Chinese Alpha?

[Rich Jordan]

Reportedly China is considering a 'national' CPU architecture (ISA at
least) decision for all government funded projects.

http://www.extremetech.com/computing/127791-china-plans-national-unified-cpu-architecture

"There are at least five existing ISAs on the table for consideration
— MIPS, Alpha, ARM, Power, and the homegrown UPU — but the Chinese
leadership has also mooted the idea of defining an entirely new
architecture."

[MG]

On 27-4-2012 17:11, Rich Jordan wrote:
> Reportedly China is considering a 'national' CPU architecture (ISA at
> least) decision for all government funded projects.

Weren't they looking into MIPS already a year or two ago? At the time,
I believe it was misreported and gave the wrong impression, that they
were looking into Alpha. However, this article claims --- on top of
those --- that they're looking into POWER, ARM and this "UPU" thing as
well. Now I don't know what to believe.

Only time will tell, I guess.

- MG

[VAXman-]

Well, unless they resurrect Alpha, they're hedging their bets on a dead
horse. However, it is nice to see the SPARC is not in contention.

--
VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)ORG

Well I speak to machines with the voice of humanity.

[JF Mezei]

VAXman- @SendSpamHere.ORG wrote:

> Well, unless they resurrect Alpha, they're hedging their bets on a dead
> horse. However, it is nice to see the SPARC is not in contention.

Buying the rights for a dead platform may be much easier and cheaper
than buying rights to copy a living one.


Doesn't the 8086 have special status that allows competitors to buid
their own copies ? Seems to me that this would be the way to go since
China could export those chips to compete against Intel and ARM.

[Dirk Munk]

VAXman- @SendSpamHere.ORG wrote:
> In article<4f9ac1a4$0$6893$e4fe...@news2.news.xs4all.nl>, MG<marc...@SPAMxs4all.nl> writes:
>> On 27-4-2012 17:11, Rich Jordan wrote:
>>> Reportedly China is considering a 'national' CPU architecture (ISA at
>>> least) decision for all government funded projects.
>>
>> Weren't they looking into MIPS already a year or two ago? At the time,
>> I believe it was misreported and gave the wrong impression, that they
>> were looking into Alpha. However, this article claims --- on top of
>> those --- that they're looking into POWER, ARM and this "UPU" thing as
>> well. Now I don't know what to believe.
>
> Well, unless they resurrect Alpha, they're hedging their bets on a dead
> horse. However, it is nice to see the SPARC is not in contention.
>

Well that would be a joke. I remember there were roadmaps up till EV12.
Just imagine the Chinese turning up with somethinng like an Alpha EV12
cpu. HP might buy them to replace Itanium :-) .

[glen herrmannsfeldt]

JF Mezei <jfmezei...@vaxination.ca> wrote:

(snip)

> Doesn't the 8086 have special status that allows competitors to buid
> their own copies ? Seems to me that this would be the way to go since
> China could export those chips to compete against Intel and ARM.

Last I remember, there were some second source agreements with AMD
in the 8086 days. (That is, 80x86 with x='').

There is always the patent vs. copyright issue. Remember, copyright
protects the expression of the idea. A clone could be a new
expression of an existing ISA. There could also be patents.
There is also the trademark on the name to consider.

I remember a story that the SPARC licence agreement requires one
to use a boldface font when writing SPARC. (My newsreader doesn't
have that option, though.)

-- glen

[David Froble]

Some of you might remember far back, when Japan was going to develop a "new and better
world beating computer". Guess they are still working on it.

Politicians love to come up with ideas such as these. Somehow, it never happens.
Politicians seem to be long on ideas, and short on achievements.

[Arne Vajhøj]

On 4/27/2012 2:07 PM, JF Mezei wrote:
> VAXman- @SendSpamHere.ORG wrote:
>> Well, unless they resurrect Alpha, they're hedging their bets on a dead
>> horse. However, it is nice to see the SPARC is not in contention.
>
> Buying the rights for a dead platform may be much easier and cheaper
> than buying rights to copy a living one.

True.

But I still do not believe in China going Alpha.

> Doesn't the 8086 have special status that allows competitors to buid
> their own copies ? Seems to me that this would be the way to go since
> China could export those chips to compete against Intel and ARM.

No.

Intel and AMD have some very complex cross licensing schemes
in place to deliver compatible CPU's.

Arne

[Arne Vajhøj]

There are politicians and there are politicians.

To refer to a couple of US presidents from different
parties I would mention Kennedy and Reagan as president
that got some things completed (well - in Kennedy's case
after his death).

But it requires determinations and the will to pay
what it costs.

Those characteristics is not that prominent among western
politicians in 2012.

But the Chinese??

Arne

[John Wallace]

On Apr 27, 4:11 pm, Rich Jordan <jor...@ccs4vms.com> wrote:
> Reportedly China is considering a 'national' CPU architecture (ISA at
> least) decision for all government funded projects.
>

> http://www.extremetech.com/computing/127791-china-plans-national-unif...

>
> "There are at least five existing ISAs on the table for consideration
> — MIPS, Alpha, ARM, Power, and the homegrown UPU — but the Chinese
> leadership has also mooted the idea of defining an entirely new
> architecture."

A similar rumo(u)r was reported last November, e.g. in The Register in
a throwaway comment at http://www.theregister.co.uk/2011/11/14/top_500_supers_nov_2011/page2.html

Closer to here, that article was then briefly discussed on
comp.sys.dec, post title "Alpha processors manufactured in China?"
dated November 16.

A comment on The Register's article asking for substantion for the
Alpha claim got no responses. A more forthright comment of mine was
rejected by The Register's Ministry of Truth.

I remember at the time of that article I attempted to follow any chain
of facts leading to evidence. There wasn't any I could find (as an
outsider with limited resources).

That doesn't mean it isn't happening, some might think it would be
quite sensible if it was happening (the architecture is proven, freely
available existing reference designs are proven albeit dated,
available architecture-specific software is proven. MIPS in general is
currently receiving the Last Rites, even in the embedded market, so
wouldn't be an obvious choice unless there were special
circumstances).

If you want 32bit addressing and don't need Windows, ARM has a great
deal going for it.

If you want 64bit addressing, the picture becomes less clear.

But what are they actually up to in China? We'll have to wait and see.

[John Wallace]

On Apr 27, 4:11 pm, Rich Jordan <jor...@ccs4vms.com> wrote:

> Reportedly China is considering a 'national' CPU architecture (ISA at
> least) decision for all government funded projects.
>

> http://www.extremetech.com/computing/127791-china-plans-national-unif...

>
> "There are at least five existing ISAs on the table for consideration
> — MIPS, Alpha, ARM, Power, and the homegrown UPU — but the Chinese
> leadership has also mooted the idea of defining an entirely new
> architecture."

This will be the same Extremetech and the same writer that in a
similar article last November [1] made the rather unusual assertion
that Loongson was Alpha based, whereas common industry consensus is
that Loongson is MIPS based, right?

"The ShenWei chips are based on the Loongson/Godson architecture,
which China — as in, the country itself — has been steadily developing
since 2001. It is believed that the Loongson family of processors,
including the ShenWei SW-3 found in Sunway, were created by reverse
engineering a DEC Alpha CPU."

[1] http://www.extremetech.com/computing/102461-east-vs-west-china-builds-record-breaking-homegrown-supercomputer

[JF Mezei]

Would gaining access to the Alpha IP but without the original Alpha
engineers be worth it ?

After all, what so many corporations forget, is that their biggest asset
are the employees. With the original Alpha engineers now dispersed at
various chip companies, it would be quite the steep learning curve for
someone to buy the rights for Alpha, hire some newbie team and get them
to essentially reverse engineer Alpha so they can understand it enough
to start to develop it further.

On the other hand, if they have people with experience in reverse
engineering chips, then this could be child's play for them.

[Paul Sture]

On Sat, 28 Apr 2012 03:01:16 -0700, John Wallace wrote:

> A similar rumo(u)r was reported last November, e.g. in The Register in a
> throwaway comment at
> http://www.theregister.co.uk/2011/11/14/top_500_supers_nov_2011/
page2.html

<snip>

> If you want 32bit addressing and don't need Windows, ARM has a great
> deal going for it.
>

> But what are they actually up to in China? We'll have to wait and see.

Manufacturing is what they are up to in China, and we know what OS is
useful in that field :-)



--
Paul Sture

[Simon Clubley]

On 2012-04-28, John Wallace <johnwa...@yahoo.co.uk> wrote:
>
> That doesn't mean it isn't happening, some might think it would be
> quite sensible if it was happening (the architecture is proven, freely
> available existing reference designs are proven albeit dated,
> available architecture-specific software is proven. MIPS in general is
> currently receiving the Last Rites, even in the embedded market, so
> wouldn't be an obvious choice unless there were special
> circumstances).
>

Interesting development here. Microchip have recently released a series
of 32 bit MCUs (the PIC32MX) which have a MIPS core and a PIC24 type
peripheral set. No full blown MMU on board as far as I can tell however,
which given the memory sizes the MCU support is not really a surprise.

> If you want 32bit addressing and don't need Windows, ARM has a great
> deal going for it.
>

ARM is my preferred embedded MCU of choice, all other things been equal.

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

[Arne Vajhøj]

On 4/28/2012 6:30 AM, JF Mezei wrote:
> Would gaining access to the Alpha IP but without the original Alpha
> engineers be worth it ?
>
> After all, what so many corporations forget, is that their biggest asset
> are the employees. With the original Alpha engineers now dispersed at
> various chip companies, it would be quite the steep learning curve for
> someone to buy the rights for Alpha, hire some newbie team and get them
> to essentially reverse engineer Alpha so they can understand it enough
> to start to develop it further.

If they get IP *incl.* various internal design documentation, then
it could certainly be valuable.

Arne

[David Froble]

Unless the save it in the "circular file" Samsung should have a bunch of details on Alpha.

