Re: VAX9000 unearthed

[David Brownlee via cctalk]

On Sun, 20 Feb 2022 at 21:20, Bill Gunshannon via cctalk
 wrote:
>
> On 2/20/22 15:31, Chuck Guzis via cctalk wrote:
> > On 2/20/22 10:10, Mark Kahrs via cctalk wrote:
> >> I heard Butler Lampson once exclaim that ECL design was in some ways easier
> >> than TTL.  If you terminated every line, you get controlled impedances with
> >> controlled edges.  This was the design philosophy for the Dorado.
> >
> > Indeed--ECL WW prototype boards usually had a 3rd row for SIP
> > termination resistors alongside the DIP sockets.  One nice thing about
> > ECL is that there are many fewer problems with power rail spikes.  On
> > the other hand, the constant power consumption needs beefier power supplies.
> >
> > I recall that Honeywell redid one of their mainframe designs in ECL,
> > with somewhat disappointing performance results.  I don't recall the
> > details offhand.
>
> It's long enough ago that my mind is fuzzy, but I think Primes were
> ECL.
>

Gould made a set of realtime and (via a board swap) unix systems - we
had a few PowerNode 6040s and one 9080 at university. One of the
6040's was mostly consumed by providing NFS to the other machines. I
remember at the time being impressed that the ffs had on disk padding
for 64 bit timestamps.

When the university stopped using them the CS department just left
their 6040 running unmonitored over the summer, and came in to find
the aircon had partially flooded the room then failed, and the machine
was happily still running in a shallow pool of slightly steamy water
(it was, shall we say, not in a purpose built room)

I also heard a possibly apocryphal story that the royal navy was
tendering for an onboard computer for a sub: various suppliers found
that their sample machines kept crashing, and the gould was the only
one able to run reliably - turned out the installation location was
quite close to "power generation equipment" and had been selected as
the ambient temperature was too high to put anything else there

David

Re: VAX9000 unearthed

[Jon Elson via cctalk]

On 2/20/22 14:31, Chuck Guzis via cctalk wrote:

On 2/20/22 10:10, Mark Kahrs via cctalk wrote:

I heard Butler Lampson once exclaim that ECL design was in some ways easier
than TTL.  If you terminated every line, you get controlled impedances with
controlled edges.  This was the design philosophy for the Dorado.
Yes, quite true.  Really short lines could be left 
unterminated, but you still needed a pull-down resistor.

Indeed--ECL WW prototype boards usually had a 3rd row for SIP
termination resistors alongside the DIP sockets.  One nice thing about
ECL is that there are many fewer problems with power rail spikes.  On
the other hand, the constant power consumption needs beefier power supplies.

I recall that Honeywell redid one of their mainframe designs in ECL,
with somewhat disappointing performance results.  I don't recall the
details offhand.

Yup, I was trying to speed up a baud rate divider that was 
already using AS or FAST 4-bit counters.  I worked out how 
much faster I could do it with ECL, and the result was not 
worth it. But, back in the days when TTL was king, before 
LS, AS and FAST, ECL had a REAL advantage.


Jon

Re: VAX9000 unearthed

[Bill Gunshannon via cctalk]

On 2/20/22 15:31, Chuck Guzis via cctalk wrote:

On 2/20/22 10:10, Mark Kahrs via cctalk wrote:

I heard Butler Lampson once exclaim that ECL design was in some ways easier
than TTL.  If you terminated every line, you get controlled impedances with
controlled edges.  This was the design philosophy for the Dorado.


Indeed--ECL WW prototype boards usually had a 3rd row for SIP
termination resistors alongside the DIP sockets.  One nice thing about
ECL is that there are many fewer problems with power rail spikes.  On
the other hand, the constant power consumption needs beefier power supplies.

I recall that Honeywell redid one of their mainframe designs in ECL,
with somewhat disappointing performance results.  I don't recall the
details offhand.


