Intel cowers in the face of AMD legal strategy

[Robert Myers]

http://www.pcpro.co.uk/news/76418/analysis-whats-behind-the-chipset-shortage.html

<quote>

However, another consideration for Intel is the dynamic in the market
among chipset vendors. To counter Intel's 'platformization' strategy,
which integrates processors, chipsets and network chips into one
platform, motherboard makers hinted that AMD plans to introduce a
program called the 'Commercial Stable Image Platform (CSIP) to the
enterprise PC segment in September. Since AMD does not produce
chipsets, it needs the full support of chipset vendors, including VIA
Technologies, SiS, ULi, ATI Technologies and Nvidia, to expand its CPU
market share worldwide.

By exiting the entry level segment Intel is reducing itself as a threat
to Taiwan chipset makers, and providing them more opportunities to
expand their market presence on the Intel platform. To prevent AMD from
gaining more share, Intel is opening the door for chipset vendors to
increase their chipset production, as well as their market focus, on
the Pentium 4 platform.

</quote>

Once a predator, always a predator, I guess. Everything adds up to
100%, and what Intel gets, AMD doesn't get. Isn't there an agency,
perhaps outside the US, to deal with this sort of thing?

RM

 

[Robert Myers]

I believe mr myers was being sarcastic and did not mean to be taken
seriously.

No, I did not mean to be taken seriously.

As to the morality of the behavior, if the strategy is accurately
described, I don't see it as any less predatory than what are claimed
to be Intel's pricing strategies. The pricing strategies are claimed
to be intended to deny AMD markets, and the claimed strategy of
choosing what to produce is intended to deny AMD collaborators.

If the analysis is accurate, it is an example of Intel using its market
dominance to make it more difficult for a competitor to do business.

And my sarasm about having an agency to regulate the behavior was aimed
at illustrating the futility of having agencies to enforce fair play.

RM

 

[George Macdonald]

On what planet and in what universe is this purported "behaviour",
even if accurate, require an "agency" to deal with it? What smallest
smidgen of this "behaviour" is by whatever torturous reasoning illegal
or immoral or unethical? I really want to know the answer to this,
Robert.

It's just another iTurd Felger. He can't take the heat but he can't resist
either.... sad really.

 

[Robert Myers]

It's just another iTurd Felger. He can't take the heat but he can't resist
either.... sad really.

It probably did reflect bad judgment on my part to make that post.
Just among other things, I'm at a point in my life where I really don't
want to have conversations that include bathroom jokes.

What I posted is a much more plausible explanation for why Intel would
have left the low end chipset business than anything else I've read,
although I'll admit to having given up on that thread since it has
degenerated into mutual abuse.

Far from Intel seeming to be intimidated by much of anything that's
happened, it seems to be pursuing its usual take-no-prisoners style of
business.

Someone in the business I spoke with recently seems to have taken an
Inquirer story

http://www.theinquirer.net/?article=25496

seriously. It, like what I take to be another rumor posted here, about
Intel having onboard Java support, would be fascinating, if true, but I
doubt if it's even remotely true.

Sad? Yeah, a little. My take is that the business is no longer driven
by technology. The technologists here are all google-eyed over AMD's
technology decisions, which really seem to have been shrewder than
Intel's. That's all neat, but not neat enough to drive the business.
Intel apparently stopped thinking technology long ago, they are
dominating the business, and money poured into lawsuits isn't going to
put any exciting technology on the table or change Intel's aggressive
posture, which was the point of the post.

Now, Intel is doing its best to pump up the rumor mill prior to IDF
about a new low-power strategy. In retrospect, the Apple announcement
and Jobs' remarks about performance per watt were just part of that
publicity build-up. I'm hoping they're going to put something new and
exciting on the table, but recent experience tells me not to prepare to
be less than amazed. At least they've got the correct figure of merit,
which is performance per watt.

I'm posting in a forum where there's a pluarality for carrying
microprocessor technology into the next generation on the shoulders of
lawyers. If that isn't sad, I don't know what is.

RM

 

[George Macdonald]

It probably did reflect bad judgment on my part to make that post.
Just among other things, I'm at a point in my life where I really don't
want to have conversations that include bathroom jokes.

What I posted is a much more plausible explanation for why Intel would
have left the low end chipset business than anything else I've read,
although I'll admit to having given up on that thread since it has
degenerated into mutual abuse.