[Arne Vajhøj]

On 4/29/2012 2:50 PM, David Froble wrote:
> Arne Vajhøj wrote:
>> On 4/28/2012 6:30 AM, JF Mezei wrote:
>>> Would gaining access to the Alpha IP but without the original Alpha
>>> engineers be worth it ?
>>>
>>> After all, what so many corporations forget, is that their biggest asset
>>> are the employees. With the original Alpha engineers now dispersed at
>>> various chip companies, it would be quite the steep learning curve for
>>> someone to buy the rights for Alpha, hire some newbie team and get them
>>> to essentially reverse engineer Alpha so they can understand it enough
>>> to start to develop it further.
>>
>> If they get IP *incl.* various internal design documentation, then
>> it could certainly be valuable.
>>
>

> Unless the save it in the "circular file" Samsung should have a bunch of
> details on Alpha.

Yes. But I doubt that they will sell to the Chinese.

Arne

[David Froble]

That an interesting statement.

I'd really doubt that Samsung has any future plans for the Alpha. If so, they wouldn't
have abandoned it. So, if someone offers them several million for something they don't
want, I'd be interested to know why anyone wouldn't take the money?

Not that I know if they really have anything to sell. The license agreement, or whatever,
might preclude disclosing certain information. Most likely, they threw out all that old
useless stuff.

[JF Mezei]

David Froble wrote:

> I'd really doubt that Samsung has any future plans for the Alpha.

Samsung now designs its own ARM chips for its mobile phones. (they are
not bigger than Nokia in the total mobile phone market).

The ARM guys design the cores, and various people including Samsung and
now Apple implement those designs for their chips.

Assuming that Samsung has full access to the Alpha IP, would this be of
use to them when implementing their ARM chips, or would the
designs/technilogies/ideas of Alpha apply only to work at a lower level,
done by ARM in england and and thus have no value to the "packaging"
done by Samsung ?

[glen herrmannsfeldt]

JF Mezei <jfmezei...@vaxination.ca> wrote:

(snip)

> Assuming that Samsung has full access to the Alpha IP, would this be of
> use to them when implementing their ARM chips, or would the
> designs/technilogies/ideas of Alpha apply only to work at a lower level,
> done by ARM in england and and thus have no value to the "packaging"
> done by Samsung ?

Alpha is an ISA (instruction set architecture) originally
implemented about 20 years ago. The ISA may be applicable today,
but the details of the implementations likely aren't.

As far as I know, there is no reason Alpha couldn't be implemented
in current technology, and be successful in the market.

Now, it might be that some details of old Alpha implementations
could be useful in new designs, but most of the lower level
(gate level) ideas, as far as I know, wouldn't.

-- glen

[JF Mezei]

glen herrmannsfeldt wrote:

> Now, it might be that some details of old Alpha implementations
> could be useful in new designs, but most of the lower level
> (gate level) ideas, as far as I know, wouldn't.

OK, so whatever Samsung has about Alpha wouldn't be of much value to
Samsung's endeavours in the mobile market with the ARM architecture today.

So if Alpha gives Samsung no competetitive advantage today, it makes it
easier for them to sell it off to the Chinese (assuming they are allowed
to, which I doubt they could do without HP's approval)

[Michael Kraemer]

JF Mezei schrieb:

> Would gaining access to the Alpha IP but without the original Alpha
> engineers be worth it ?

Chinese taking care about IP?
What will happen next, the hell will freeze over?

[John Wallace]

ARM today is a 32bit architecture doing very nicely in a 32bit market.
There is also a 64bit market which the ARM v8 architecture announced
last year will address. It's hard to see what ARM and their licensees
might need from the remains of Alpha that they haven't already got,
but then the future's kind of hard to predict sometimes.

Incidentally, it may well be true to say Samsung are bigger than Nokia
in mobiles, but since Elop arrived on his mission to finish off Nokia,
that's not real difficult. Perhaps more interestingly, Samsung have
two bites at the cherry in the mobile market, because they make
component level products used in product with other people's trendy
badges, and also they make stuff which sells in Samsung's own phones
too. In the iPhone 4, Samsung supply most of the most expensive
components other than the actual display [1]: the ARM processor, the
DRAM, the 16GB flash...

Now, what is there to look at outside the mobile market, and how
significant is it? In the Windows-dependent market, it still looks
like x86 or irrelevance; Windows on ARM seems to be deliberately
hobbled. But as more and more users and IT departments realise that
Windows is not "adding value", what will happen to x86 as a
consequence? Various Linuxes already run quite satisfactorily on
various ARMs, and more could be added with little effort.

In the massive non-Windows market for consumer and professional
embedded electronics, everything from a router to a TV and up probably
has an ARM Inside already.

In the server room, chips from people like Calxeda plugged into a
passive backplane can end up using less than 5W per 4GB server. E.g. a
Proliant SL6500-style box from a printer company called HP [2] can do
you a 4U unit which houses up to 288 servers (in trays of 72 servers
on cards of 4 servers). Handy for a HYPErvisor-free cloud/web
infrastructure, or maybe even for virtual desktop for the boring
routine stuff once the Board tell the IT department that the Wintel
security nightmare and upgrade treadmill is no longer necessary or
cost-effective for substantial parts of their wider business needs.

Actually the Calxeda high level concept is remarkably similar to the
Alpha 21066/21068 concept re-invented. The 21066 was a 21064 Alpha to
which you could just add DRAM and just add PCI. No Northbridge, no
Southbridge, it's all built in on "the processor". But the
21066/21068's reduced system cost and complexity (and performance)
weren't enough to bring system builders on board in an era when Linux
as an alternative to Windows was still considered radical. Have times
changed?

[1] http://www.isuppli.com/Teardowns/News/Pages/iPhone-4-Carries-Bill-of-Materials-of-187-51-According-to-iSuppli.aspx
[2] http://www.hp.com/hpinfo/newsroom/press/2011/111101xa.html and
http://www.youtube.com/watch?v=4PIajg_Htx0&feature=player_embedded and
even
http://www.hp.com/go/moonshot

[VAXman-]

In article <4f9e19a9$0$10856$c3e8da3$dbd...@news.astraweb.com>, JF Mezei <jfmezei...@vaxination.ca> writes:

>David Froble wrote:
>
>> I'd really doubt that Samsung has any future plans for the Alpha.
>

>Samsung now designs its own ARM chips for its mobile phones. (they are
>not bigger than Nokia in the total mobile phone market).
>
>The ARM guys design the cores, and various people including Samsung and
>now Apple implement those designs for their chips.
>

>Assuming that Samsung has full access to the Alpha IP, would this be of
>use to them when implementing their ARM chips, or would the
>designs/technilogies/ideas of Alpha apply only to work at a lower level,
>done by ARM in england and and thus have no value to the "packaging"
>done by Samsung ?

Here's a new hand-held chinese Alpha. <http://tinyurl.com/Hand-Held-Alpha>

[Paul Sture]

Wasn't Alpha quite power hungry? And isn't that ARM's advantage over
Intel?

http://buswk.co/IOdN51

"Until recent years, Intel (Intel) focused its efforts on what’s called
the “clock speed” of CPUs, rapidly increasing the performance of computer
chips to handle desktop operating systems and processor-intensive
applications better. Less thought was given to reducing the power
consumption requirements of these chips.

RAMPING UP PERFORMANCE

Contrast that with chips built on the ARM architecture, which is licensed
to chipmakers such as Nvidia, Qualcomm, Texas Instruments, Freescale, and
a host of others. Instead of the “top down” strategy of boosting
performance first and focusing on power requirements second, ARM chips
have used a “bottom up” approach. Early ARM chips weren’t capable of
running complex software but could run for days between charges. Once the
power requirements of the silicon were effectively managed, ARM chips
began to ramp up performance, most recently with quad-core chips that can
offer 16 hours of high-definition playback on a tablet."


--
Paul Sture

[Jan-Erik Soderholm]

Paul Sture wrote 2012-04-30 13:45:
> On Mon, 30 Apr 2012 05:11:08 +0000, glen herrmannsfeldt wrote:
>
>> JF Mezei<jfmezei...@vaxination.ca> wrote:
>>
>> (snip)
>>> Assuming that Samsung has full access to the Alpha IP, would this be of
>>> use to them when implementing their ARM chips, or would the
>>> designs/technilogies/ideas of Alpha apply only to work at a lower
>>> level, done by ARM in england and and thus have no value to the
>>> "packaging" done by Samsung ?
>>
>> Alpha is an ISA (instruction set architecture) originally implemented
>> about 20 years ago. The ISA may be applicable today, but the details of
>> the implementations likely aren't.
>>
>> As far as I know, there is no reason Alpha couldn't be implemented in
>> current technology, and be successful in the market.
>>
>> Now, it might be that some details of old Alpha implementations could be
>> useful in new designs, but most of the lower level (gate level) ideas,
>> as far as I know, wouldn't.
>
> Wasn't Alpha quite power hungry?

First released Alpha processorn was built using a 750 nm process.
The volume models (EV5/EV6) used a 500 or 350 nm process.
The last sold (EV7/EV7z) used 180 nm
The canceled EV78/EV79/EV8 would have used 130 or 125 nm.

Latest Xeon (and IA64 Poulson) uses/will use a 32 nm process.

125 => 32 nm = aprox *30* times less chip area.

Yes, the Alpha chips was rather power hungry, but one can not
compare 10 yers old technology with todays chip processes.
It's amazing what speed the Alphas run at on those processes...

Jan-Erik.

[Jan-Erik Soderholm]

Sorry, a slight error there. Should have been :

180 => 32 nm = aprox *30* times less chip area.

Jan-Erik.

[Rich Jordan]

"... It is believed that the Loongson family of processors, including

the ShenWei SW-3 found in Sunway, were created by reverse engineering
a DEC Alpha CPU."

When you care enough to steal the very best...

[glen herrmannsfeldt]

Reverse engineering is one thing, copying another.

There are stories of a Russian 8080 made by generating masks
directly from an extracted 8080 chip, including the intel
copyright symbol. (Slightly bigger and slower, though.)

One should be able to avoid copyright by reimplementing the ISA
from the architecture description. One then only has to worry about
any patents (which might have expired) and trademarks.