It's long enough ago that my mind is fuzzy, but I think Primes were
ECL.

bill

Re: VAX9000 unearthed

[Chuck Guzis via cctalk]

On 2/20/22 10:10, Mark Kahrs via cctalk wrote:
> I heard Butler Lampson once exclaim that ECL design was in some ways easier
> than TTL.  If you terminated every line, you get controlled impedances with
> controlled edges.  This was the design philosophy for the Dorado.

Indeed--ECL WW prototype boards usually had a 3rd row for SIP
termination resistors alongside the DIP sockets.  One nice thing about
ECL is that there are many fewer problems with power rail spikes.  On
the other hand, the constant power consumption needs beefier power supplies.

I recall that Honeywell redid one of their mainframe designs in ECL,
with somewhat disappointing performance results.  I don't recall the
details offhand.

--Chuck

Re: VAX9000 unearthed

[Mark Kahrs via cctalk]

I heard Butler Lampson once exclaim that ECL design was in some ways easier
than TTL.  If you terminated every line, you get controlled impedances with
controlled edges.  This was the design philosophy for the Dorado.

Re: VAX9000 unearthed

[Paul Koning via cctalk]

> On Feb 19, 2022, at 1:28 PM, Jon Elson via cctalk  
> wrote:
> 
> On 2/18/22 21:43, ben via cctalk wrote:
>> 
>> The 70's was all low scale tech. I suspect it was the high speed/edge rates 
>> more the power that kept ECL from common use. Any other views on this topic. 
>> Ben, who only had access to RADIO SHACK in the 70's.
>> PS: Still grumbling about buying life time tubes at a big price,
>> just to see all tubes discontinued a year or two later.
> 
> Edge rates on pedestrian MECL 10K were not crazy fast.  Rise and fall of 
> about 1 ns, but the gate propagation delay was ALSO about 1 ns, so that was a 
> lot faster than TTL.  ECL was very easy to work with, crosstalk was not a 
> common issue.  But, you HAD to terminate any line over a foot, and better to 
> make it 6" to be sure.  And, the termination and pulldown resistors ate a LOT 
> of power!

I think there are a number of reasons why ECL was niche technology.  One is 
that TTL was fast enough for most applications.  Another is that more people 
knew TTL, and ECL requires (some) different design techniques.  Yet another is 
that higher levels of integration appeared in CMOS but not ECL.  Yet another is 
that ECL was expensive compared to the alternatives, partly because of the low 
integration and partly because of the low volume.

In the mid-1980s (I think) there was a very interesting project at DEC Western 
Research Lab to build a custom VSLI ECL processor chip.  A lot of amazing 
design was done for it.  One is power and cooling work; it was estimated to 
consume about 100 watts which in that day was utterly unheard of, by a 
substantial margin.  This was solved by a package with integral heat pipe.  
Another issue was the fact that ECL foundries each had their own design rules, 
and they were shutting down frequently.  So the CAD system needed to be able to 
let you specify a design where the fab rules were inputs to the layout 
algorithms.  The design took great advantage of ECL-specific logic capabilities 
like wire OR or stacked pass transistors.  I remember that the CAD system let 
the designer work at multiple levels in the same chip: at the rectangle level 
(for memory arrays), transistor level, gate level, and even write some 
constructs as programming language notations.  For example a 64-bit register 
could be specified as:

for (i = 0; i < 64; i++) { transistor-level schematic of a one-bit 
register }

Originally the idea was to use this for a 1 GHz Alpha; I think it ended up 
being a 1 GHz MIPS processor.  Possibly the project was killed before it quite 
finished.

That seems to have been one of the very few examples of ECL going beyond SSI.  
The physical possibility existed; the economics did not.

paul

Re: VAX9000 unearthed

[Jon Elson via cctalk]

On 2/18/22 21:43, ben via cctalk wrote:


The 70's was all low scale tech. I suspect it was the high 
speed/edge rates more the power that kept ECL from common 
use. Any other views on this topic. Ben, who only had 
access to RADIO SHACK in the 70's.
PS: Still grumbling about buying life time tubes at a big 
price,

just to see all tubes discontinued a year or two later.


