Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!decwrl!ucbvax!VIOLET.BERKELEY.EDU!mwm
From: mwm@VIOLET.BERKELEY.EDU (Mike  Meyer, My watch has windows)
Newsgroups: comp.sys.amiga
Subject: CBM and upgrade paths
Message-ID: <8711040542.AA29099@violet.berkeley.edu>
Date: Wed, 4-Nov-87 00:42:41 EST
Article-I.D.: violet.8711040542.AA29099
Posted: Wed Nov  4 00:42:41 1987
Date-Received: Sat, 7-Nov-87 09:02:21 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Lines: 458

Yet another round of bitching at CBM about their total lack of an
upgrade path.

Let's get to the important thing first. I claimed that a posting from
a CBM employee asking people to contemplate what the A1000->A2000
upgrade deal means was scary, because so contemplating leads to the
conclusion that CBM is going to do something *truly* nasty to A1000
owners.

That was weeks ago. I've gotten private mail from CBM employees, and
most of the CBM names seen on the net have replied.

None - NOT ONE - has said that no such thing was planned. They haven't
even used the cop-out "To the best of my knowledge, no such thing is
planned." Any of you CBM people care to say that? Better yet, you want
to provide an "insurance" bet on such?

The best that came from them was when I postulated a specific series
of events. Dave said "it can't happen." The key point was:

<< Any hardware upgrade would have to happen to ALL systems out
<< there if it were to be made a viable required upgrade.

I agree, it should be viable on all systems. But tell me about the KS
1.2 that's now required for new software, or the "optional" 256K of
ram (that is required for the blasted TUTORIAL shipped by CBM!!!).
Given that track record, what assurance do we have that CBM won't do
it again, this time with something that won't run on an A1000?

Most of the replies from CBM on the net were just "Shut up" in more or
less (mostly more) polite tones. I got one piece of e-mail claiming
that my posting had internal inconsistencies. I said "point them out
to me" and am still waiting for a reply.

On the other hand, when I complained about DECs (more later) bus
change, the first thing they said (right after "Not so LOUD! You're
hurting my ears!" :-) was "We think you're wrong. Tell us what you
think is broken, and we'll attempt to fix it, or explain why we
won't."

As I said before, DEC is doing a *much* nastier job of screwing
customers with old hardware, but they are at least willing to *try*
and make things less painfull. CBM is trying to pretend that the
customers with old hardware don't exist, with the exception of trying
to turn them into customers with new hardware.

But let's look at some posted examples of what other manufacturers are
doing:

<< IBM changed from the PC bus to the AT bus and announced that most of
<<   the PC cards would not be supported in the AT.

One case so far of a company acting as CMB has acted.

<< IBM dumped the PC/AT bus, an industry defacto standard for the
<< micro-channel.

Not a valid case - IBM doesn't dominate the IBM-PC clone market the
way CBM dominates the Amiga market. People are already dumping Zorro I
products for Zorro ][. The MCA has as yet to prove that it's going to
survive.

<< IBM constantly announces new products with only enough backward
<< compatibility to allow exiting users to upgrade before it's too late.

This must be the micro market. The mainframe market tends to have
software hang around for decades. The latest/greatest systems now
support 30 bit addressing - but still run software that uses the upper
8 bits for other things. So, if the micro market acts like that,
that's two cases - both by IBM.

<< TI walked away from away from the home computer market entirely, turning a
<<   computer that was a slow as beans, but had a real operating system and
<<   real expansion capability into scrap metal and plastic.
<< TI dumped their low-end minicomputer compatible office systems in favor
<<   of incompatible PC compatibles.
<< Apple canned the Apple III completely.
<< Apple canned the Lisa and had nothing to offer but the Macintosh, only
<<   the MAC II finally comes close in features and market placement.

These are cases of a company dropping a machine completely. It's not
the same thing, and I'll have more to say about that later.

<< Radio Shack changes models every year or so with compatibility a sometimes
<< sort of affair,

But has RS quit carrying the expansion products for the old machines?
In fact, as long as you stayed inside of a "line" (the z80 model toys,
the z80/68K model business machines, and the IBM-PC clones), RS has
only one compatability boondogle that I know of. They made the mistake
of producing a machine that was superior to, but incompatible with,
the IBM-PC.

<< SUN changed their expansion bus between the SUN-2 and SUN-3, did they change
<<  it between the SUN-1 and SUN-2 also?

They didn't change it from the SUN-1 to the SUN-2. And there are ways
to make the old cards work in the SUN-3.

<<SUN has decided that some SUN-3 peripherals will not be supported on SUN-4
<<  configurations, even though you can plug them in and the software supports
<<  them.

Note, _some_ aren't supported. That means you have to run them at your
own risk. Note also that some _are_ supported, even though Sun changed
processors in going from the 3 to the 4. Contrast this to CBMs
complete lack of support for SOTS and Zorro I in the A2000.