-- glen

[John Wallace]

21064 started life at around 150W. Xeons have been up there too,
although currently they're a little bit less. Then you need to add all
the Northbridge/Southbridge or similar gubbins traditionally
associated with an x86 CPU.

[JF Mezei]

John Wallace wrote:

> 21064 started life at around 150W. Xeons have been up there too,
> although currently they're a little bit less. Then you need to add all
> the Northbridge/Southbridge or similar gubbins traditionally
> associated with an x86 CPU.

My Nehalem quadcore mahines consume roughly 150 watts when not too busy,
but this is the total system consumption at the plug. Goes up to about
220watts when busy.

Remember that Alpha was done before the days where reducing power/heat
was a priority. When Merced came out, the power consumption issue was
just beginning to be important to the market. (remember that initial
IS64s had bad power consumption to unit of computation ratios).

Had EV7 work not be stretched and delayed over many years, it would have
come out before power consumtion was a big issue, and my guess is that
any shrinks and EV8 would have then dealt with it.

Note that a lot of the "design" issues for power have to do with
shutting down portions of the CPU not currently needed (such as extra
cores, turbo-boost etc).

When you are single core, there is less you can do to lower power
consumption. However, in the case of IA64, its gargantual and excessive
number of transistors probably makes it very hard to get much lower
power consumption.

Had Digital and Alpha survived, I am pretty sure that they would have
lead the market with power efficient Alpha, especially once they would
have gone multi core.

[Michael Unger]

On 2012-04-30 20:20, "John Wallace" wrote:

> 21064 started life at around 150W. Xeons have been up there too,
> although currently they're a little bit less. Then you need to add all
> the Northbridge/Southbridge or similar gubbins traditionally
> associated with an x86 CPU.

"Green IT" (i.e., achieving significant savings on the electric power
bill) is a rather recent trend in the hardware business ...

There have been and are of course "mobile editions" of the x86
architecture (for notebooks, e.g.) running at a very low CPU core
voltage. And as has already been mentioned there are extremely low power
designs like ARM with a reasonable computing power. I don't know what
GPS chipsets (tracking devices, running several days from a single or
two AA cells) are using though.

Michael

--
Real names enhance the probability of getting real answers.
My e-mail account at DECUS Munich is no longer valid.

[John Wallace]

"IA64 ... gargantual and excessive number of transistors probably

makes it very hard to get much lower power consumption."

Take away most of the XXXXL (24MB?) cache and IA64 chips would not
only use less power they would use less silicon too. But presumably
that might expose the underlying performance.

Intel's x86s aren't going to be able to do anything radical with power
consumption without losing compatibility with existing software (ie
Windows). Lose the need for Windows, and where's the big incentive to
use x86? Any consumer electronics or embedded system builder can
answer that one, and you can see their answers all around you (you've
got examples of their decisions in your home, in the car, in the
office, and in your pocket). You can also see the answer in some of
the more sensible datacentres.

[Arne Vajhøj]

On 4/29/2012 10:44 PM, David Froble wrote:
> Arne Vajhøj wrote:
>> On 4/29/2012 2:50 PM, David Froble wrote:
>>> Arne Vajhøj wrote:
>>>> On 4/28/2012 6:30 AM, JF Mezei wrote:
>>>>> Would gaining access to the Alpha IP but without the original Alpha
>>>>> engineers be worth it ?
>>>>>
>>>>> After all, what so many corporations forget, is that their biggest
>>>>> asset
>>>>> are the employees. With the original Alpha engineers now dispersed at
>>>>> various chip companies, it would be quite the steep learning curve for
>>>>> someone to buy the rights for Alpha, hire some newbie team and get
>>>>> them
>>>>> to essentially reverse engineer Alpha so they can understand it enough
>>>>> to start to develop it further.
>>>>
>>>> If they get IP *incl.* various internal design documentation, then
>>>> it could certainly be valuable.
>>>>
>>>
>>> Unless the save it in the "circular file" Samsung should have a bunch of
>>> details on Alpha.
>>
>> Yes. But I doubt that they will sell to the Chinese.
>

> That an interesting statement.
>
> I'd really doubt that Samsung has any future plans for the Alpha. If so,
> they wouldn't have abandoned it. So, if someone offers them several
> million for something they don't want, I'd be interested to know why
> anyone wouldn't take the money?

If there is a commercial market then Samsung would want to sell
the chips themselves.

If there is a non commercial market, then I don't think South Korea
will sell to the closest ally of North Korea.

Arne

[JF Mezei]

Arne Vajhøj wrote:

> If there is a non commercial market, then I don't think South Korea
> will sell to the closest ally of North Korea.

HP would definitly donate Alpha IP to China in exchange for some favours
from the chinese government to help HP produce in china and perhaps even
sell its products in China.

[Michael Moroney]

glen herrmannsfeldt <g...@ugcs.caltech.edu> writes:

>Rich Jordan <jor...@ccs4vms.com> wrote:

>> "... It is believed that the Loongson family of processors, including
>> the ShenWei SW-3 found in Sunway, were created by reverse engineering
>> a DEC Alpha CPU."

>> When you care enough to steal the very best...

>Reverse engineering is one thing, copying another.

>There are stories of a Russian 8080 made by generating masks
>directly from an extracted 8080 chip, including the intel
>copyright symbol. (Slightly bigger and slower, though.)

In case you missed it, the phrase "When you care enough to steal the very
best" has significance.

In the old days, the VAX was one of the favorite computers of the Soviets.
VAXen were export-restricted, so the Soviets reverse-engineered them
and made their own. DEC engineers knew this, so on one of the Microvax
chips, they had added the phrase "When you care enough to steal the very
best", in Russian, on the silicon, a little Easter egg for the Soviets
to find.

Perhaps someone can fill in details, such as which VAX chip it was.

[JF Mezei]

Say HP would donate Alpha to the chinese for $9.99.


Would the current DEC compilers (maintained by HP) be of use to the
chinese if they wanted to go Linux ? (aka: is it easy to port an alpha
compiler from VMS/Tru64 to Linux for instance ?)


How long would it realistically take for the chinese between purchase of
Alpha and compilers and being able to compete against Sun/IBM in the
enterprise market with an Alpha based solution ?

It would be sweet irony if the chinese could get Tru64 and Alpha for
peanuts, and within 5 years come back with a quadcore Alpha with current
process shrink and speeds and a updated True64 and negotiate to get
Oracle on it.

HP would have donated the very technology that could have gotten it out
of the Itanium dead end and now the chinese would steal market share
from HP.

[Paul Sture]

On Tue, 01 May 2012 03:58:34 +0000, Michael Moroney wrote:

> In case you missed it, the phrase "When you care enough to steal the
> very best" has significance.
>
> In the old days, the VAX was one of the favorite computers of the
> Soviets. VAXen were export-restricted, so the Soviets reverse-engineered
> them and made their own. DEC engineers knew this, so on one of the
> Microvax chips, they had added the phrase "When you care enough to steal
> the very best", in Russian, on the silicon, a little Easter egg for the
> Soviets to find.
>
> Perhaps someone can fill in details, such as which VAX chip it was.

Here's a picture:

http://micro.magnet.fsu.edu/creatures/pages/russians.html

According to the Wiki entry for VAX:

https://en.wikipedia.org/wiki/VAXen#History

"Further VLSI VAX processors followed in the form of the V-11, CVAX, SOC
("System On Chip", a single-chip CVAX), Rigel, Mariah and NVAX
implementations. The VAX microprocessors extended the architecture to
inexpensive workstations and later also supplanted the high-end VAX
models. This wide range of platforms (mainframe to workstation) using one
architecture was unique in the computer industry at that time. Sundry
graphics were etched onto the CVAX microprocessor die. The phrase CVAX...
when you care enough to steal the very best was etched in broken Russian
as a play on a Hallmark Cards slogan, intended as a message to Soviet
engineers who were known to be both purloining DEC computers for military
applications, along with reverse engineering their chip design."

And circa 1983/4 the Russians were advertising in France for VMS skills.
They were offering a _lot_ of money for that time (~50 GBP p.a. IIRC).


--
Paul Sture

[Phillip Helbig---undress to reply]

In article <nde479...@news.sture.ch>, Paul Sture <pa...@sture.ch>
writes:

> > In case you missed it, the phrase "When you care enough to steal the
> > very best" has significance.
> >
> > In the old days, the VAX was one of the favorite computers of the
> > Soviets. VAXen were export-restricted, so the Soviets reverse-engineered
> > them and made their own. DEC engineers knew this, so on one of the
> > Microvax chips, they had added the phrase "When you care enough to steal
> > the very best", in Russian, on the silicon, a little Easter egg for the
> > Soviets to find.
> >
> > Perhaps someone can fill in details, such as which VAX chip it was.
>
> Here's a picture:
>
> http://micro.magnet.fsu.edu/creatures/pages/russians.html

Maybe it's time to update the phrase: first, write it in Chinese instead
of Russian; second, make it "for those who are rich enough to buy the
very best". :-|

[Bob Koehler]

In article <4f9f8a8b$0$6911$c3e8da3$9b4f...@news.astraweb.com>, JF Mezei <jfmezei...@vaxination.ca> writes:
> Say HP would donate Alpha to the chinese for $9.99.
>
>
> Would the current DEC compilers (maintained by HP) be of use to the
> chinese if they wanted to go Linux ? (aka: is it easy to port an alpha
> compiler from VMS/Tru64 to Linux for instance ?)

Why use DEC compilers with Linux? Linux was ported to Alpha and used
the gnu compilers.

OK, I know the DEC compilers often generate better code, but the
world doesn't seem to care. Nobody even seems to turn on compiler
optimization in UNIX land.