Edge rates on pedestrian MECL 10K were not crazy fast.  Rise 
and fall of about 1 ns, but the gate propagation delay was 
ALSO about 1 ns, so that was a lot faster than TTL.  ECL was 
very easy to work with, crosstalk was not a common issue.  
But, you HAD to terminate any line over a foot, and better 
to make it 6" to be sure.  And, the termination and pulldown 
resistors ate a LOT of power!


Jon

Re: VAX9000 unearthed

[Gary Grebus via cctalk]

On 2/18/22 15:35, Paul Koning wrote:




On Feb 18, 2022, at 3:18 PM, Gary Grebus  wrote:

On 2/18/22 09:46, Paul Koning wrote:

...The 9000 also had its own I/O bus, XMI, different from BI.  I don't know how 
its performance compares, whether it was worth the effort.


XMI already existed as the system bus for the VAX 6000 series machines.   I/O 
on the VAX 6000's was via an XMI-to-BI bridge.  I don't remember the exact 
performance specs on XMI, but it was wider and faster than BI.

XMI was then also used as one of the possible I/O buses on the VAX 1 and 
AlphaServer 7000 and 8000 series machines, via a system bus to XMI bridge.   So 
the XMI I/O adapters were common across all these series of machines.


I didn't remember all those details, thanks.

There also was an effort at one point to adopt FutureBus in DEC systems.  We 
did a pile of design in the network architecture group to figure out how to 
handle interrupts and bus cycles efficiently; I don't remember if anything 
actually shipped with that stuff.



There was a FutureBus I/O subsystem for the AlphaServer 8000 series.  It 
was a qualified, orderable option, but I can't imagine we sold very many 
(if any).  It was done supposedly because the US Navy was standardizing 
on FutureBus for some application.  I vaguely recall DEC made an 
Ethernet adapter that went on FutureBus, but you would have needed 
another I/O bus to have a usable system.


The native I/O interface on the AlphaServer 8000 (aka TurboLaser) was 
one or more system bus to "hose" modules, where a "hose" was a pair of 
cables that provided a 32 bit data path in each direction.  The hose 
connected to a bridge module on the target I/O bus.  There was support 
for XMI, PCI, and FutureBus.


-- Gary

Re: VAX9000 unearthed

[ben via cctalk]

On 2022-02-18 8:21 p.m., Mark Linimon via cctalk wrote:

On Fri, Feb 18, 2022 at 05:25:26PM -0500, Patrick Finnegan via cctalk wrote:

Yours will be a lot cheaper to run.


Custom ECL chips?

I think I can go with "relatively cheaper".

Make sure you have a bazillion BTU of air conditioning ...

(Yes, I have had experience with ECL, albeit 1970s low-scale tech.  The
power consumption ... eek ...)

mcl


The 70's was all low scale tech. I suspect it was the high speed/edge 
rates more the power that kept ECL from common use. Any other views on 
this topic. Ben, who only had access to RADIO SHACK in the 70's.

PS: Still grumbling about buying life time tubes at a big price,
just to see all tubes discontinued a year or two later.

Re: VAX9000 unearthed

[Mark Linimon via cctalk]

On Fri, Feb 18, 2022 at 05:25:26PM -0500, Patrick Finnegan via cctalk wrote:
> Yours will be a lot cheaper to run.

Custom ECL chips?

I think I can go with "relatively cheaper".

Make sure you have a bazillion BTU of air conditioning ...