<<DEC had three incompatible I/O busses on the PDP-8 family, two or three on
<<  the PDP-11, and god knows how many on the VAX family.

I never dealt with PDP-8s, so...  The PDP-11 and VAX expansion bus
class will now come to order. Skip to the "class dismissed" string if
you aren't interested in the gory details

Yup, the PDP-11 line has three expansion busses. The micro-11 line
used the q-bus, and only the q-bus. This is also what you find in the
microvax line. In fact, you can turn your micro-11 into a micro-vax by
plugging in the microvax CPU card and installing a VAX OS.  I'll note
that there's not much way to move things from the micro* machines to
larger machines, and leave that line of DEC products. From one on,
when I say "PDP-11," I mean the non-micro 11s. Ditto for "VAX."

The smaller PDP-11s (everything but the /70, the never-distributed
/80, and maybe the /60) only had one IO bus, the unibus. Some of them
(the /45, for instance) had two unibuses.

When DEC announced the /70, the put a new, faster bus with more
address bits in it, called the "massbus." They also put had places to
hang unibus adapters in the machine, so you could use unibus
peripherals in the machine. In fact, that was really the only place to
hang serial cards and some others.

Now, the 11s did have a seperate "bus" that the CPU cards and main
memory hung off of. It corresponded to the motherboard in an A1000/C64
type machine. You could put more memory on it, and that was about it.

Then came the Virtual Address eXtension 11, the PDP/VAX-11/780. It had
the same basic architechture as the PDP-11/70 - a bus for CPU cards
and memory cards, plus one or more unibus and massbuss adapters. In
other words, you could yank the boards from your 11, roll out the CPU,
roll in a VAX cpu, plug the new cards in, and go. It would even run
PDP-11 code, if you wanted it to. Of course, DEC didn't support some
cards in the new machine. But people ran them anyway, as they were
better than what DEC would support.

The 750, 785, and 782 all followed this pattern. The 730 had places on
the backplane to plug things, and is best ignored as the abortion it is.

The 8600 and 8650 changed this only slightly, in that you could plug
two 780-like backplanes into the real backplane. This let you get
twice the IO bandwidth from the machine. But note that you could move
unibus cards from an old, cantankerous 11/45 into an
brand-spanking-new 8650, and they would work. In fact, I did that,
except the cards stopped in 11/70s for a decade or so along the way.

With the 8600, DEC also announced the "VAX Cluster" concept. A VAX
Cluster is basically a Sun cluster done right (details upon demand).
The IO "bus" that supports this is the CI, or Computer Interconnect.
It's more of a 70 MB/sec ethernet than a bus. You hang VAXen and HSCs
off of it. An HSC is nothing more than a file server that does nothing
but serve files - no OS needed.  HSCs speak to standard DEC devices -
just move the cables.

This is called an upgrade path. It's what CBM hasn't provided between
the A1000 and the A2000.

Now, you get to the latest series of DEC machines - the
82??/83??/85??/87??/88??. These have a brand new bus, called the BI
(Backplane Interconnect). You can hang CI connecters, massbus
connecters and unibus connecters (though you can't run a unibus flat
out on it) off of them. This is the bus that DEC started getting nasty
with. No serial devices for it (buy LATs), it's proprietary, etc, etc,
etc. But you can *still* take cards from the old machines, and have
them work. Just don't expect them to work well.

[class dismissed]

Now, let's see what people have to say about the new CBM machines:

<< Still, a company that publishes a buss standard, changes it
<< while the machine is still being sold, and then replaces it for the
<< next upgrade, obsoleting all the add on hardware (yours and mine
<< included) DOES have a "game machine" mentality.  Responsible companies
<< don't do this kind of stuff to their users.

<< Considering [...] Commodore's own history of introducing
<< incompatible machines, 

Of course, I've found out that CBM isn't shipping a serious memory
card yet. Just the little 2Meg jobs using 256Kb chips. Filling the
available address space means filling all but one slot of the bus. Of
course, ASDG will sell you a serious memory card. The same one they
sell for the Zorro I, and have been seling for months.

Now add:

<< However, Commodore hasn't been able to ship us the bridgeboard which
<< we ordered over nine weeks ago, although we were assured the day
<< before making the order that "bridgeboards are in stock".

And you arrive at the conclusion that the "easy expandability" is more
hype than fact. At least until the A2000 market catches up with the
A1000 market.

And from "the B2000 guy" himself, we hear:

<< So you don't REALLY have to upgrade to stay in the Amiga family and
<< use nearly every software product you can expect for this generation.
       ^^^^^^ ^^^^^

Ah, so you won't be able to run *all* A2000/A500 software on the
A1000? Neat, really neat. Of course, it's only natural for a company
to junk an old software standard after junking an old hardware
standard.

