything, like Sun seems to believe, is
	not something that needs to be tackled in this note.)
	IBM is coming into this, as RS/6000s get sold as commercial
	multi-user machines, which is perhaps a little odd (given the
	ISA's emphasis on floating point), but not surprising.

	c) Then there are things that are really big systems, but not UNIX
	or general-purpose, i.e., like telephone switches.  here, I'd
	guess the RISC war is between MIPS (which has at least one public,
	and some more that aren't), and the 88K.  Little trace of SPARC.
	In lower parts, all kinds of things, such i960s, AMD 29Ks,  lots
	of 68Ks.

	d) Then there are things that really look like embedded control,
	like: laser printers, avionics, controllers, automobiles, etc.
	MIPS is actually doing OK in the (currently)
	small fraction of this market for which 32-bit micros make any sense.
	There are already plenty of chips in high-end laser printers
	(because of Adobe) and Canon Color Laser Copiers, and some of
	the more-dedicated embedded-control chips will be abel to compete
	for the lower-priced versions.  (BIG WAR IN LASER PRINTERS).
	You WILL see MIPS chips in HDTVs, DATs, maybe VCRs,
	and automobiles (I can't name
	names of course).  In any case, Motorola and Intel (960) will remain
	strong in this, and the new NSC 32K derivatives look interesting,
	and in general, there's room for lots more things.  Anyway, the
	message was: MIPS is actually OK in selected parts of this already.
	This area is important for volume reasons, of course, and some of the
	most successful chips out there have had both systems applications
	and embedded applications.  Little trace of SPARC here, so far,
	in anything actually shipping, except maybe, VME boards.

I don't know what the true metric is: chip unit volume, system unit volume,
or total value.  It's hard to tell what's happening unless you have these,
and it depends on who you are as to what you think is important.
(This is why it's hard to declare anything #1 or #2, even in systems.
Let me pick the market segment, and whether it's units or value,
and I can prove almost anything about who's ahead!)

>I personally don't know what the outcome of the RISC wars will be yet,
>but I think that it is a bit early to count Moto out.  Even if Moto was
>out, what about the i860?  What about HP PA?  What about the IBM RS/6000
>processor?  What data is there to support the notion that MIPS will beat
>out all of these to take the #2 slot?
As usual, it depends on what you're counting, and it's hard to tell.
I think the original thing was for chips used in lots of people's systems,
although I certainly don't expect HP PA and RS/6000 to go away.
So far the i860 seems mostly used as a non-reprogrammmable graphics unit.

It is the case that a fair number of serious companies ship MIPS-based systems,
already, in a wide range or markets.

>The last time I heard, MIPS was still a little (basically one product)
>company that was bleeding red ink.  Also, the last thing I heard about
>DEC's push in the (MIPS-based) RISC business was that they were really
>not selling very many DECstations at all.
>
>(If this posting doesn't draw a response from John Mashey, I'll eat my
>modem! :-)
Well, your modem is safe.  Of course, some facts would help here:
1) MIPS sells machines ranging from $9K desktops thru $150K servers,
and has cranked out quite a few system products in a few years.
We do have 700+ people, and have done >$100M so far in 1990.
(This is NOT big, of course, but it's not a little 1-product company.)
2) MIPS has designed both CMOS and ECL chipsets of various kinds,
and has all kinds of technology-license products, as well as lots of
software products, both of its own, and third-party.
MIPS third-party s/w catalog by itself is 700+ packages
(without counting mice, NCD X-terminals, VME extenders, etc, like at least
one other vendor does :-) of products one can actually order and get.
(This begs the question of bean-counting of applications, but does say
somebody out there msut beleive something is happening, as most of
these applications came on in the last 12 months.)

3) We just announced results for last quarter, and I'd hardly call it
bleeding red ink (slight profit), although life is certainly not easy out
there right now for almost everybody in this business.  As is well-known,
we are working hard with our colleagues at B.I.T. to improve yields on
the ECL chips.  It doesn't take much arithmetic to see what happens
when you have, for instance, 20 $150K computers you'd like to ship,
and each is missing one chip..... $3M takes a big byte from a quarter,
at our size.

>My own opinion is that (over time) the heir to the #2 spot will be some
>company with *both* deep pockets and a willingness to second source.  That
>most likely means either Moto or HP (although one never knows when, if ever,
>Intel will wake up and realize that second sourcing is Good and not Bad).

Well, do note that MIPS also has plenty of money in the bank, although
hardly in this league.  As I think another poster noted, the web of
partnerships, relationships, investors, etc, around MIPS is much bigger
than MIPS itself. 

Anyway, nothing is certain in this business, and many players have
various strengths in this.  However, the rules in this game keep
changing, and sometimes surprising huge established players...
-- 
-john mashey	DISCLAIMER: <generic disclaimer, I speak for me only, etc>
UUCP: 	 mash@mips.com OR {ames,decwrl,prls,pyramid}!mips!mash 
DDD:  	408-524-7015, 524-8253 or (main number) 408-720-1700
USPS: 	MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086