(Yes, I have had experience with ECL, albeit 1970s low-scale tech.  The
power consumption ... eek ...)

mcl

Re: VAX9000 unearthed

[Don Stalkowski via cctalk]

There's an issue of DTJ dedicated to the 9000:

http://dtjcd.vmsresource.org.uk/pdfs/dtj_v02-04_1990.pdf


On Fri, Feb 18, 2022 at 11:01:40PM +, Antonio Carlini via cctalk wrote:
> On 18/02/2022 17:16, Lee Courtney via cctalk wrote:
> > Paul,
> > 
> > What was the timeframe for the MPP?
> > 
> > Lee
> 
> 
> The earliest DECmpp reference I can find is from 1991:
> 
> https://eisner.decus.org/anon/htnotes/note?f1=INDUSTRY_NEWS=551.1
> 
> You can peruse the service manual here:
> 
> http://manx-docs.org/collections/mds-199909/cd2/decmpp/decacsmc.pdf
> 
> 
> There are other docs that you can most easily find by searching for "DECmpp"
> on manx.
> 
> 
> Antonio
> 
> 
> -- 
> Antonio Carlini
> anto...@acarlini.com
> 
> 

Re: VAX9000 unearthed

[Antonio Carlini via cctalk]

On 18/02/2022 17:16, Lee Courtney via cctalk wrote:

Paul,

What was the timeframe for the MPP?

Lee



The earliest DECmpp reference I can find is from 1991:

https://eisner.decus.org/anon/htnotes/note?f1=INDUSTRY_NEWS=551.1

You can peruse the service manual here:

http://manx-docs.org/collections/mds-199909/cd2/decmpp/decacsmc.pdf


There are other docs that you can most easily find by searching for 
"DECmpp" on manx.



Antonio


--
Antonio Carlini
anto...@acarlini.com

Re: VAX9000 unearthed

[Maciej W. Rozycki via cctalk]

On Fri, 18 Feb 2022, Paul Koning via cctalk wrote:

> From what was just reported, the 6000 series indeed did it that way.  
> But I think on the 9000 it was an I/O bus too.  I definitely remember 
> some work on XMI based I/O devices, in particular an FDDI card.  And 
> indeed you can find a spec for that device in 
> http://www.bitsavers.org/pdf/dec/xmi/ .

 Also: 
.

  Maciej

Re: VAX9000 unearthed

[Patrick Finnegan via cctalk]

Cool! It's the baby version of the two 9000/440s I recently rescued.

Yours will be a lot cheaper to run. I'm still working on digesting the
documentation for mine before I try to get them running.

Pictures of mine in their new home: https://i.imgur.com/sVgkuG9.jpg

Patrick Finnegan

On Fri, Feb 18, 2022, 07:53 Joerg Hoppe via cctalk 
wrote:

> Hi,
>
> my computer club c-c-g.de could acquire the remains of a VAX9000 !
> The machine ran at the GWDG computing center in Göttingen, Germany,
> around 1993.
> Parts of it were in stock of their museum for 20+ years.
>
> See lots of hires-pictures at
>
> https://c-c-g.de/fachartikel/359-vax-9000-ein-starker-exot
>
> (scroll to the bottom for a slide show).
>
> Joerg
>
>

Re: VAX9000 unearthed

[Paul Koning via cctalk]

> On Feb 18, 2022, at 4:30 PM, Chris Zach via cctalk  
> wrote:
> 
>> XMI already existed as the system bus for the VAX 6000 series machines.
>> I/O on the VAX 6000's was via an XMI-to-BI bridge.  I don't remember the 
>> exact performance specs on XMI, but it was wider and faster than BI.
> 
> I thought XMI was only supposed to be a CPU/memory bus, with IO being done by 
> multiple VaxBI busses. That's what we had on the 6000 at the computer 
> Society: 2 CPUs, memory, and two VaxBi with a SCSI disk controller on each.

From what was just reported, the 6000 series indeed did it that way.  But I 
think on the 9000 it was an I/O bus too.  I definitely remember some work on 
XMI based I/O devices, in particular an FDDI card.  And indeed you can find a 
spec for that device in http://www.bitsavers.org/pdf/dec/xmi/ .

paul

Re: VAX9000 unearthed

[Chris Zach via cctalk]

XMI already existed as the system bus for the VAX 6000 series machines. 
   I/O on the VAX 6000's was via an XMI-to-BI bridge.  I don't remember 
the exact performance specs on XMI, but it was wider and faster than BI.


