[Info-vax] Out with Hurd, in with OpenVMS

[John Wallace]

On Aug 21, 10:43 am, Michael Kraemer <M.Krae... at gsi.de> wrote:
> JF Mezei schrieb:
>
> > Michael Kraemer wrote:
>
> >>But that was still Olsen/VAX era, not Palmer/Alpha,
> >>so you cannot hold Palmer responsible for that.
>
> > Palmer is responsible for not fixing the price/competitiveness
> > sufficiently and quickly enough.
>
> By the time Palmer became boss this *was* fixed.
> Alpha hardware was roughly in sync with the competitors,
> in contrast to VAXen being overpriced by factors 3 to 6.
>
> > That should have been job #1, instead
> > of focusing on changing the Digital logo all over the world.
>
> > Which is why RD54 prices should have dropped in price to match current
> > prices. I used RD54 as example, but this applied to all of DEC's
> > hardware. The price set at launch would remain even years later when the
> > product's value had gone way down.
>
> That's not unusual. For example I don't remember IBM ever dropped
> the price of obsolete hardware, e.g. to clean out their
> stock. It was just discontinued, at (almost) the same price
> when it was introduced.
> Even next generation often came at (almost) the same price as
> the previous one, but was of course more powerful.
>
>
>
> >>What slash and burn?
> >>Alpha machines appeared about the same time as Palmer became CEO.
> >>Whatever "slash and burn" had occured before,
>
> > You may not remember, but it was Palmer that instituted the slash and
> > burn of Digital's software portfolio. Some sold off to others, many
> > abandonned. Many never made it to Alpha because they were abandonned
> > right at the start of Palmer's tenure.
>
> Nothing wrong here. Changing hardware platform allows to get
> rid of old habits.
> Again: the problem with Alpha,
> on which DEC had bet its future, was that it came way too late
> and almost *everything* within this ecosystem
> (hardware, support, VMS, Unix, NT) was at V1.0 at best,
> at a time when competitors had (for the time) mature
> complete solutions to reap the benefits of downsizing
> and open systems wave.
> To believe that this or that single piece of VMS software
> would have changed that is just wishful thinking.
> Even VMS as a whole is insignificant here.
> What might have helped a bit would have been focusing
> solely on Unix, leaving VMS on VAX and NT on x86,
> at least for the time being.
>
> But all that mess was decided upon long before Palmer
> got the helm.
>
>
>
> >>They did, and failed, due to the nature of the product.
> >>You can't make a $1000 PC with a $2000 CPU.
> >>Of course one can cripple the CPU to lower the price,
> >>but then, where's the advantage over commodity stuff?
>
> > Crippling a machine is just not done in a competitive environment. If
> > you can produce a machine for $1000 with X CPU in it, you will market
> > that machine with the fastest speed that CPU can do.
>
> But you couldn't have produced a $1000 machine with the fastest
> Alpha in it without making a loss, because this CPU alone
> costs $2000 to make.

A) $2000 to for who to make *what*, please?
B) Who says $2000?
C) It's the *system* cost that counts, not just the chip cost.

If the bus interface on the 21066 had been done right, you could have
built a dead simple reasonable performance Alpha board (with VMS
support) consisting of the CPU, the DRAMS, the IO controller, and very
little else - no northbridge, no southbridge. This chip was the heart
of the AlphaBook and the Multia and the early VMEbus Alphas, but
because the memory interface was "sub optimal" the performance was
poor.

I still have a PC164SX here. It has one of the (Samsung-built?) low
cost EV5s in it. That chip is smaller than today's commodity chips,
and the heatsink certainly is. It didn't cost anywhere near $2000 to
make (or for system builders to buy). When Gates decided NT/Alpha was
dead, perhaps along the lines of Palmer either being Larry's friend or
Bill's friend and Palmer didn't choose Bill, suddenly there wasn't
much of a market for a non-x86 NT chip.

Earlier, Michael wrote "even in 1990 nobody would have bought a 150Meg
diskanymore, the standard was 660 or even 990. "

I don't remember it quite like that. I bought my first PC three years
after that, in 1993 when Gateway 2000 opened up the European market
for sensibly priced PCs in cow-logo'd boxes. I wanted a PC to play NT
3.x on. What I got was a 486/33 with 12MB and a ~400MB IDE drive.
1993, not 1990, and 400MB not 660. Getting the next bigger drive above
420MB required a BIOS update and/or some weird disk address
translation software. In 1993.

On a related subject (PC prices and chip prices): Have readers been
following the recent coverage of the US court case concerning Dell's
mis-statement of results for the last few years? Intel had secretly
been paying Dell a fortune every year to ensure that no other player
got a look in at Dell (this is in addition to the fact that Dell get
most of the expensive bits of system design done for them by Intel).
There is evidence in court that this behaviour goes back quite a few
years. How far back does this behaviour actually go? I've heard some
interesting stories from the era of the "Intel Inside" stickers...
When the market is rigged like that, what chance does any new player
stand?

http://www.theregister.co.uk/2010/07/26/after_the_dell_settlement/
[Sorry, it's an Orlowski article, but the facts are clear enough]