In other words, the only company from that nice long list to behave as
badly as CBM has towards owners of their old hardware is IBM. Acting
like IBM is one of the best reasons I can think of to stop dealing
with a company.

As I said before, an upgrade path would have been easy: just put an
A1000 SOTS slot on the A2000. The replies are:

<< If the A2000 accepted SOTS, we'd have all the same A1000 problems all over
<< again.

But only the people who already had SOTS boxes would have those
problems. At least they would if CBM went to some pains to make sure
that people understood the purpose of SOTS. Then again, expecting CBM
to take pains to prevent their customers pain is probably unreasonable
- as witnessed by the pains they are taking to make sure that A500
users don't buy SOTS expansion for their machine.

But we also get:

<< SOTS is a real bad idea, and never should have even been mentioned in the
<< Amiga literature.

<< I think the time to drop SOTS was before SOTS existed. 

Of course, this is in the same message as:

<< Want to run any SOTS box for the A1000 on the A500, add a power supply
<< and a short extender.

Or: "SOTS is lousy, and we should never have done it. But we do have a
brand new SOTS machine if you really want it." Unfortunately, the SOTS
interface is on the *wrong bloody machine*.

Then there's this alternative:

<< Imagine if Commodore had released an incompatible machine, like,
<< one that used the 386 instead of the 68000.

<< It could be worse.  C= could have died.  They could have come out with
<< an incompatible machine.  They didn't.

In other words, if they'd given up on the Amiga. Well, I *know* what
the affect of what they are doing now is: the A1000 peripherals market
is drying up as people abandon it for the A2000 market, and we can
expect to see Amiga software that won't run on the A1000. The second
of these is trivial; the first shortens the usuable life of my machine.

On the other hand, suppose that CBM had walked away from the Amiga.
There'd probably be less new software. On the other hand, the people
producing peripherals for it almost certainly wouldn't have abandoned
it as quickly as they did. The first is (once again) trivial. The
second is the mirror of the above - it signifies a shortening of the
usefull life of the machine.

When I set out to buy a machine, I *don't* try to buy something
completely compatable with what I had before. I set out to buy the
most machine I can get for the dollar that will do the job I want it
to do. That's how I selected the Amiga then.

I had planned on hanging onto my A1000 for about four years. At that
point, I'd probably buy a '386 machine because they're going to be
delivering (actually, already are delivering) a lot more bang for the
buck than anything CBM sold, and will be able to do the job I want the
machine to do.

Except now CBM has cut the feet out from under the A1000 peripheral
market. So I either need to upgrade to the new A2000 now, or wait
until I need something that's not on the market anymore, and then buy
a new computer.

Only CBM treated owners of old expansion hardware the same way Reagan
treats the poor. No reasonable way to use it on the new boxes. Guess
how many alternatives that leaves? And then consider the first few
paragraphs in this article, and notice how many paddles I've got to
get out of the creek CBM left me up.

To summarize the above, I would have been better off if CBM had
abandoned the Amiga, rather than introduce a new machine that's has
enough compatibility to steal the market from the old one.

Now, let's deal with some random comments:

<< Nope, Commodore decides to throw it's weight behind the Amiga, makes another
<< big $investment$ and introduces two new products intended to correct the
<< perceived market limitations of the original.  One retains the features that
<< made the A1000 attractive to the recreational users, but through tradeoffs
<< like the integral keyboard and separate power supply, can be priced low
<< enough to appeal to a broader user community, and can be sold without
<< complicated technical explanations of why it's really twice as good as the
<< cheaper Atari machines.  The other model adds features intended to address
<< the concerns of the professional or business user who sees the computer as a
<< tool or a means to an end.

Of course, there's no upgrade path from the cheaper of the new
machines to the more expensive, or from the old machine to *either* of
the new ones.

<< Now I suppose it's natural for someone bought an A1000 to argue that this
<< was the perfect machine, since it's the one they decided to buy.

No, it wasn't perfect. I spent to much time fighting SOTS, GURUs and
the like to know that. On the other hand, it was a long sight better
ergonomically than either of it's replacments.

I didn't even think that:

<< the Amiga [...] is the coolest home computer by far ever.

I've publicly stated that there were two home computers I'd seriously
consider trading my Amiga for (but both retail for significantly more
$'s), and I've continually pointed out that the Amiga didn't introduce
multitasking to the home computer market. I've even been pointing out
that I can buy a cheaper multi-tasking, windowing system than an A500.

The Amiga doesn't have any *single* feature that makes it a real win.
It just has a large collection of *right* features.

<< Some even like the paint the people in West Chester as either evil
<< beings or dullards who don't understand the difference between an
<< Amiga and an IBM PC!

I never said anything like that. I accused them of not knowing
anything about ergonomics, and being silly enough to build an Amiga
inside of an IBM PC. The second implies that they understand the
difference between an IBM PC and an Amiga.