[John Wallace]

On May 1, 2:29 pm, koeh...@eisner.nospam.encompasserve.org (Bob
Koehler) wrote:

> In article <4f9f8a8b$0$6911$c3e8da3$9b4ff...@news.astraweb.com>, JF Mezei <jfmezei.spam...@vaxination.ca> writes:
>
> > Say  HP would donate Alpha to the chinese for $9.99.
>
> > Would the current DEC compilers (maintained by HP) be of use to the
> > chinese if they wanted to go Linux ? (aka: is it easy to port an alpha
> > compiler from VMS/Tru64 to Linux for instance ?)
>
>    Why use DEC compilers with Linux?  Linux was ported to Alpha and used
>    the gnu compilers.
>
>    OK, I know the DEC compilers often generate better code, but the
>    world doesn't seem to care.  Nobody even seems to turn on compiler
>    optimization in UNIX land.

Why use DEC compilers with Linux Alpha? Same reason you'd run with
optimisation enabled with gcc - because performance is important to
some people?

There are still people interested in performance; they may not be very
visible in the general market but they include big spenders in well
established markets such as compute farms (everything from generic HPC
to commercial 3D render farms) and *maybe* cloud infrastructure
providers and others I forget to mention.

[Keith Parris]

Found some interesting quotes here:
http://vr-zone.com/articles/chinese-high-end-cpus-are-now-in-the-game--details--part-2-alpha/14347.html

"China saw the value and capability of Alpha, and built a number of
Alpha systems, some of them very large for the time. It also fully
licenced the Digital / Tru64 UNIX and related software stack, including
getting the full source code, from Compaq after the latter bought DEC."

"After over a decade of work and three generations of CPUs, Jiangnan
Reseach Lab has shown the ShenWei (Sunway) SW-3 processor, the Chinese
flavour of Alpha, not in a small workstation, not in a server, but in no
less than a huge petaflop-class supercomputer machine in Jinan, Shandong
- the Sunway BlueLight MPP."

"SW3 aka SW1600 is a 16-core, 64-bit RISC processor, with each core
looking a lot like an improved version of the 21164A EV56 Alpha core."

I wonder if the Chinese got TruClusters along with Tru64.

[John Wallace]

On May 1, 9:55 pm, Keith Parris <keithparris_deletet...@yahoo.com>
wrote:
> Found some interesting quotes here:http://vr-zone.com/articles/chinese-high-end-cpus-are-now-in-the-game...

>
> "China saw the value and capability of Alpha, and built a number of
> Alpha systems, some of them very large for the time. It also fully
> licenced the Digital / Tru64 UNIX and related software stack, including
> getting the full source code, from Compaq after the latter bought DEC."
>
> "After over a decade of work and three generations of CPUs, Jiangnan
> Reseach Lab has shown the ShenWei (Sunway) SW-3 processor, the Chinese
> flavour of Alpha, not in a small workstation, not in a server, but in no
> less than a huge petaflop-class supercomputer machine in Jinan, Shandong
> - the Sunway BlueLight MPP."
>
> "SW3 aka SW1600 is a 16-core, 64-bit RISC processor, with each core
> looking a lot like an improved version of the 21164A EV56 Alpha core."
>
> I wonder if the Chinese got TruClusters along with Tru64.

Written by Nebosja Novakovic, a name I think I recognise from industry
comic The Inquirer? The Alpha facts in the article may well be correct
(I didn't check them all), but there's nothing I could see to really
substantiate the claims of a Chinese Alpha connection.

A different article on the same supercomputer, again by a name which
might be recognised round here (Cade Metz), says the system uses "a
new instruction set" without saying how that conclusion is reached:
http://www.wired.com/wiredenterprise/2011/10/sunway_super/

Both those articles have pictures which are a subset of those at
http://laotsao.wordpress.com/2011/10/29/sunway-bluelight-mpp-%E7%A5%9E%E5%A8%81%E8%93%9D%E5%85%89/

That page appears to be a (partial?) translation of the page it links
to as "Other detail". It has lots of mentions of generic supercomputer
hardware stuff, no visible mention of Alpha, no significant mention of
software.

So I'm still looking for a *credible* source for this "it's derived
from Alpha" rumour. I can't help wondering slightly if some machine
translator has confused "it is an alpha version" with "it is a version
of Alpha", or something along those lines. That would be a laugh.

Unless I'm missing something (always possible), I'd probably be more
likely to believe something like a screen grab of a Tru64 or Linux on
Alpha boot sequence, or some other software-derived evidence, than the
generic largely hardware-independent stuff I've seen so far.

Don't forget the Chinese allegedly already have at least one builder
of IA64 systems too: Huawei (another name which should be familiar),
according to
http://forums.theregister.co.uk/forum/1/2011/04/14/huawei_inspur_itanium_servers/

Again, not saying it's not true, there are plenty of technical reasons
why it might make sense, just saying show me some real evidence (maybe
in English, or from Chinese via a trustworthy source?).

[Single Stage to Orbit]

On Tue, 2012-05-01 at 14:55 -0600, Keith Parris wrote:
> "SW3 aka SW1600 is a 16-core, 64-bit RISC processor, with each core
> looking a lot like an improved version of the 21164A EV56 Alpha
> core."

But can it boot VMS? :)
--
Tactical Nuclear Kittens

[Michael Kraemer]

JF Mezei schrieb:

> Had EV7 work not be stretched and delayed over many years, it would have
> come out before power consumtion was a big issue, and my guess is that
> any shrinks and EV8 would have then dealt with it.

I don't get it what magical design features the Alpha has, which
would make it still attractive, even if it had been developed further.
It's a design more than twenty years old, patents would have expired,
and even by DEC's exaggerated claims it would reach its predicted
lifetime in five years from now.

> Had Digital and Alpha survived, I am pretty sure that they would have
> lead the market with power efficient Alpha, especially once they would
> have gone multi core.

I don't know what feeds your pipe dreams,
but the Alpha never had a good track record in power efficiency.
In fact, its inability to adapt to the needs of embedded devices
(with the exception of AXPvme, which failed miserably)
was one reason for its demise.

[Michael Kraemer]

Keith Parris schrieb:

> "SW3 aka SW1600 is a 16-core, 64-bit RISC processor, with each core
> looking a lot like

what does that mean?
At some point, all CPUs look very similar.

> an improved version of the 21164A EV56 Alpha core."

That would be an EV6, no?
Now, how old is this part?
Should I be impressed?

[David Froble]

When it was being competitively developed, nothing was faster than Alpha.

As has been mentioned elsewhere, some people will pay for performance.

To get to your question, you need to ask, "why was it fast?" Perhaps because the design
was better than the competition? If so, and if there has not been any new designs that
might be better, then a competitively developed Alpha might still be the fastest single
processor available. When talking multiple cores, the "glue" on the chip, memory
controllers, interprocessor communications, and such in EV7 and EV7z, even with the larger
die size was doing things that out performed competing CPUs that had continued development
and smaller die size.

So, if the Alpha could be successfully shrunk down to say 35 nm, or even 22 nm which Intel
has just released, and with a large on chip cache as the IA-64 has, perhaps it might still
out-perform anything else available.

Now, that is a mighty big "if", and what it would cost I have no idea, other than it
wouldn't be cheap. So, commercially, might it be viable? I really have no idea. It
would depend upon how many might pay for such performance.

My perception is that the majority of people using VMS are not doing so for ultimate
performance. If this is so, then the majority just might be satisfied with VMS running
(and being adequately supported) on x86-64. Since I feel very strongly that some of the
reason for VMS losing market share is the perception to some that the HW that supports VMS
might easily die off, having it on x86 would quash that issue.

Actually, VMS is supported, in emulation, on x86 right now, and the performance appears to
be satisfactory, else the emulator vendors wouldn't be in business.

[JF Mezei]

Michael Kraemer wrote:
>
> I don't get it what magical design features the Alpha has, which
> would make it still attractive, even if it had been developed further.

Alpha is the only native 64 bit architectire that worked and performed
well. The only other native 64 bit is IA64 and we all knwo who well it
TANKED !

The thing about Alpha is that it was cleanly designed. And this is what
alled DEC engineers to implement features way before others.

Remember that 64 bit x86 came from the 8086. So there is a lot of
baggage there that makes improving it harder.

[JF Mezei]

David Froble wrote:

> My perception is that the majority of people using VMS are not doing so for ultimate
> performance.

What is left of the VMS installed base is mostly the "installed base"
for whom migrating to a newer platform would be difficult because they
rely on some unique feature of VMS. And VMS no longer runs whole
corporations, it just runs those legacy application that could not be
migrated.

[Jan-Erik Soderholm]

JF Mezei wrote 2012-05-02 08:29:
> David Froble wrote:
>
>> My perception is that the majority of people using VMS are not doing so for ultimate
>> performance.
>
>
> What is left of the VMS installed base is mostly the "installed base"
> for whom migrating to a newer platform would be difficult because they
> rely on some unique feature of VMS. And VMS no longer runs whole
> corporations,

It does, at some places.

[Hans Vlems]

No, it's an EV5 manufactured in a process designed to produce the EV6
generation.
The EV56 is found in the Alpha Server 1200 and 4100 series, among
others, and in the latter ran at its highest clockspeed of 600 MHz
(IIRC).
In 1998 you'd surely have been impressed (given what Intel was doing
with the Pentium II and III at the time).
Today it is of interest to hobbyists like me ;-)
Alpha always was power hungry, and until the blades arrived I guess
nobody noticed or cared because compared to their predecessors
(mainframes) an Alpha wasn't that bad at all.
Today handhelds are more important if only because the potential
market is a lot bigger. Power is a function of die size and clock
speed and perhaps on a modern production line an EV5 or EV6 class
Alpha might be a very cost effective (read dirt cheap) way to produce
a 64 bit computing engine running at, say, 1.2 GHz without spending
years of development or paying royalties to others (something the
Chinese don't like doing anyway).
If you're looking for just such a solution then it might be feasible.
I wouldn't invest money in such a plan...
Hans

[Paul Sture]

On Wed, 02 May 2012 03:21:19 -0700, Hans Vlems wrote:

> No, it's an EV5 manufactured in a process designed to produce the EV6
> generation.
> The EV56 is found in the Alpha Server 1200 and 4100 series, among
> others, and in the latter ran at its highest clockspeed of 600 MHz
> (IIRC).
> In 1998 you'd surely have been impressed (given what Intel was doing
> with the Pentium II and III at the time). Today it is of interest to
> hobbyists like me ;-)

I bought a Pentium at the beginning of 1997 when 200 MHz was the latest
offering (I was advised to go for a twin cpu 133 MHz configuration
instead because that was substantially cheaper).