Pity that - if you'd read the thread, and weren't quite so quick on the
draw, you'd have seen that that explanation had already been covered.

My take is that Intel just doesn't want to have their technology,
especially a "platform", being sold for $299./per system - it just doesn't
fit a platform strategy. OTOH, leaving the low-end to 3rd party chipsets
ruins their "single image" advantage in the business low-end... and they've
already ceded the gaming high-end to nVidia. It's going to be interesting
to hear just what it is that Otellini is going to pull from his magic hat
in the next couple of weeks.

Far from Intel seeming to be intimidated by much of anything that's
happened, it seems to be pursuing its usual take-no-prisoners style of
business.

Someone in the business I spoke with recently seems to have taken an
Inquirer story

http://www.theinquirer.net/?article=25496

seriously. It, like what I take to be another rumor posted here, about
Intel having onboard Java support, would be fascinating, if true, but I
doubt if it's even remotely true.

I don't see what's so exciting about Java-assist - nuthin' new there.
As for the Elbrus connection and VLIW, I hope this is not Merced-redux.

Sad? Yeah, a little. My take is that the business is no longer driven
by technology. The technologists here are all google-eyed over AMD's
technology decisions, which really seem to have been shrewder than
Intel's. That's all neat, but not neat enough to drive the business.
Intel apparently stopped thinking technology long ago, they are
dominating the business, and money poured into lawsuits isn't going to
put any exciting technology on the table or change Intel's aggressive
posture, which was the point of the post.

We've heard all this guff from you before... umpteen times. Repeating it
won't make AMD and their "technology", which you apparently refuse to even
sample, go away.

Now, Intel is doing its best to pump up the rumor mill prior to IDF
about a new low-power strategy. In retrospect, the Apple announcement
and Jobs' remarks about performance per watt were just part of that
publicity build-up. I'm hoping they're going to put something new and
exciting on the table, but recent experience tells me not to prepare to
be less than amazed. At least they've got the correct figure of merit,
which is performance per watt.

I'm posting in a forum where there's a pluarality for carrying
microprocessor technology into the next generation on the shoulders of
lawyers. If that isn't sad, I don't know what is.

That is utter BS and you know it; hell just above, you are blaming people
for being "google-eyed" by technology - make up your mind. For obvious
reasons, you have never been frustrated by the lack of availability of
AMD64 systems in the mainstream market, both desktop and notebook. Now
that we know precisely why -- the dominant player has been running a
protection racket -- I for one am hoping that the situation can be
repaired; if it takes lawyers and court cases, so be it. A better product
just deserves recognition for what it is.

 

[Robert Myers]

Robert Myers wrote:


I don't see what's so exciting about Java-assist - nuthin' new there.
As for the Elbrus connection and VLIW, I hope this is not Merced-redux.

I think people who write for the inquirer probably smoke alot of weed.
Management knows it and doesn't mind because they know how to get stuff
on slashdot, which is good for business.

Java support is nothing new? Since when?

We've heard all this guff from you before... umpteen times. Repeating it
won't make AMD and their "technology", which you apparently refuse to even
sample, go away.

I expect the "supercomputer" of the future to be a cluster with "Intel
Inside" labels all over the place. Whether it's x86 or Itanium I'd
hesitate to place a bet right now.

All prognostications about technology are risky, and when I find out
just how wrong I was, it'll be interesting to look back and try to see
what it was I missed I why I should have seen it.

In the meantime, the figure of merit that matters is, indeed,
performance per watt, and, while I've never been a big admirer of Steve
Jobs, he's gotten to see what everybody's got, and he thinks Intel has
the strategy of the future. Without that data point, I might think
Power architecture stood a chance. If Jobs' claims are anywhere near
correct, I don't think Power stands a chance.

Opteron has all the right stuff for supercomputing, and it *still* can
barely crack the Top 500. Based on what I've seen so far, even if AMD
managed to meet or beat Intel's new low power offerings, they still
wouldn't have much of a future.

Markets don't care what I think. They do whatever they want to do
without asking me, but my bet is that Intel is the future for the kind
of computing I do. You're free to put your own energies wherever you
think best.

And I don't like lawyers.

RM

 

[Del Cecchi]

I think people who write for the inquirer probably smoke alot of weed.
Management knows it and doesn't mind because they know how to get stuff
on slashdot, which is good for business.

Java support is nothing new? Since when?