I thought XMI was only supposed to be a CPU/memory bus, with IO being 
done by multiple VaxBI busses. That's what we had on the 6000 at the 
computer Society: 2 CPUs, memory, and two VaxBi with a SCSI disk 
controller on each.


I think it also had serial connections via terminal servers because the 
ceiling in the computer room collapsed shortly after I started working 
there from all the RS232 cables piled up there. Landed right on the 
SparcStation 20 and "crashed" our WAIS/Gopher server.


Learn something new every day.

Re: VAX9000 unearthed

[Gary Grebus via cctalk]

On 2/18/22 09:46, Paul Koning wrote:




On Feb 18, 2022, at 7:08 AM, Joerg Hoppe via cctalk  
wrote:

Hi,

my computer club c-c-g.de could acquire the remains of a VAX9000 !
The machine ran at the GWDG computing center in G?ttingen, Germany, around 1993.
Parts of it were in stock of their museum for 20+ years.

See lots of hires-pictures at

https://c-c-g.de/fachartikel/359-vax-9000-ein-starker-exot

(scroll to the bottom for a slide show).

Joerg


Excellent photos!

I didn't realize the 9000 had a vector processor.

One reason the design was so expensive is that it was originally planned as a water-cooled machine 
-- code name "Aquarius".  At some point that idea was dropped and switched to air cooling 
-- code name "Aridus".  I guess those skinny pipes with red and blue markers carry jets 
of cooling air, but were originally going to carry water.

The 9000 also had its own I/O bus, XMI, different from BI.  I don't know how 
its performance compares, whether it was worth the effort.



XMI already existed as the system bus for the VAX 6000 series machines. 
  I/O on the VAX 6000's was via an XMI-to-BI bridge.  I don't remember 
the exact performance specs on XMI, but it was wider and faster than BI.


XMI was then also used as one of the possible I/O buses on the VAX 1 
and AlphaServer 7000 and 8000 series machines, via a system bus to XMI 
bridge.   So the XMI I/O adapters were common across all these series of 
machines.


Gary

Re: VAX9000 unearthed

[Chris Zach via cctalk]

https://en.wikipedia.org/wiki/MasPar says that MasPar was founded by ex-DEC 
chip VP Jeff Kalb.  He took a design done at DEC, for a massively parallel 
machine inspired by the Goodyear MPP with some changes.  DEC decided not to 
build that so MasPar did and DEC then resold it.  The description sounds 
vaguely familiar.  The manual I downloaded says it has 1024 cores per board, 
and up to 16 boards.  Neat.


If you have a university library nearby go check the old Computer 
Society archives. I was involved with the Supercomputing conferences in 
the early 90's (and built the E-commerce system for SC94's registration) 
and talk of this system has knocked loose a marble in my brain from the 
late 80's. There might be more documentation in the various proceedings 
or in the Computer magazines and Transactions on Parallel Computing from 
the time

Re: VAX9000 unearthed

[Antonio Carlini via cctalk]

On 18/02/2022 20:35, Paul Koning via cctalk wrote:


There also was an effort at one point to adopt FutureBus in DEC systems.  We 
did a pile of design in the network architecture group to figure out how to 
handle interrupts and bus cycles efficiently; I don't remember if anything 
actually shipped with that stuff.

paul


The DEC 4000 systems (COBRA and the follow-on upgrade, FANG) use FB as 
an I/O bus. DECnis also used FB as its backplane. They couldn't share 
cards though.