If I have interpreted the serial number correctly, my PWS 600au was made
in week 9 of 1997, putting it in the same time frame as my Pentium.

Alpha was streets ahead of Intel at the time. It definitely wasn't cheap
though. I did find a price of 20,000 USD for an Alpha 600au in that era,
but there wasn't a detailed listing of the configuration, so cannot say
definitively how much more expensive than the Pentium of the day they
were.

--
Paul Sture

[Bob Koehler]

In article <jnq1mq$no$1...@solani.org>, Michael Kraemer <M.Kr...@gsi.de> writes:
>
> I don't get it what magical design features the Alpha has, which
> would make it still attractive, even if it had been developed further.

> It's a design more than twenty years old, patents would have expired,
> and even by DEC's exaggerated claims it would reach its predicted
> lifetime in five years from now.
>

DEC's claims were based mostly on the state of the art when Alpha was
new. By the time alpha was abandoned DEC was putting things into
Alpha that I don't think DEC had considered when it was new.

DEC made mistakes, like not realizing how important byte and word
operations are to some applications. And later DEC fixed some of
those mistakes.

Processors have gotten faster since the day Alpha was abandoned,
but the notions folks had about future processors then don't seem to
have all panned out. EPIC isn't the dragon slayer. CISC x86 is
still competing with RISC.

When Alpha was built, it was THE fastest processor for many
applications, and competitive for all. I'm sure it's competitors
could make the same claim. If the advances applie to other
processors were made to Alpha it would still be competitive.

In addition to performance, DEC was looking at the doubling of
address space every 25 years. That hasn't happened, few applications
are stretching 64 bits.

[David Froble]

You seem to forget IBM and Power, which is I believe 64 bit (I don't know much about
Power, could be wrong) and as far as performance, was close to Alpha. It was the only CPU
that was competitive with Alpha in the past, and I doubt IBM has slacked off on
development, so might be the best performer today.

As for the itanic, are you using one? The process shrinks and on chip memory (or
something) has produced some decent performance. We now (I believe) have all of our
customers on IA-64 systems, performance is good, and the pricing of VMS is much lower than
it was with DEC.

The other side of that issue is we can now see the development efforts, or lack there-of,
that Intel makes when there is no competition. IA-64 in comparison to Intel's x86
products is at a virtual stand-still. Some had predicted this. Some feel IA-64 has
fulfilled Intel's real purpose, killing off competition. If so, then a valid question is
of what use is IA-64 to Intel? As much competition as they could kill off is gone, what
choice does anyone really have? It's Intel or count on your fingers, or the remaining
competition such as AMD which is still x86.

I do hate (hawk, spit) "good enough", but what are the options? At least we don't have
"not good enough and no options".

[David Froble]

Consider it lack of comprehension. Going from the number 32 to the number 64 is just one
more bit. But going from 32 bit numbers to 64 bit numbers just might be a bit more than
people normally can comprehend. Talking about such is cheap, but a recent discussion here
on c.o.v mentioned that it's not practical to physically get that much memory into a
computer. At least at today's densities.

64 bit data paths might be useful. 64 bit addresses probably aren't. Always might be the
exception.

[David Froble]

Bob Koehler wrote:
> In article <jnq1mq$no$1...@solani.org>, Michael Kraemer <M.Kr...@gsi.de> writes:
>> I don't get it what magical design features the Alpha has, which
>> would make it still attractive, even if it had been developed further.
>> It's a design more than twenty years old, patents would have expired,
>> and even by DEC's exaggerated claims it would reach its predicted
>> lifetime in five years from now.
>>
>
> DEC's claims were based mostly on the state of the art when Alpha was
> new. By the time alpha was abandoned DEC was putting things into
> Alpha that I don't think DEC had considered when it was new.

When the 11/780 was introduced, I think the CPU was 4 large boards plugged into the
backplane. 2 MB of memory worked. The microprocessor, CPU on one chip, really changed
things. It was basically miniaturization. Even though much smaller, it was much more
powerful. Shorter data paths with a constant on the speed of electricity is a big gain.
Still is in today's continued process shrinks.

The continued shrinking of the die size allowed additional things to be placed on the
chip, instead of on the motherboard, or additional boards, and finally to multiple cores
on one chip. DEC lead some of this with the on-chip memory controllers, inter-processor
communications and such. Then AMD used some of this technology and ended up dragging
Intel (relatively) out of the stone age.

The things added to the basic CPU were not Alpha specific, though the Alpha developers
seemed to "get it" ahead of others. Most likely none of this was imagined when the Alpha
design was implemented. Such is progress.

The original 25 year target could not imagine what might be developed during that time.
Some of those developments could affect the validity of the 25 year target, for better or
worse. For example, could an OoO design run in 22 nm process? Got no idea. Don't even
know what I'm talking about.

[JF Mezei]

Power started off 32 bité. Was upgraded to 64 later.

Sparc started off 32 bits , was upgraded later.

PA-Risc started off 32 bits. Was upgraded later.

8086 started off 16 bits, was upgraded to 32 bits and later 64.

Alpha started off 64 bits.

[glen herrmannsfeldt]

JF Mezei <jfmezei...@vaxination.ca> wrote:

(snip)

> 8086 started off 16 bits, was upgraded to 32 bits and later 64.

Actually, the 8086 is the 16 bit extension of the 8080.

The registers and instruction set were carefully designed to make
it easy to port 8080 code with minimal rewriting.

Some consider the 8080 as an extension of the 4004 or 4040, but
that might be stretching it too far.

-- glen

[JF Mezei]

David Froble wrote:

> The original 25 year target could not imagine what might be developed during that time.

I know that the words "Marketing" and "Digital" should never be used
together in the same paragraph, but in this case, I think the predicted
25 year lifespan was more of a marketing thing than some scientific
prediction.

Consider that Alpha's competitors were still 32 bits and it was not know
how long they would last, whereas Alpha was going to outlast them all
because it was already at 64 bits.

So the 25 year thing was more of an annoucement that "we're already at
64 bits, and our competitors will have to go through some transition to
64 bits later on so their architecture may change and disrupt you.

[glen herrmannsfeldt]

JF Mezei <jfmezei...@vaxination.ca> wrote:
> David Froble wrote:

>> The original 25 year target could not imagine what might
>> be developed during that time.

> I know that the words "Marketing" and "Digital" should never be used
> together in the same paragraph, but in this case, I think the predicted
> 25 year lifespan was more of a marketing thing than some scientific
> prediction.

Well, about that time S/360 was 25 years old, with extensions to
S/370 and XA/370. The idea that an architecture could last 25 years
was there.

> Consider that Alpha's competitors were still 32 bits and it
> was not know how long they would last, whereas Alpha was
> going to outlast them all because it was already at 64 bits.

VAX was the 32 bit extension of the PDP-11, which they didn't
decide to extend again. Probably the right decsision.

> So the 25 year thing was more of an annoucement that "we're already at
> 64 bits, and our competitors will have to go through some transition to
> 64 bits later on so their architecture may change and disrupt you.

I suppose, but also that, given history, it is harder to see farther
than 25 years ahead.

-- glen

[Arne Vajhøj]

On 5/2/2012 12:35 AM, David Froble wrote:
> When it was being competitively developed, nothing was faster than Alpha.
>
> As has been mentioned elsewhere, some people will pay for performance.
>
> To get to your question, you need to ask, "why was it fast?" Perhaps
> because the design was better than the competition? If so, and if there
> has not been any new designs that might be better, then a competitively
> developed Alpha might still be the fastest single processor available.
> When talking multiple cores, the "glue" on the chip, memory controllers,
> interprocessor communications, and such in EV7 and EV7z, even with the
> larger die size was doing things that out performed competing CPUs that
> had continued development and smaller die size.
>
> So, if the Alpha could be successfully shrunk down to say 35 nm, or even
> 22 nm which Intel has just released, and with a large on chip cache as
> the IA-64 has, perhaps it might still out-perform anything else available.
>
> Now, that is a mighty big "if", and what it would cost I have no idea,
> other than it wouldn't be cheap. So, commercially, might it be viable? I
> really have no idea. It would depend upon how many might pay for such
> performance.

A shrink to current technology could make it run at 3-4 GHz.

Or about 3-5 times faster than back then.

That is not enough.

But modify it from 1 to 8 cores per socket.

Then it will be 24-40 times faster than back then
(for multi thread/process usage).

It is starting to look like something.

Double all 3 levels of cache and do some optimizations
based on what is learned since then to double performance.

Then it will be 48-80 times faster than back then.

But the first item require a B$ investment in production
facility.

And the two last items require significant design work that
will take a lot of time (years).

Arne

[Arne Vajhøj]

On 5/2/2012 1:34 AM, JF Mezei wrote:
> Michael Kraemer wrote:
>> I don't get it what magical design features the Alpha has, which
>> would make it still attractive, even if it had been developed further.
>
> Alpha is the only native 64 bit architectire that worked and performed
> well. The only other native 64 bit is IA64 and we all knwo who well it
> TANKED !

They are the only architectures that did not have a 32 bit predecessor.

But having a 32 bit predecessor does not make a CPU less native 64 bit.

> Remember that 64 bit x86 came from the 8086. So there is a lot of
> baggage there that makes improving it harder.

Not true.

A x86-64 in 64 bit mode is not compatible with 32 bit so there
are no required baggage.

Arne

[Arne Vajhøj]

On 4/30/2012 10:52 PM, JF Mezei wrote:
> Arne Vajhøj wrote:
>> If there is a non commercial market, then I don't think South Korea
>> will sell to the closest ally of North Korea.
>
> HP would definitly donate Alpha IP to China in exchange for some favours
> from the chinese government to help HP produce in china and perhaps even
> sell its products in China.

Have all restrictions on selling IT to China been lifted?