I expect the "supercomputer" of the future to be a cluster with "Intel
Inside" labels all over the place. Whether it's x86 or Itanium I'd
hesitate to place a bet right now.

Not gonna be Itanium. And Intel has dropped the SC business and appears
to have no interest. Of course one could plug a bunch of PCIX IB HCA
cards into some cheap dell boxes.... like the guy did with the G5 Macs.
Cray picked Opterons and IBM is using Power or Intel X86, depending.

All prognostications about technology are risky, and when I find out
just how wrong I was, it'll be interesting to look back and try to see
what it was I missed I why I should have seen it.

That's the trouble with the future. It's so hard to predict.

In the meantime, the figure of merit that matters is, indeed,
performance per watt, and, while I've never been a big admirer of Steve
Jobs, he's gotten to see what everybody's got, and he thinks Intel has
the strategy of the future. Without that data point, I might think
Power architecture stood a chance. If Jobs' claims are anywhere near
correct, I don't think Power stands a chance.

? What is it about power that makes it less efficient than x86? One
could design a Power M if one chose. We could go into my opinion about
why Jobs did what he did, but I like to eat.

Opteron has all the right stuff for supercomputing, and it *still* can
barely crack the Top 500. Based on what I've seen so far, even if AMD
managed to meet or beat Intel's new low power offerings, they still
wouldn't have much of a future.

There is no reason that AMD shouldn't surpass Intel in low power should
they choose to. SOI has some advantage that way. And the design
techniques are not trade secrets.

Markets don't care what I think. They do whatever they want to do
without asking me, but my bet is that Intel is the future for the kind
of computing I do. You're free to put your own energies wherever you
think best.

What kind of computing do you do? If it is top end HPC, then the future
is likely very large clusters of special purpose hardware ala blue gene.
If it is fair to middling HPC on a shoestring then clusters of white
boxes or blades with intel or opteron as seems cheapest at teh moment.
If it's high availability transaction processing, a nice Z or I series
will do the job.

del

 

[George Macdonald]

I think people who write for the inquirer probably smoke alot of weed.
Management knows it and doesn't mind because they know how to get stuff
on slashdot, which is good for business.

Java support is nothing new? Since when?
Sun?


I expect the "supercomputer" of the future to be a cluster with "Intel
Inside" labels all over the place. Whether it's x86 or Itanium I'd
hesitate to place a bet right now.

The party's over. When is it going to sink in?

All prognostications about technology are risky, and when I find out
just how wrong I was, it'll be interesting to look back and try to see
what it was I missed I why I should have seen it.

In the meantime, the figure of merit that matters is, indeed,
performance per watt, and, while I've never been a big admirer of Steve
Jobs, he's gotten to see what everybody's got, and he thinks Intel has
the strategy of the future. Without that data point, I might think
Power architecture stood a chance. If Jobs' claims are anywhere near
correct, I don't think Power stands a chance.

Jobs is a technical chump... and he's just been chumped again. AMD64 is
leagues ahead of Intel on performance per watt and has been for ~2 years.
My favorite quote from this recently discussed Intel presentation
http://www.macworld.com/news/2005/08/17/dualcore/index.php is "We were

Opteron has all the right stuff for supercomputing, and it *still* can
barely crack the Top 500. Based on what I've seen so far, even if AMD
managed to meet or beat Intel's new low power offerings, they still
wouldn't have much of a future.

That is again, utter BS. In fact I gave you a reference very recently
which showed that in MP super computing, AMD was at the top - you never
listen! Your "opinions" are all based on your prejudiced personal picture
of computing and have no basis in documented fact.

Markets don't care what I think. They do whatever they want to do
without asking me, but my bet is that Intel is the future for the kind
of computing I do. You're free to put your own energies wherever you
think best.

And again we differ - that's allowed... as I've pointed out umpteen times.

And I don't like lawyers.

How fashionable!

 

[Robert Myers]

"Robert Myers" <[email protected]> wrote in message

Not gonna be Itanium. And Intel has dropped the SC business and appears
to have no interest.

If Intel ever was interested in supercomputers, it was only from the
point of view of publicity, and I doubt if their interest in publicity
has changed--any more than IBM's. I read recently about a comment from
Intel that they're not at all unhappy with the Top 500 benchmark or the
attention the Top 500 list gets. Of course not. They can pile up
boxes faster than anyone else on the planet.

Of course one could plug a bunch of PCIX IB HCA
cards into some cheap dell boxes.... like the guy did with the G5 Macs.
Cray picked Opterons and IBM is using Power or Intel X86, depending.

