Re: Singapore Server Rescue

In article <fiu163$hqq$03$1@xxxxxxxxxxxxxxxxx>,
Michael Kraemer <M.Kraemer@xxxxxx> writes:
JF Mezei schrieb:

I should remind you that the PDP-11 was probably superior to the 8086 at
the time. And when you consider that Intel has been able to bring the
8086 from a lousy toy controller into a respectable enterprise 64 bit
architecture that still commands the market despite the RISC and EPIC
attempts at ousting it, it is wrong to assume that sticking with PDP-11
architecture would have doomed Digital.

the big difference was that the 8086 was a single chip solution
which could be produced relatively cheap in masses.
It could be sold to the masses and be incorporated in a lot
of industrial designs. OTOH, the PDP-11 I remember is a bulky
device, looking like a dinosaur. In fact it was,
since it belongs to an earlier step of IT evolution than
the microprocessor. If DEC had squeezed it into a single chip
and mass-marketed it in a timely (and non-proprietary) manner,
they might have got away with it for some time.

How many chips in the PDP-11 CPU used on the Strobe Data Osprey?

bill


Digital refused to compete at the low end not because it couldn't but
because it didn't want its highly lucrative ricgh customers to stop
buying the old expensicve dinausors and start buying the better lower
priced machines. As a result of wanting to retain a few high end
customers, Digital priced itself out of the market.

DEC didn't manage the "divide and conquer" strategy,
which IBM did (and does) very well.
IBM sells low-end as well as high-end equipment,
whatever the customers want, with the one not
necessarily cannibalizing the other.
This worked out even in the times when "downsizing"
and "open systems" were the motd.
Of course they lost a lot of mainframe business
back then, but they also had a replacement solution
(the RS/6000) ready for prime time to catch at least
part of the lost sales. And they had no problems
to sell equipment spanning two orders of magnitude
in raw price/performance to their respective customer base.

DEC didn't have a single vision in late 1980s. It had different
departments with their own pet projects (VAX 9000, N-vax, prism and
later alpha). Those competed for funding and attention.

Alpha won the internal PR war and convinced the powers that VAX was a
dead end bla bla bla bla and got the funding to go ahead. Meanwhile,
the vax group kept on working with N-VAX and came out with pretty
impressive speed improvements (proving wrong the Alpha guys' premise
that VAX was a dead end).

It may have been impressive for the VAX camp,
but it wasn't when compared to the contemporary RISC camp.
A VS4000-90 runs at what, 80+x MHz ?
This might be sufficient for the less power hungry,
but certainly wasn't for the typical audience in technical
and academic computing (which once was DECs main turf,
rather than banks and such).
At the same time (1992), RISC designs were already above 100 MHz,
executing at least 1 instruction per cycle.

Surely people here realize that comparing clock speeds is totally
meaningless. I doubt a 100 Mhz RISC system is functionally as fast
as an 80 Mhz VAX.


I am not so sure that there was a objective debate within all of DEC on
whether VAX was really a dead end or whether it could be improved to
remain competitive.

Look at the other CISC design with tragic fate, the 68K.

It is likely that the 68K died because they opted not to be the one
used for the IBM PC (they did have first shot) and the sheer size of
the created market made Intel (which had one foot int he grave at the
time) a success instead of another blank page in IT history.

Even a dedicated chip maker like Motorola couldn't evolve
it beyond the 50MHz mark.

Yes, but it could have. Imagine a world where the IBM PC was Motorola
based and not Intel based. :-)


And at that time, even if VAX wasn't as fast, DEC could have simply
priced its machine accordingly and remain in the game. The 8086s back
then weren't too quick either, yet they priced themselve low enough to
gain market share and look at where they are now.

Unlike DEC's architectures they were designed to be produced in
masses and sold over the counter to everybody right from the start.

Well, while that may have been Intel's dream, it was only the use by
IBM that kept the company from vanishing. At the time IBM decided on
using Intel the company was all but dead. IBM bought a majority share
of them so they could guarantee they would get first choice on sales.
Something Motorola would not guarantee.


When I recollect the 1990/91 price lists,
a reasonably equipped VS3176 workplace
should have costed 10 to 15 kDEM ($5000..$8000) max to be competitive
with a contemporary RISC machine, i.e. RS/6000
(or DECstation, for that matter). In fact,
the price was about 3 times higher, even after including
all kinds of discounts and OEM equipment.
I'm not sure if DEC could have afforded
to offer their main hardware at the price of
a somewhat better X-terminal.
In fact, it may well be that a contemporary
88k-driven X-terminal had more compute power
than those poor VAXen.

Sadly, we will never know. But it is fun to speculate on how things
could have been. :-)

bill

--
Bill Gunshannon | de-moc-ra-cy (di mok' ra see) n. Three wolves
bill@xxxxxxxxxxxxxxx | and a sheep voting on what's for dinner.
University of Scranton |
Scranton, Pennsylvania | #include <std.disclaimer.h>
.

Relevant Pages

  • Re: Singapore Server Rescue
    ... I doubt a 100 Mhz RISC system is functionally as fast ... than the fastest VAX. ... IBM 320H 25 43 ... and the workstation makers. ...
    (comp.os.vms)
  • Re: The Development of the Vital IBM PC in Spite of the Corporate Culture of IBM
    ... I also led the engineering side of the fight against the VAX. ... Anyone who thinks the RISC chips killed the 68K is off base. ... Motorola allied with IBM on RISC only after Intel had destroyed Motorola's market for the 68K. ... For IBM-MAIN subscribe / signoff / archive access instructions, ...
    (bit.listserv.ibm-main)
  • RE: The Development of the Vital IBM PC in Spite of the Corporate Culture of IBM
    ... I also led the engineering side of the fight against the VAX. ... Anyone who thinks the RISC chips killed the 68K is off base. ... Motorola allied with IBM on RISC only after Intel had destroyed Motorola's market for the 68K. ... For IBM-MAIN subscribe / signoff / archive access instructions, ...
    (bit.listserv.ibm-main)
  • Re: Curly soon to be out of a job
    ... >> The former technology unit includes IBM's Microelectronic Division, ... Recently, IBM ... > absurdly high-volume commodity chips which currently sustain it. ... > that was the rationale Intel used to justify undertaking its development ...
    (comp.os.vms)
  • Re: just for laughs
    ... both TI and Fujitsu are ahead of TSMC. ... use more of their own fab capacity than IBM. ... But Xbox is not really foundry work. ... or Intel. ...
    (comp.sys.ibm.pc.hardware.chips)