Arne

[JF Mezei]

Arne Vajhøj wrote:

> Have all restrictions on selling IT to China been lifted?

If the USA doesn't forget about those restrictions, China could also
forget to buy so much of the USA government's debts.

The USA no longer has the luxury of criticising China because China is
using american's money to keep the USA afloat.

In other words, Americans send money to China. China then uses that
money to bail the USA government out with conditions attached.

People greatly underestimate the economic power of China.

[David Froble]

The real problem is, the cost per chip must pay for the development and production. Here
is where numbers matter. If you cannot sell enough chips, then the cost will be too high.

I was very disappointed to see the end of Alpha. But I'm not blind to the economic
realities. The CPUs going into tablets, phones, and such are a much better economic
investment than what might power a much smaller number of very powerful servers. Such is
reality.

[Hans Vlems]

JF, I think you are quite right there. A lot of production was shifted
to China because the wages were lower.
In a very short time that country has developed its internal economy
and knowledge base. Not too different
from what has happened in India and other "third world countries".
There are two differences though: the
internal market of China is huge and the Chinese government has shown
the ability to guide the development properly.
Put differently: start learning Chinese because in 10 years it'll be
taught in school as a second language.
Hans

[Hans Vlems]

Exactly my point Paul. The problem with US commercial thinking is that
"better equals cheaper".
Occasionally a good product demands a higher pricetag. Digital built
quality equipment and sold it at
a higher pricelevel than the competition. That quality allows me to
run VAX systems that are 20 years
old and these systems keep on going. The same applies for Alpha
systems and apparently we've all
clearly forgotten the awe when we first saw an Alpha booting VMS and
the speed our (VESTed!) applications
ran at! I remember the reactions of users of a library catalogue
program. It was written in Business Basic, an
interpreted language. I had VESTed the interpreter and the program ran
on an Alpha Server 2100/190.
On the VAX 6420 it took minutes just to display the program banner and
even more to get a command prompt.
On Alpha it was instantaneous. Users were sure the system was
faulty.... Everything worked well however and
two weeks later the users were quite used to the new performance.
The potential for Alpha in 1998 was enormous. Provided its owner was
willing to sink a lot of money in continuous
development and could find buying custumers willing to spend the
money. Customers vote with their wallets and
the alection was won by Intel's 64 bit incarnation of the Pentium
line. Dirt cheap and none of the systems based
on that technology last longer than 7 years.
Hans

[Michael Kraemer]

Hans Vlems schrieb:

> Put differently: start learning Chinese because in 10 years it'll be
> taught in school as a second language.

This is one of those predictions which pop up every now and then
and may or may not come true.
Back in the 1980s the motd was "lean production" and thus
"learn Japanese because they'll take over the world soon".
Hasn't happened, and I doubt it'll happen with Chinese.
Undoubtedly they have big economic success
(with the help of the capitalist Western countries, who
were stupid enough to outsource their industries),
but they also have big problems, which easily could
lead to the implosion of the system.

Btw, Chinese (as well as Japanese)
is already offered everywhere as a second language everywhere,
I even tried it myself thirty years ago, just for fun.
I don't think it will win, it's simply to old-fashioned
and inappropriate for the technical/scientific world.

[Michael Kraemer]

Hans Vlems schrieb:

> On the VAX 6420 it took minutes just to display the program banner and
> even more to get a command prompt.
> On Alpha it was instantaneous. Users were sure the system was
> faulty.... Everything worked well however and
> two weeks later the users were quite used to the new performance.

Comparing with a then ten years old VAX or an intel box
is low hanging fruit.

> The potential for Alpha in 1998 was enormous.

In 1998 the game was already over.

> Provided its owner was
> willing to sink a lot of money in continuous
> development and could find buying custumers willing to spend the
> money.

But that's the crucial point.
"Clean design" is worthless if it can't be manufactured
ecomonically and the customers just aren't interested.

[Paul Sture]

On Thu, 03 May 2012 10:12:26 +0200, Michael Kraemer wrote:

> Hans Vlems schrieb:
>
>> Put differently: start learning Chinese because in 10 years it'll be
>> taught in school as a second language.
>
> This is one of those predictions which pop up every now and then and may
> or may not come true.
> Back in the 1980s the motd was "lean production" and thus "learn
> Japanese because they'll take over the world soon". Hasn't happened, and
> I doubt it'll happen with Chinese.

I did a few months with a Japanese vehicle manufacturer in the UK. They
were very successful and I believe they still are. However, they did
that by starting a brand new facility rather than inheriting an existing
one (BMW tried that and failed).

The Japanese were certainly pioneers in Just In Time assembly and
properly managed that can work well, but I am sure others have copied
that now.

> Undoubtedly they have big economic
> success (with the help of the capitalist Western countries, who were
> stupid enough to outsource their industries), but they also have big
> problems, which easily could lead to the implosion of the system.

And that is where the danger lies. Can the West go back to manufacturing
if China does implode?

> Btw, Chinese (as well as Japanese)
> is already offered everywhere as a second language everywhere, I even
> tried it myself thirty years ago, just for fun. I don't think it will
> win, it's simply to old-fashioned and inappropriate for the
> technical/scientific world.

--
Paul Sture

[Paul Sture]

On Wed, 02 May 2012 23:42:26 -0700, Hans Vlems wrote:

> On 2 mei, 13:38, Paul Sture <p...@sture.ch> wrote:
>> On Wed, 02 May 2012 03:21:19 -0700, Hans Vlems wrote:
>> > No, it's an EV5 manufactured in a process designed to produce the EV6
>> > generation.
>> > The EV56 is found in the Alpha Server 1200 and 4100 series, among
>> > others, and in the latter ran at its highest clockspeed of 600 MHz
>> > (IIRC).
>> > In 1998 you'd surely have been impressed (given what Intel was doing
>> > with the Pentium II and III at the time). Today it is of interest to
>> > hobbyists like me ;-)
>>
>> I bought a Pentium at the beginning of 1997 when 200 MHz was the latest
>> offering (I was advised to go for a twin cpu 133 MHz configuration
>> instead because that was substantially cheaper).
>>
>> If I have interpreted the serial number correctly, my PWS 600au was
>> made in week 9 of 1997, putting it in the same time frame as my
>> Pentium.
>>
>> Alpha was streets ahead of Intel at the time. It definitely wasn't
>> cheap though.  I did find a price of 20,000 USD for an Alpha 600au in
>> that era, but there wasn't a detailed listing of the configuration, so
>> cannot say definitively how much more expensive than the Pentium of the
>> day they were.
>>
>

> Exactly my point Paul. The problem with US commercial thinking is that
> "better equals cheaper".
>
> Occasionally a good product demands a higher pricetag. Digital built
> quality equipment and sold it at a higher pricelevel than the
> competition. That quality allows me to run VAX systems that are 20 years
> old and these systems keep on going.

I started getting frustrated with this "better equals cheaper" thinking
about a decade ago. I didn't buy all my furniture from IKEA, nor did I
buy the cheapest vehicle on four wheels. Not the cheapest washing
machine either: I have learned that that can be a false economy.

Why should I suddenly reverse my purchasing habits for the special case
of computer gear?

> The same applies for Alpha systems
> and apparently we've all clearly forgotten the awe when we first saw an
> Alpha booting VMS and the speed our (VESTed!) applications ran at! I
> remember the reactions of users of a library catalogue program. It was
> written in Business Basic, an interpreted language. I had VESTed the
> interpreter and the program ran on an Alpha Server 2100/190.

Going the other way, Peter Weaver (IIRC) wrote a piece of DCL to convert
the VMS FAQ into VMS help file format. That ran well on Alpha, but I was
horrified at how long it took on a VAXstation 3100. That told me that on
an Alpha you could probably afford to write large chunks of code in DCL
or interpreted languages for one-off jobs, something you wouldn't have
dreamed of on a VAX.

> On the VAX 6420 it took minutes just to display the program banner and
> even more to get a command prompt.
>
> On Alpha it was instantaneous. Users were sure the system was faulty....
> Everything worked well however and two weeks later the users were quite
> used to the new performance. The potential for Alpha in 1998 was
> enormous. Provided its owner was willing to sink a lot of money in
> continuous development and could find buying custumers willing to spend
> the money. Customers vote with their wallets and the alection was won by
> Intel's 64 bit incarnation of the Pentium line. Dirt cheap and none of
> the systems based on that technology last longer than 7 years. Hans

Also in the mid to late nineties, customers could finance a large chunk
of their Alpha purchases by taking their old VAXen off maintenance.



--
Paul Sture

[ChrisQ]

On 05/02/12 01:12, Michael Kraemer wrote:

>
> I don't know what feeds your pipe dreams,
> but the Alpha never had a good track record in power efficiency.
> In fact, its inability to adapt to the needs of embedded devices
> (with the exception of AXPvme, which failed miserably)
> was one reason for its demise.
>

Sounds like rewriting history ?. I ran alpha machines for a decade
or more and don't remember them being any more power hungry than
equivalent x86 machines of the time, though they were much, much
faster.

With shrinks and improved process technology, i'm sure that Alpha
would have been more than competitvve with current designs, just
as ibm power is now for it's intended market.

Wonder how fast the last of the microvaxen would have been with
similar upgrades ?...

Regards,

Chris

[Bob Koehler]

In article <4fa17923$0$2204$c3e8da3$c8b7...@news.astraweb.com>, JF Mezei <jfmezei...@vaxination.ca> writes:
>
> 8086 started off 16 bits, was upgraded to 32 bits and later 64.
>

> Alpha started off 64 bits.

8086 inherits archiecture lessons from the 4 bit 4004 via the 8 bit
8008. But current IA32 inherits actual instructions from 8086. 8 bit
register instructions were part of the 32 bit 80386 and and 486, and
I think they are still supported.

Alpha inherits lessons learned from 16 bit PDP-11, 32 bit VAX, and
36 bit PDP-10. I suspect there are lessons learned from other PDP
and I think I see a few things from IBM 360. But it does not inherit
instructions from any of those processors.

[Michael Unger]

On 2012-05-03 01:55, "Arne Vajhøj" wrote:

> [...]