That's more or less what's happening, and that's more or less what's
going to continue to happen, except that the IB won't necessarily be
going through PCIX.

? What is it about power that makes it less efficient than x86?

I don't think there is anything about Power that makes it less
efficient than x86. If anything, I'd guess the advantage would be the
other way around.

One could design a Power M if one chose. We could go into my opinion about
why Jobs did what he did, but I like to eat.

I'm sure IBM could design lots of things. The question is, will they?

My comment wasn't based on Jobs deciding to go with Intel. It was
based on the claims of relative performance per watt. Integer
performance, IIRC, but that's what most bioclusters are doing anyway,
and it's not irrelevant as a clue to what relative floating point
performance might be.

As to why Apple went with Intel rather than IBM, I think IBM and AMD
have the same problem: they just can't come close to Intel in terms of
the variety of specialized solutions they can afford to develop. IBM
hasn't been able to provide what Apple has thought it needed and Intel
has promised them the moon.

Even discounting a significant part of the guessing as blue smoke and
mirrors, it all confirms what I see as the big picture: the only thing
that matters in the long run is cash flow because the capital needs of
the business just keep climbing. Not quite the same thing happened in
automobiles, but more or less exactly the same thing happened in
commercial aircraft (and jet engines, for that matter).

There is no reason that AMD shouldn't surpass Intel in low power should
they choose to. SOI has some advantage that way. And the design
techniques are not trade secrets.

The only reason is money.

I'll happily admit that what Intel has shown recently, aside from the
ability to generate cash, is the apparent ability to squander it. From
the point of view of sentimental favorites, AMD looks great: scrappy
little competitor does more with less. Nevertheless, I'm following
Deep Throat's advice and following the money.

What kind of computing do you do? If it is top end HPC, then the future
is likely very large clusters of special purpose hardware ala blue gene.

What Blue Gene has going for it is low power operation. When low power
clusters are purchased as a commodity in terms of $/performance, IBM is
going to have a tough time of it.

RM

 

[Robert Myers]

Sun?

And since when does Sun make x86? Since when, for that matter, does
Sun make a chip that matters to any but a shrinking base of existing
customers?

Jobs is a technical chump... and he's just been chumped again. AMD64 is
leagues ahead of Intel on performance per watt and has been for ~2 years.
My favorite quote from this recently discussed Intel presentation
http://www.macworld.com/news/2005/08/17/dualcore/index.php is "We were
behind".<gulp>

That's all very well, but the problem wasn't that IBM couldn't sell
chips to Apple for servers, the problem was that they couldn't sell
chips for mobile operation, a market in which Intel has been doing very
nicely.

And, at that, Jobs didn't (apparently) base his decision on what's on
the table now. Intel has plenty on the table in terms of low power
operation to be credible, and Intel clearly promised much better things
to come, with the focus on performance/watt. That's just what Jobs
wanted to hear.

That is again, utter BS.

There's no BS. All you have to do is get a Top 500 list and count, on
your fingers if you have to. My disdain for the Top 500 list is well
known, but I'd have to be persuaded that what it seems to be saying
about who owns the market, both in terms of who is currently dominant
and what seems to be an unmistakeable trend, is irrelevant.

RM

 

[chrisv]

And, at that, Jobs didn't (apparently) base his decision on what's on
the table now. Intel has plenty on the table in terms of low power
operation to be credible, and Intel clearly promised much better things
to come, with the focus on performance/watt. That's just what Jobs
wanted to hear.

Oh, I'm sure they told him exactly what he wanted to hear. Salesmen
tend to do that.

 

[George Macdonald]

Oh, I'm sure they told him exactly what he wanted to hear. Salesmen
tend to do that.

And of course, salesmen, like Jobs, are ironically easily conned - funny
that.

 

[George Macdonald]

George Macdonald wrote:
That's all very well, but the problem wasn't that IBM couldn't sell
chips to Apple for servers, the problem was that they couldn't sell
chips for mobile operation, a market in which Intel has been doing very
nicely.

That is/was a very obvious crock... The articles I looked at claimed that
Jobs was pining for a 3+GHz notebook chip, which Intel didn't/doesn't have,
any more than IBM. In fact IBM's current (at the time) notebook chips were
very closely matched to Intel's P-Ms at ~1.8GHz.