I've also accused the CBM planning people of not understanding what's
involved in marketing computers. The A1000 succeeded in spite of them.
The A2000 and A500 will probably succeed in spite of the A1000.

<< Boy, a lot of people are flaming Commodore for the 1000 to 2000 upgrade.

No, a lot of people are flaming CBM because they didn't provide an
upgrade path. I introduced the 1000->2000 upgrade deal into this,
becasue that's what triggered the thing. I've already apologized for
that.

<<	Like the typical Commodore fanatic the emergence of the A2000 was
<< a threat to my "state-of-the-art" A1000.  Immediately I thought about
<< getting CSAs +$1000 tower.  Then I looked at the A2000 but sighed in despair
<< as the price shot up close to $2000.  Well I finally settled upon a
<< philosophy that I will use when the A3000 and A4000 come about.  It
<< essentially amounts to analyzing how much you are spending and what
<< percentage of that money will start going to work for you immediately.  

Naw, the A2000 isn't a threat to the A1000. It wasn't even serious
competition, as far as I was concerned. It took about 15 minutes of
typing at the A2000 to convince me that upgrading was a silly idea.

Unfortunately, it only took me about 30 minutes of thought to realize
that CBM probably had something nasty in the works for A1000 owners,
and when the screaming started, they'd just say "So why didn't you
upgrade?"

It *isn't* that I think that the A2000 is a wonderful machine, and I
feel shat upon because I didn't buy one. It's because the market for
cards to plug into my expansion box is drying up, there probably won't
be any new cards, and CBM is apparently up to something nasty. That,
coupled with lack of any kind of upgrade path is what the problem is.

<< John Campbell, a former (and possibly current) editor of Analog once
<< wrote that pioneering amounts to finding new and more horrible ways
<< to die.  Amiga buyers, particularly early ones, were pioneers, and
<< as such they risked seeing their investment in the Amiga, in both time 
<< and money, die.

Yes, but JWC didn't cover the people benefitting from your pioneering
stabbing you in the back as one of the risks.

And just for those who haven't gotten it yet - the backstab isn't
announcing the A2000, it's failing to provide for compatability to one
generation back. Even MicroShaft does that much!

<< ...and Commodore still has the hackers.  If you don't think so then have 
<< a look at the cat demo from the BADGE killer demo contest.  (and the 
<< others too, they're totally incredible)

And how many of them were actually done on either an A2000 or an A500?

<< So, to the people who are bummed out about the minor incompatibilities:

I don't call having to disgard three times the value of the cpu in
peripheral boxes minor!!!!!

<< Commodore is doing a *wonderful* job supporting 1000 users by maintaining
<< software compatibility

But they blew off hardware compatibility.  They've introduced a door
for software incompatibilities, and haven't said anything about not
using it!

<< I think that upgrade deal demonstrates a continuing committment to
<< their early customers

I think it demonstrates a commitment to killing the A1000 so the
screams of pain when they shaft it's owners won't be very loud.

<< Incidentally, I was in a Federated here in Houston today.  They have a big
<< new Atari display, and the salesman confidently informed me that the Amiga
<< 2000 had been discontinued.

We're all entitled to be able to dream. Odd that an Atari salesman and
an Amiga fan would have the same one, though.

Finally, I've gotten lots of mail from people who assume I *don't*
know about ways to make the A2000 resemble something other than an
Amiga in an IBM-PC box. Some even got their facts wrong.

1) I know I can put a keyboard garage underneath the A2000. Known it
	for a long time. It would happen almost as soon as an A2000
	arrived.

2) I know that Zorro I cards plug into Zorro II slots; you just can't
	close the case. My A1000 gets enough dust in it as is. I'm not
	about to run anything with an open case.

3) I know about the microbotics SOTS adapter. It won't work with
	any of the expansion boxes I've used because it requires
	leaving the cover off. The things I've put in the SOTS slot
	almost certainly weigh to much, and would cover several slots
	anyway. Not acceptable.

4) I know about the A2000-n-1. To date, it's varporware. But ASDG does
	good things, and helps provide an upgrade path that CBM
	didn't see fit to provide, so I expect it to appear. On the
	other hand, it doesn't solve the problem of CBM producing
	stuff that won't run on/in the A1000.

5) I even recall (as apparently nobody else does) a posting by a CBM
	employee when we first started seeing that bogus A2000 keyboard
	that an adapter so you could use an A1000 keyboard on an A2000.

The bottom line is I made a stupid mistake. I assumed that a company
marketing a serious computer would act like a serious computer
company, even if they'd only sold toys in the past.

Of course, if I'd listened to Ed Chaban (who was the first person I
saw say that the A2000 was what the A1000 should have been), I would
have known that long ago. That's hilarious, all by itself.

	<mike