>
> But modify it from 1 to 8 cores per socket.
>
> Then it will be 24-40 times faster than back then
> (for multi thread/process usage).

Isn't exactly that quite normal for _server_ CPUs?

> [...]

Michael

--
Real names enhance the probability of getting real answers.
My e-mail account at DECUS Munich is no longer valid.

[David Froble]

Hans Vlems wrote:

> Exactly my point Paul. The problem with US commercial thinking is that
> "better equals cheaper".
> Occasionally a good product demands a higher pricetag. Digital built
> quality equipment and sold it at
> a higher pricelevel than the competition. That quality allows me to
> run VAX systems that are 20 years
> old and these systems keep on going. The same applies for Alpha
> systems and apparently we've all

The problem isn't so simple. Digital had a huge investment in people. Just about every
major city had an office for hardware support, and software support. At one time such was
needed. But over time the entire thing became obsolete, at least in comparison with what
the competition was offering. The cost of people also continued to rise.

No longer do computer manufacturers (for the most part) have local offices, and HW people
who come out to fix problems with occiliscopes and soldering irons.

I don't have any details to back up my guesses, but I've got to think that the cost of the
computers helped to support the local support that DEC provided.

Does Intel have local offices to support and repair their CPUs?

I have to think that one of the problems that DEC had was being able to change quickly
enough. There was no way to maintain some things that made DEC such a good vendor while
the competition was lean, mean, and usually not there when you needed support.

As for the superior products, I have to agree, since I'm still running 10-20 year old
equipment. I feel very lucky if I get a year out of current PC power supplies.

The downside of the DEC equipment is, when it finally fails, there will not be anything
off the shelf to replace failed pieces. I think I've reconciled myself that some day SIMH
will be part of my future. But today is not that day! :D

[Arne Vajhøj]

On 5/3/2012 11:21 AM, Michael Unger wrote:
> On 2012-05-03 01:55, "Arne Vajhøj" wrote:
>> [...]
>>
>> But modify it from 1 to 8 cores per socket.
>>
>> Then it will be 24-40 times faster than back then
>> (for multi thread/process usage).
>
> Isn't exactly that quite normal for _server_ CPUs?

It is.

Arne

[glen herrmannsfeldt]

ChrisQ <me...@devnull.com> wrote:

(snip, someone wrote)

>> I don't know what feeds your pipe dreams,
>> but the Alpha never had a good track record in power efficiency.

(snip)

> Sounds like rewriting history ?. I ran alpha machines for a decade
> or more and don't remember them being any more power hungry than
> equivalent x86 machines of the time, though they were much, much
> faster.

The problem with the comparison is that the power requirements
of just about everything except the processor have gone down over
the years. In the PC/AT days, it was the disk drive that used
most of the power in the box. (At least relative to the processor.)

> With shrinks and improved process technology, i'm sure that Alpha
> would have been more than competitvve with current designs, just
> as ibm power is now for it's intended market.

It might take more than that, but should be possible.

Many now turn off functional units, such as floating point,
when they are not being used.

> Wonder how fast the last of the microvaxen would have been with
> similar upgrades ?...

There were problems with VAX. The 512 byte page was already too
small just about when it came out. The very large number of different
instruction sizes, and the complication with decoding them, also
made it hard. (I believe somewhat worse than IA32.)

-- glen

[glen herrmannsfeldt]

Bob Koehler <koe...@eisner.nospam.encompasserve.org> wrote:

(snip)

> 8086 inherits archiecture lessons from the 4 bit 4004 via the 8 bit
> 8008. But current IA32 inherits actual instructions from 8086. 8 bit
> register instructions were part of the 32 bit 80386 and and 486, and
> I think they are still supported.

The 8086 inherits instructions from the 8080, but the actual opcodes
have changed. You can map the 8080 registers onto the appropriate
8086 registers, and rewrite assembly source using new assembly
opcodes, assemble the result, and run it.

> Alpha inherits lessons learned from 16 bit PDP-11, 32 bit VAX, and
> 36 bit PDP-10. I suspect there are lessons learned from other PDP
> and I think I see a few things from IBM 360. But it does not inherit
> instructions from any of those processors.

I think I believe that. Without going through them one at a
time, though, I wouldn't be so sure that there weren't any.

Using only 32 and 64 bit load/store made it pretty different
from the previous processors.

-- glen

[John Wallace]

Alphas weren't notably more power hungry than comparable x86 from the
same era. I'll offer an EV4 and an EV5 system as examples.

The EV4 example is the AlphaStation 400, which was a EV4/21064 (at up
to 233MHz if I remember rightly). The very same box and the very same
power supply and cooling was available with different part numbers and
processor daughterboard as either a high(ish) end x86 desktop (the
Celebris XL, which went all the way up to a dual Pentium 166 or a
Pentium Pro) or an entry level AlphaStation (or with the same
hardware but different software and licensing, as an AlphaServer).

The EV5 system I'm thinking of was the Personal Workstation (Miata
etc) with an EV5 at up to 600MHz, again sold in NT-only and more
expensive Unix/VMS-licenced versions. These again shared a box and
power supply and pretty much everything except the processor
daughtercard (in industry standard but not widely used NLX format)
with their x86 equivalent.

I have an unconfirmed suspicion that something similar went on with
the AlphaServer 1000 (Mikasa) but I can't remember the name of the
corresponding PCBU/Intel server box. Supporting evidence (or
correction) welcome.

These were not cheap bog-standard low-end systems, either in x86 or
Alpha flavours, but the NT versions of the Alpha systems weren't
wildly different in price from their x86 relatives, and performance-
senstive customers I dealt with were happy to pay a little extra for
the extra performance, and those using them for VMS or Tru64 were also
happy that they had an affordable reasonable-performance VMS/UNIX box,
quite competitive with other RISC boxes of the era (though courtesy of
generally relying on "the channel" to do marketing and promotion, not
many prospective customers actually knew this).

So, on the subject of power consumption, I'd say Chris's recollection
is perfectly reasonable and supported by the evidence.

[Arne Vajhøj]

I would think that x86-64 is similar to x86 in somewhat
the same way as VAX was similar to PDP-11.

Alpha and VAX do not really share any characteristics. All
the VAX specific stuff was done as PALcode.

Arne

[David Froble]

glen herrmannsfeldt wrote:

>> Wonder how fast the last of the microvaxen would have been with
>> similar upgrades ?...
>
> There were problems with VAX. The 512 byte page was already too
> small just about when it came out. The very large number of different
> instruction sizes, and the complication with decoding them, also
> made it hard. (I believe somewhat worse than IA32.)
>
> -- glen

I believe that the N-VAX microprocessor addressed those issues rather well. It was (if my
memory and facts are correct) basically a RISC processor that implemented the VAX
instruction set.

There are also some issues with expectations. I have no idea what type of processing you
do, or are referring to when you claim there were problems with VAX. However, as the
current unknown demand for old MV3100 systems seems to indicate, there are those who could
get along just fine with the then current levels of performance.

It makes me wonder how the product line would have prospered if DEC has made the business
decision to continue the VAX product line, not the large ones, (but I could be mistaken
with that choice), just the MV3100 type of systems. Go for low cost to manufacture,
generic components, upgrades as computer components changes, ie; S-ATA disks, memory
SIMMs, power supplies, and such. Sell the OS at some minimal price, or just include it
with the system. Perhaps it would have stopped some customers from migrating to PC based
systems, if they had some confidence in being able to continue to get new systems for
their needs.

For those who have seen a MV3100 model 98, you might notice some similarities to a
mid-tower PC ....

[Arne Vajhøj]

> The real problem is, the cost per chip must pay for the development and
> production. Here is where numbers matter. If you cannot sell enough
> chips, then the cost will be too high.
>
> I was very disappointed to see the end of Alpha. But I'm not blind to
> the economic realities. The CPUs going into tablets, phones, and such
> are a much better economic investment than what might power a much
> smaller number of very powerful servers. Such is reality.

The fixed costs is very large and increasing relative to the
variable cost for CPU's.

This will inevitable lead to fewer CPU producers.

Intel and AMD will survive because they sell desktop
CPU's in the 2 or 3 digit millions per design.

I do not have much confidence in Power or SPARC long term.

Arne

[Arne Vajhøj]

On 5/3/2012 5:54 PM, David Froble wrote:
> It makes me wonder how the product line would have prospered if DEC has
> made the business decision to continue the VAX product line, not the
> large ones, (but I could be mistaken with that choice), just the MV3100
> type of systems. Go for low cost to manufacture, generic components,
> upgrades as computer components changes, ie; S-ATA disks, memory SIMMs,
> power supplies, and such. Sell the OS at some minimal price, or just
> include it with the system. Perhaps it would have stopped some customers
> from migrating to PC based systems, if they had some confidence in being
> able to continue to get new systems for their needs.

It would have reduced Alpha sales, which at the time of the
decision had a chance.

And in the end it would not have made much of a difference. VMS's
problem is not that VAX is no longer. Applications are what matters.

Arne

[Arne Vajhøj]

On 5/2/2012 10:57 PM, JF Mezei wrote:
> Arne Vajhøj wrote:
>
>> Have all restrictions on selling IT to China been lifted?
>
> If the USA doesn't forget about those restrictions, China could also
> forget to buy so much of the USA government's debts.
>
> The USA no longer has the luxury of criticising China because China is
> using american's money to keep the USA afloat.
>
> In other words, Americans send money to China. China then uses that
> money to bail the USA government out with conditions attached.

It is a myth that the US government is funding most of
its huge deficit directly from China. It is less than 10%
that has come from China.

China could stop accepting dollars tommorrow if they wanted to.

It would cause Chinese manufacturing to tank and US ditto to soar.

Sure US consumers would see stores with US goods twice as
expensive as the Chinese goods there today, but ...

That threat is not really a threat.

Arne

[David Froble]

Do be careful when betting against IBM.

Got some IBM stock that was purchased at $104 and 2 days ago it was at $206

The thing with IBM is that they are not just a semiconductor mfg. They can afford to take
a loss in one department, if that allows other departments to do well. If Power gives
them an edge in sales, perhaps that's better for the company than just being another "me too".