And, at that, Jobs didn't (apparently) base his decision on what's on
the table now. Intel has plenty on the table in terms of low power
operation to be credible, and Intel clearly promised much better things
to come, with the focus on performance/watt. That's just what Jobs
wanted to hear.

For low power, Intel has P-M and its future derivataives.... plenty?...
we'll see. Jobs must be loving the IDF VIIV BS.

There's no BS. All you have to do is get a Top 500 list and count, on
your fingers if you have to. My disdain for the Top 500 list is well
known, but I'd have to be persuaded that what it seems to be saying
about who owns the market, both in terms of who is currently dominant
and what seems to be an unmistakeable trend, is irrelevant.

The Top 500 doesn't say much about what is actually moving in the market -
it's easy, in the MP space, to just add CPUs and claim to be pissing
higher. A Google on "Cray XD1 wins" shows sufficient "wins" for Opteron in
actual implementations that it's clear that AMD has more than a little
elbow-room in the current "unmistakeable trend".

 

[Robert Myers]

That is/was a very obvious crock... The articles I looked at claimed that
Jobs was pining for a 3+GHz notebook chip, which Intel didn't/doesn't have,
any more than IBM. In fact IBM's current (at the time) notebook chips were
very closely matched to Intel's P-Ms at ~1.8GHz.

Obvious? Point me to the slam dunk study that led you to that
conclusion, please.

For low power, Intel has P-M and its future derivataives.... plenty?...
we'll see. Jobs must be loving the IDF VIIV BS.

I don't think VIIV has anything particularly to do with low power,
other perhaps than fanless operation for home theaters. Apple will be
putting the same marketing stickers on its computers as Dell? I doubt
it very much.

What is Intel *really* going to produce for low power products? Enough
to get by, I'm sure.

Intel is gearing up for a fight over the consumer electronics space.
Both VIIV and the Apple deal are a part of that, and I'm sure we
haven't been told the real story.

The Top 500 doesn't say much about what is actually moving in the market -
it's easy, in the MP space, to just add CPUs and claim to be pissing
higher. A Google on "Cray XD1 wins" shows sufficient "wins" for Opteron in
actual implementations that it's clear that AMD has more than a little
elbow-room in the current "unmistakeable trend".

Cray may survive, or not. SGI may survive, or not. *Somebody* will be
making high-performance clusters with Intel chips, and IBM will be
building what it calls supercomputers whether it makes money doing it
or not. The future is racks of no-name hardware with Intel Inside
stickers. If there is anything fancy, it will be a switch.

The future wouldn't have to be so unexciting. Japan is apparently
willing to spend money on something more interesting. If somebody put
the money on the table, IBM could build somthing really interesting,
too, but it wouldn't be using x86.

In the meantime, the only safe bet is Intel.

RM

 

[Del Cecchi]

Cray may survive, or not. SGI may survive, or not. *Somebody* will be
making high-performance clusters with Intel chips, and IBM will be
building what it calls supercomputers whether it makes money doing it
or not. The future is racks of no-name hardware with Intel Inside
stickers. If there is anything fancy, it will be a switch.

The future wouldn't have to be so unexciting. Japan is apparently
willing to spend money on something more interesting. If somebody put
the money on the table, IBM could build somthing really interesting,
too, but it wouldn't be using x86.

If somebody puts money on the table, IBM (E&TS that is) will build
whatever the customer wants. Intel, AMD, Power. Ethernet, IB,
Quadrics, Myrinet, custom interconnect.

Blue Gene was done the way it was to get to the performance target they
had. What is your estimate of the size pile of white boxes it takes to
get to a Petaflop? What kind of interconnect is required? How much
power does it take?

In the meantime, the only safe bet is Intel.

What are we betting on? And what is the point spread? Or the odds.?
If the bet is on a stock, or buying a compute server, or building a
Supercomputer the answers might all be different.

 

[YKhan]

That is/was a very obvious crock... The articles I looked at claimed that
Jobs was pining for a 3+GHz notebook chip, which Intel didn't/doesn't have,
any more than IBM. In fact IBM's current (at the time) notebook chips were
very closely matched to Intel's P-Ms at ~1.8GHz.

You mean Motorola/Freescale. The Apple laptops were based on the G4
chips from Freescale, not the IBM G5.

Yousuf Khan

 

[YKhan]

I don't think VIIV has anything particularly to do with low power,
other perhaps than fanless operation for home theaters. Apple will be
putting the same marketing stickers on its computers as Dell? I doubt
it very much.

