[Info-vax] Vaxes shutting off this week

[Bill Gunshannon]

In article <bf2aaa5d-0102-4afe-8f80-117ab0ac575d at c36g2000yqn.googlegroups.com>,
	Neil Rieck <n.rieck at sympatico.ca> writes:
> On Mar 2, 9:34 pm, "Tom Linden" <t... at kednos.company> wrote:
>> On Sun, 01 Mar 2009 13:40:21 -0800, Michael Kraemer <M.Krae... at gsi.de>  
>> wrote:
>>
> [...snip...]
>>
>> I agree, Mips was in my view a better design, and Digital had after all
>> invested in the venture.  But as I have said before, abandoning the
>> VAX was the seminal event that led to Digital's failure.
>>
> I'm not so sure. VAX had to die 

Why?  IBM has had no problem (and no loss of ROI that has been mentioned by
anyone) while keeping the 360 Architecture alive and well.  A good design
is a good design.

>                                  but Alpha did not.
> I was sad to see Digital switch from PDP to VAX until I worked on VAX.
> While PDP was only 16-bit, addressing hardware extended the memory
> range to 24-bits. 

22 bits.

>                   VAX addressing allowed for a full 32-bits of
> addressing. 

While I liked the VAX, too, I see no reason why the PDP-11 could not
have been expanded while keeping its core architecture the same.  But
then, like with VMS, it would have been just as ssatisfactory to have
ported the OSes to other architectures.

>              Initial VAX models were able to run PDP software in
> hardware compatibility mode( for almost 10 years), so most software
> guys were happy. 

I believe that capability was very limited.  I certainly know you couldn't
run RSTS on a VAX in any mode.  :-)

>                   DEC still built their own peripherals which mean
> these machines were too expensive for most companies.
> I was sad to see Digital switch from VAX go to Alpha until I worked on
> Alpha. While VAX was only 32-bit, addressing hardware extended the
> memory range to 40-bits (and maybe even higher?). Alpha addressing
> allowed for 64-bits of addressing. 

And, again, is there any real reason why they cold not have built a
64bit VAX?  Was there something in the underlying concept that limited
the possibilities of larger bit sizes?  Somehow, I doubt it.  The
real reason for abandoning VAX was the warped notion that the whole
world had to go RISC.

>                                     The Alpha CPU added other important
> features including IEEE floating point. 

Vax supported a number of different kinds of FP.  Why would you think 
that adding IEEE FP was not doable?

>                                         Except for the CPU, Alpha was
> mostly built from COTS (commercial off the shelf) parts which means
> the purchase price dropped while performance went through the roof
> (mostly due to the switch from CISC to RISC) but faster memory an
> buses also helped. 

Mostly a matter of the bus being used.  There was no reason why EISA or PCI
VAX cards cold not have been made.  The rest would have been software.  Have
you ever seen the Strobedata Osprey?  PDP-11 on ISA or PCI.

>                     Many Alphas only required single-phase power and no
> special cooling.

I run VAXen in my house all the time with single-phase power and no
special cooling.  More modern devices were likely to be much less
energy hungry requiring even more modest power and less cooling.

> I never saw any reason for Compaq/HP killing Alpha and can only assume
> that new management thought they were repeating the transitions of the
> past. Intel fell into this trap too, and I am convinced that they will
> kill Itanium. This is especially true since Pentiums went from 32-bit
> to 64-bit compatible with most people not even noticing the
> changeover, as well as their Core2 technology going from single core
> to dual, then quad, and now hex while Itanium is still stuck at dual
> core. Itanium will soon be irrelevant.

On this, we agree.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999 at cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>