Antonio


--
Antonio Carlini
anto...@acarlini.com

Re: VAX9000 unearthed

[Paul Koning via cctalk]

> On Feb 18, 2022, at 3:18 PM, Gary Grebus  wrote:
> 
> On 2/18/22 09:46, Paul Koning wrote:
>> ...The 9000 also had its own I/O bus, XMI, different from BI.  I don't know 
>> how its performance compares, whether it was worth the effort.
> 
> XMI already existed as the system bus for the VAX 6000 series machines.   I/O 
> on the VAX 6000's was via an XMI-to-BI bridge.  I don't remember the exact 
> performance specs on XMI, but it was wider and faster than BI.
> 
> XMI was then also used as one of the possible I/O buses on the VAX 1 and 
> AlphaServer 7000 and 8000 series machines, via a system bus to XMI bridge.   
> So the XMI I/O adapters were common across all these series of machines.

I didn't remember all those details, thanks.

There also was an effort at one point to adopt FutureBus in DEC systems.  We 
did a pile of design in the network architecture group to figure out how to 
handle interrupts and bus cycles efficiently; I don't remember if anything 
actually shipped with that stuff.

paul

Re: VAX9000 unearthed

[Bill Gunshannon via cctalk]

On 2/18/22 13:30, Josh Dersch via cctalk wrote:

On Fri, Feb 18, 2022 at 6:55 AM Paul Koning via cctalk <
cctalk@classiccmp.org> wrote:






Speaking of vector processors: there's a very obscure DEC processor, the
DEC MPP.  I remember seeing the processor architecture document when it was
being designed, not sure why.  It's a very-RISC machine, just a few
instructions, but lots of cores especially for that time -- 256?  More?
Recently I saw it mentioned in some documents, apparently it did get
produced and shipped, perhaps only in small numbers.  I wonder if any have
been preserved.



There was one listed on eBay for an obscene (like $500k?) price for quite
awhile, don't see it listed right now though.  Somehow I doubt it sold.
But at least there's one out there, somewhere...



Maybe, maybe not.  Failure to sell may have resulted in scrapping.
When I had to get rid of my last big iron vaxen at the University
it was made quite clear that if I didn't find them a home in a very
short time they would be scrapped.

bill

Re: VAX9000 unearthed

[Josh Dersch via cctalk]

On Fri, Feb 18, 2022 at 6:55 AM Paul Koning via cctalk <
cctalk@classiccmp.org> wrote:

>
>
>
>
> Speaking of vector processors: there's a very obscure DEC processor, the
> DEC MPP.  I remember seeing the processor architecture document when it was
> being designed, not sure why.  It's a very-RISC machine, just a few
> instructions, but lots of cores especially for that time -- 256?  More?
> Recently I saw it mentioned in some documents, apparently it did get
> produced and shipped, perhaps only in small numbers.  I wonder if any have
> been preserved.


There was one listed on eBay for an obscene (like $500k?) price for quite
awhile, don't see it listed right now though.  Somehow I doubt it sold.
But at least there's one out there, somewhere...

- Josh



> As far as I know there is no family connection between that machine and
> anything else DEC did before or since.
>
> paul
>
>
>

Re: VAX9000 unearthed

[Paul Koning via cctalk]

> On Feb 18, 2022, at 12:16 PM, Lee Courtney  wrote:
> 
> Paul,
> 
> What was the timeframe for the MPP?

I thought late 1980s.  Just did some searching, which turns up some manuals for 
the "DecMPP 12000".  And a trade press article that says it's a rebadged MasPar 
machine.

https://en.wikipedia.org/wiki/MasPar says that MasPar was founded by ex-DEC 
chip VP Jeff Kalb.  He took a design done at DEC, for a massively parallel 
machine inspired by the Goodyear MPP with some changes.  DEC decided not to 
build that so MasPar did and DEC then resold it.  The description sounds 
vaguely familiar.  The manual I downloaded says it has 1024 cores per board, 
and up to 16 boards.  Neat.

paul

Re: VAX9000 unearthed

[Lee Courtney via cctalk]

Paul,

What was the timeframe for the MPP?

