Re: Alpha remembrance day


JF Mezei wrote:
Andrew wrote:
So who would Digital have sold FAB capacity to?

I have been told some time ago that AMD had been very interestedbut that
Digital intentionally made it diffidult for AMD to buy FAB capacity for
Hudson. In the end, DEC got its hands on the Arm, built StrongArm and
used Hudson to build those chips. But that wasn't anywhere near enough
to fill the shoes that Alpha had been expected to fill.

That and the rather grandiose plans for Alpha meant that Palmer had to
reserve FAB capacity ofr Alpha. Its rather tricky just to turn a
contracted FAB run off in order to make your own units if demand rises.

Palmer took steps to prevent Alpha from growing, such as preventing it
from competing against Wintel. And by 1996, Palmer was already talking
to Pfeiffer about Compa buying Digital and Pfeiffer made it very clear
that he wasn't interested in the FAB, so from that point on, there was
no point in trying to fix Hudson to get more customers.


"Competing with Wintel" why would you think that Alpha was ever going
to manage that ? It was never going to hit the price points required to
compete with x86. The cheapest Alpha PC's cost $1500 and had a
miniscule software portfolio compared with x86 or for that matter
SPARC. Sun at least managed to build a sub $1000 workstation and make
money with it. Yes it picked up a number of vendors who were prepared
to develop Alpha PC's but I doubt if you can remember the names of any
of them and even if you do everyone else will say who.


inherited was the decision to FAB Alpha in house. The cost of designing
Alpha and then fabricating Alpha proved to be business decisions which
bled Digital dry.

Yes and no. It was the mismanagement and refusal to take on more work
which mad Husdon bleed money profusely. And Palmer hid that fact for a
very long time. I think it was Terry Shannon who got the scoop that
Hudson had been bleeding far more money than had been previously though,
and this was at the time Digital donated Hudson as a gift to Intel to
thank Intel for stealing Alpha intellectual property.


Hang on Palmer started talking to Compaq in 1996 and some time after
Palmer aledgedly started running down the FAB because Compaq didn't
want it.

Ok well aside from the rather dubious business logic that implies that
this course of action was a good move if Palmer wanted to unload the
FAB lets examine what happened prior to 1996.

Are you now suggesting that Palmer somehow guessed that Compaq would
not want the FAB and started running it down before even beginning
negociations with Compaq.

JF I can hear the thud, thud, thud of the black helicopters
approaching, sorry I forgot black helicopters are of course silent.

Note that Digital made a lot of its own chips. And had Digital
succesfully marketed its FAB capabilities and made it easy for others to
use it, it might have been a succesful enterprise.

By the time Alpha arrived many of the RISC processors Alpha was
designed to compete against were in their 2 and or third iteration.
MIPS had gone from R2000 to R3000 to R4000. Sun had gone from SPARC v7
to SPARC v8 etc.

Yes, but Alpha was designed from scratch with new knowldge/technologies
that allowed it to scale much higher than existing risc processors. And
consider that when Alpha came out, HP pissed in its pants and decided to
abandon its own PA-Risc and beg Intel to design something that could
compete against Alpha.


Really, or do you actually mean that Alpha design started latter and
therefore benefited from the general developments in architecture and
FAB technology.

Alpha may have been a late comer to the RISC party, but it was
definitely a class above the rest from a technology point of view.


Really, PA-RISC always gave it a serious run for its money on a per CPU
basis and SPARC generally trashed it on throughput because Sun were
better at designing Server platforms tham Digital. Compare the Sun
E10000, E6500, E4500 with the best Digital could produce at the time
the TurboLaser and guess what the TurboLaser was only comfortable
competing with the E4500.

You have to remember that the early 1990's was an period where Alpha
was one of a number of processor architectures trying to unseat x86.

It can be argued that Alpha succeeded. Consider the Pentium III which
contains a lot of "Alpha Inside"... :-)


Akkk, Alpha is a success because Intel nicked a few bits of Alpha IP in
the Pentium III, hardly the only success there was that it gave Digital
the lever to get the Hudson FAB off their back.

The reality is that whatever Palmer did he started from the standpoint
of having been delt a really terrible hand of cards.

Loy Gerstner inherited an IBM that was months away from bankrupcy and
processes already begun to sell parts of IBM off to pay the creditors.
He got an IBM in a much worse condition than Palmer got Digital. Palmer
didn't inherit a healthy Digital. But he worsoned the situation BIG
TIME.

IBM didn't have a rotten product strategy they just hadn't executed
very well and they had got very used to being awash with cash.

Regards
Andrew

.

Relevant Pages

  • Re: Alpha remembrance day
    ... wouldn't say that Alpha was late. ... having its own FAB was a way to be to market before any others by ... its FAB business, it may have had enough business to pay for upgrades. ... Compare that with the confuse DEC actions (even before Palmer). ...
    (comp.os.vms)
  • Re: Alpha remembrance day
    ... used Hudson to build those chips. ... to fill the shoes that Alpha had been expected to fill. ... reserve FAB capacity ofr Alpha. ... Palmer took steps to prevent Alpha from growing, ...
    (comp.os.vms)
  • Re: Alpha remembrance day
    ... wouldn't say that Alpha was late. ... having its own FAB was a way to be to market before any others by ... In the 3 years prior to the official announcement of the Compaq buying ... The decisions which led to Digitals downsizing were made in ...
    (comp.os.vms)
  • Re: Alpha remembrance day
    ... The decisions to design and build Alpha and the ... The only thing Palmer could have done at that point was cancelled the ... So who would Digital have sold FAB capacity to? ... target customers to use Digitals FAB. ...
    (comp.os.vms)
  • Re: Alpha remembrance day
    ... It died because the mistakes made by DEC prior to ... So Palmer did not ... Alpha which was a great opportunity that had been presented "ready to ... Microsoft reseller. ...
    (comp.os.vms)
Loading