DEC could have been the same, but poor management destroyed them.

[David Froble]

Ok, applications matter.

Consolidated Data has a rather extensive application that runs on VMS. In my opinion, it
cannot be ported to any other OS. It can be re-written, with much expense, and loss of
features most likely. But not ported. Quite frankly, maybe I'm just stupid, but I cannot
see any justification for blowing $50 million and not getting anything for doing so.

There used to be many manufacturing customers that used VMS. There must have been reasons
why they used VMS. I'm sure that they were not interested in moving to another OS, if
they had any choice. It wasn't cheap, and it wasn't the business they were/are in.

As for the "reduce Alpha sales" bullshit, I've read it before, and it's just as stupid, or
more, now as it was then. When a vendor starts telling a customer what they're going to
get, instead of asking what they want, that vendor will soon have no customers.

So, from a particular perspective, what killed Alpha? Perhaps VAX ??

[JF Mezei]

Big existentialist question that may keep me awake at night:

Will the Chinese Alphas have english/ASCII SRM or will be all chinese ?

[Arne Vajhøj]

> Do be careful when betting against IBM.
>
> Got some IBM stock that was purchased at $104 and 2 days ago it was at $206
>
> The thing with IBM is that they are not just a semiconductor mfg. They
> can afford to take a loss in one department, if that allows other
> departments to do well. If Power gives them an edge in sales, perhaps
> that's better for the company than just being another "me too".

IBM is a business.

If they make money on Power CPU's then they will make them. If not
then they will drop them.

IBM will still make a ton of money on all the other things
they do.

Arne

[JF Mezei]

David Froble wrote:

> As for the "reduce Alpha sales" bullshit, I've read it before, and it's just as stupid, or
> more, now as it was then. When a vendor starts telling a customer what they're going to
> get, instead of asking what they want, that vendor will soon have no customers.

In the Apple bible (the book of Jobs), Steve Jobs says something very
important: If you are not willing to cannabalise your own products with
a better/different one, someone else will.

This became very important when phones started to include built-in music
players and cameras. Apple coming out with its onw phone would
cannabalise sales of ipods, but the money would still stay at Apple. Not
doing so meant the money would go to SonyEriccson/Nokia/Motorola and the
then nascent asian players.

Digital's insistence that smaller system not be able to perform the
duties of the bigger systems (so as to not cannabalise sales of the more
profitable bigger machines) meant that customers simply left digital.

The problem with VAX to Alpha transition isn't Alpha, but rather the
fact that Digital not only failed to port all of its software portfolio,
but also failed to provide a built-in software emulator or one time
binary translator.


Consider that Apple has been able to do 2 platform changes very
succesfully (MacOS between 86k and PowerPC, and OS-X between PowerPC and
32 bit 8086 and later 64 bit) The last transition involved Rosetta
which was able to very smartly emulate PPC on an Intel, including the
conversion between big and litle endian.

Had DEC provided VEST with all alpha systems and perhaps spruced it up a
bit, allowing the vesting of those packages that DEC was not porting
itself (such as message router etc), DEC would have been able to retain
many more customers through the transition.

The added performance of Alpha did not compensate for the business
practice problems for licencing and recompiling (and for C, updatig from
VAXC to DECC)

[JF Mezei]

Arne Vajhøj wrote:

> If they make money on Power CPU's then they will make them. If not
> then they will drop them.
>

Not that simple. IBM makes money from its chip fabbing business. And the
sell plenty of PowerPC chips to the game console makers.

So the Power chip, *IF* not profitable, could very well be like its "Big
Blue" computer that plays chess and car makers building formula 1 racing
cars: advertising to show how capable IBM is.

[glen herrmannsfeldt]

David Froble <da...@tsoft-inc.com> wrote:

(snip, I wrote)

>> There were problems with VAX. The 512 byte page was already too
>> small just about when it came out. The very large number of different
>> instruction sizes, and the complication with decoding them, also
>> made it hard. (I believe somewhat worse than IA32.)

> I believe that the N-VAX microprocessor addressed those issues
> rather well. It was (if my memory and facts are correct)
> basically a RISC processor that implemented the VAX
> instruction set.

You can do that, but it is much harder than evan IA32.

> There are also some issues with expectations. I have no idea
> what type of processing you do, or are referring to when you
> claim there were problems with VAX.

In the VAX days, scientific number crunching.

> However, as the current unknown demand for old MV3100 systems
> seems to indicate, there are those who could get along just
> fine with the then current levels of performance.

I suppose, but as others came along cheaper people would tend
to start using them.

(snip)

-- glen

[Michael Kraemer]

JF Mezei schrieb:

But even this strategy has limits. Development of the "big" Powers
is costly, probably as much as Alpha was. If it turns out to be too
much, they'll drop it without regret. What keeps the platform alive,
is the fact that there are a lot of Power derivatives which keep
the fabs running. For example, the Nintendo Wii alone translates
into 90 million PPCs since 2006 or so.
Add the Xbox, the PS, and all that invisible embedded stuff
and one might end in annual sales of
the order of all PowerMacs ever sold.

So this is similar to intel's Itanic/x86 model.
(with the exception that Power is a much better chip of course :-)

[Michael Kraemer]

JF Mezei schrieb:

> David Froble wrote:
>
>
>>As for the "reduce Alpha sales" bullshit, I've read it before, and it's just as stupid, or
>>more, now as it was then. When a vendor starts telling a customer what they're going to
>>get, instead of asking what they want, that vendor will soon have no customers.
>
>
>
> In the Apple bible (the book of Jobs), Steve Jobs says something very
> important: If you are not willing to cannabalise your own products with
> a better/different one, someone else will.

But DEC did just that: one of their few wise decisions
was to start offering their RISC/Unix in 1989/90,
just in time to keep a significant chunk of technical
computing going with DEC. Otherwise those customers
would have been lost to Sun, SGI, HP, IBM etc.
Too bad they let it die on the vine,
in favour of that unproven Alpha thing.

> Digital's insistence that smaller system not be able to perform the
> duties of the bigger systems (so as to not cannabalise sales of the more
> profitable bigger machines) meant that customers simply left digital.

I think customers were smarter.
Of course DEC wanted them to spend $$$$$$ on VAX 6000 or 9000,
but (un)fortunately there were also the VS 4000's, which could
do almost the same job for $$$$$

[Paul Sture]

On Thu, 03 May 2012 20:00:53 +0000, glen herrmannsfeldt wrote:

> Many now turn off functional units, such as floating point, when they
> are not being used.

A lot of cheaper end stuff does that nowadays too. I ran some diagnostic
or other on my HP Microserver (AMD inside) when I first got it and could
see it downclocking itself under a light load.

--
Paul Sture

[MG]

On 3-5-2012 10:12, Michael Kraemer wrote:
> [...] Western countries, who were stupid enough to
> outsource their industries

That sums it up nicely.

- MG

[Paul Sture]

On Thu, 03 May 2012 23:20:04 -0400, JF Mezei wrote:

> Digital's insistence that smaller system not be able to perform the
> duties of the bigger systems (so as to not cannabalise sales of the more
> profitable bigger machines) meant that customers simply left digital.

I see that very same mistake today with Microsoft's products.

--
Paul Sture

[Paul Sture]

On Thu, 03 May 2012 12:50:38 -0400, David Froble wrote:

> The problem isn't so simple. Digital had a huge investment in people.
> Just about every major city had an office for hardware support, and
> software support. At one time such was needed. But over time the
> entire thing became obsolete, at least in comparison with what the
> competition was offering. The cost of people also continued to rise.
>
> No longer do computer manufacturers (for the most part) have local
> offices, and HW people who come out to fix problems with occiliscopes
> and soldering irons.
>
> I don't have any details to back up my guesses, but I've got to think
> that the cost of the computers helped to support the local support that
> DEC provided.

We had monthly Preventative Maintenance on our first PDP 11/34, performed
by the DEC office in our neighbouring city. We were very pleased when DEC
decided to change that to once every 3 months, as not only was it a
disruption to our work, but the system often failed a few hours after a
PM, sometimes as soon as the engineer had driven off.

--
Paul Sture

[Bob Koehler]

In article <jnuol3$1db$1...@speranza.aioe.org>, glen herrmannsfeldt <g...@ugcs.caltech.edu> writes:
>
> Using only 32 and 64 bit load/store made it pretty different
> from the previous processors.

Which is why I mentioned PDP-10. Almost had only 18 bit and 36 bit
load/store, but also had bit field load/store (PDP-10 called them
user defined "bytes").

When Alpha architecture was first revealed at a national symposium,
and we found out it was 32 and 64 bit load/store some folks got up
and thanked DEC for remembering PDP-10 design principles.

[Bob Koehler]

In article <4fa2eef6$0$288$1472...@news.sunsite.dk>, =?ISO-8859-1?Q?Arne_Vajh=F8j?= <ar...@vajhoej.dk> writes:
>
> I would think that x86-64 is similar to x86 in somewhat
> the same way as VAX was similar to PDP-11.

I was under the impression that adding 64 bit to x86 was something
like adding 23 bit to PDP-10, or 32 bit to 8086. Existing code
continues to run.

VAX to PDP-11 was different. The VAX instruction set is not the
PDP-11 instruction set, although early VAXen had two instruction
sets: VAX and compatability mode (a subset of PDP-11). CHMU was
used to switch to the compatability mode instruction set when
running PDP-11 user mode code.

[Bob Koehler]

In article <jnvf63$kaa$1...@dont-email.me>, David Froble <da...@tsoft-inc.com> writes:
>
> There used to be many manufacturing customers that used VMS. There must have been reasons
> why they used VMS. I'm sure that they were not interested in moving to another OS, if
> they had any choice. It wasn't cheap, and it wasn't the business they were/are in.

The reason they used VMS might have been the same reasons folks now
use Windows or UNIX.

While there are some applications that can't survive Windows or UNIX;
back in the 80's if you had a problem and needed a computer to solve
it, you just went out and bought a VAX.