Lee

On Fri, Feb 18, 2022 at 6:47 AM Paul Koning via cctalk <
cctalk@classiccmp.org> wrote:

>
>
> > On Feb 18, 2022, at 7:08 AM, Joerg Hoppe via cctalk <
> cctalk@classiccmp.org> wrote:
> >
> > Hi,
> >
> > my computer club c-c-g.de could acquire the remains of a VAX9000 !
> > The machine ran at the GWDG computing center in Göttingen, Germany,
> around 1993.
> > Parts of it were in stock of their museum for 20+ years.
> >
> > See lots of hires-pictures at
> >
> > https://c-c-g.de/fachartikel/359-vax-9000-ein-starker-exot
> >
> > (scroll to the bottom for a slide show).
> >
> > Joerg
>
> Excellent photos!
>
> I didn't realize the 9000 had a vector processor.
>
> One reason the design was so expensive is that it was originally planned
> as a water-cooled machine -- code name "Aquarius".  At some point that idea
> was dropped and switched to air cooling -- code name "Aridus".  I guess
> those skinny pipes with red and blue markers carry jets of cooling air, but
> were originally going to carry water.
>
> The 9000 also had its own I/O bus, XMI, different from BI.  I don't know
> how its performance compares, whether it was worth the effort.
>
> Speaking of vector processors: there's a very obscure DEC processor, the
> DEC MPP.  I remember seeing the processor architecture document when it was
> being designed, not sure why.  It's a very-RISC machine, just a few
> instructions, but lots of cores especially for that time -- 256?  More?
> Recently I saw it mentioned in some documents, apparently it did get
> produced and shipped, perhaps only in small numbers.  I wonder if any have
> been preserved.  As far as I know there is no family connection between
> that machine and anything else DEC did before or since.
>
> paul
>
>
>

-- 
Lee Courtney
+1-650-704-3934 cell

Re: VAX9000 unearthed

[Paul Koning via cctalk]

> On Feb 18, 2022, at 7:08 AM, Joerg Hoppe via cctalk  
> wrote:
> 
> Hi,
> 
> my computer club c-c-g.de could acquire the remains of a VAX9000 !
> The machine ran at the GWDG computing center in Göttingen, Germany, around 
> 1993.
> Parts of it were in stock of their museum for 20+ years.
> 
> See lots of hires-pictures at
> 
> https://c-c-g.de/fachartikel/359-vax-9000-ein-starker-exot
> 
> (scroll to the bottom for a slide show).
> 
> Joerg

Excellent photos!

I didn't realize the 9000 had a vector processor.

One reason the design was so expensive is that it was originally planned as a 
water-cooled machine -- code name "Aquarius".  At some point that idea was 
dropped and switched to air cooling -- code name "Aridus".  I guess those 
skinny pipes with red and blue markers carry jets of cooling air, but were 
originally going to carry water.

The 9000 also had its own I/O bus, XMI, different from BI.  I don't know how 
its performance compares, whether it was worth the effort.

Speaking of vector processors: there's a very obscure DEC processor, the DEC 
MPP.  I remember seeing the processor architecture document when it was being 
designed, not sure why.  It's a very-RISC machine, just a few instructions, but 
lots of cores especially for that time -- 256?  More?  Recently I saw it 
mentioned in some documents, apparently it did get produced and shipped, 
perhaps only in small numbers.  I wonder if any have been preserved.  As far as 
I know there is no family connection between that machine and anything else DEC 
did before or since.

paul

VAX9000 unearthed

[Joerg Hoppe via cctalk]

Hi,

my computer club c-c-g.de could acquire the remains of a VAX9000 !
The machine ran at the GWDG computing center in Göttingen, Germany, 
around 1993.

Parts of it were in stock of their museum for 20+ years.

See lots of hires-pictures at

https://c-c-g.de/fachartikel/359-vax-9000-ein-starker-exot

(scroll to the bottom for a slide show).

Joerg