VIIV is the friendly face of DRM. All of the feel-good press releases
this week pretty much didn't mention anything about DRM at all, and for
good reason too.

They've buried the DRM story behind announcements and prognostications
of smaller PCs able to fit into home theatre cabinets, running cooler,
quieter, etc., etc. None of which isn't already being done now in media
centre PCs. So why the hoopla to get this accepted? Because they want
to sneak DRM in without alerting people.

The future wouldn't have to be so unexciting. Japan is apparently
willing to spend money on something more interesting. If somebody put
the money on the table, IBM could build somthing really interesting,
too, but it wouldn't be using x86.

The reason there is any sort of IBM Opteron servers at all is because
they were forced to build one by one of their Japanese supercomputer
customers who wanted to buy Opterons from IBM. Until then, IBM was
avoiding Opteron.

Yousuf Khan

 

[Robert Myers]

If somebody puts money on the table, IBM (E&TS that is) will build
whatever the customer wants. Intel, AMD, Power. Ethernet, IB,
Quadrics, Myrinet, custom interconnect.

Blue Gene was done the way it was to get to the performance target they
had. What is your estimate of the size pile of white boxes it takes to
get to a Petaflop?

*My* estimate? I don't have one. Intel is advertising a teraflop in
20 boards:

http://www.intel.com/pressroom/archive/releases/20050707corp.htm

So, let's see, that would make, um, click, whir, 20,000 boards for a
petaflop, presumably at 90nm. Make your own guesses for low power
operation and/or smaller feature size.

Interconnect? Whatever can be bought white-box standard, infiniband
on-board, presumably. As I said, the only interesting hardware left is
the switch.

What are we betting on? And what is the point spread? Or the odds.?
If the bet is on a stock, or buying a compute server, or building a
Supercomputer the answers might all be different.

I don't bet on stocks. What is a compute server?

My crystal ball is pretty dark at the moment. Intel is acting the way
IBM used to: 50,000 in India, "trained on Intel hardware," and heaven
only knows how many thousands of hires, mostly software engineers. And
who knows what Intel is really up to with Apple, other than that Intel
and Apple are lined up on one side and Sony, IBM, and Toshiba are lined
up on the other, with Microsoft who knows where. Could get ugly. I'm
betting on Intel to survive and to be the dominant provider of hardware
for compute farms...to avoid that "supercomputer" word, which probably
should have been buried with Seymour.

IBM can play? Of course they can. And they will. Haven't been to
LLNL in a long time, but if I wanted to call IBM onsite, I'll bet I
could find a number, and there'd be a management type there to answer
the phone. IBM *could* do something really interesting, but since the
order book is open first for the national labs, we won't get any such
thing.

RM

 

[Del Cecchi]

*My* estimate? I don't have one. Intel is advertising a teraflop in
20 boards:

http://www.intel.com/pressroom/archive/releases/20050707corp.htm

So, let's see, that would make, um, click, whir, 20,000 boards for a
petaflop, presumably at 90nm. Make your own guesses for low power
operation and/or smaller feature size.

Interconnect? Whatever can be bought white-box standard, infiniband
on-board, presumably. As I said, the only interesting hardware left is
the switch.




I don't bet on stocks. What is a compute server?

My crystal ball is pretty dark at the moment. Intel is acting the way
IBM used to: 50,000 in India, "trained on Intel hardware," and heaven
only knows how many thousands of hires, mostly software engineers. And
who knows what Intel is really up to with Apple, other than that Intel
and Apple are lined up on one side and Sony, IBM, and Toshiba are lined
up on the other, with Microsoft who knows where. Could get ugly. I'm
betting on Intel to survive and to be the dominant provider of hardware
for compute farms...to avoid that "supercomputer" word, which probably
should have been buried with Seymour.

IBM can play? Of course they can. And they will. Haven't been to
LLNL in a long time, but if I wanted to call IBM onsite, I'll bet I
could find a number, and there'd be a management type there to answer
the phone. IBM *could* do something really interesting, but since the
order book is open first for the national labs, we won't get any such
thing.

RM

Microsoft is on both sides. Xbox with IBM, Windows with Intel.

What is this "order book open first for national labs"? IBM has a
rapidly growing technical services organization capable of designing and
building any kind of system, from a rugged laptop to Blue Gene.

The services of this organization are available to pretty much any
customer with money to spend. So if you or someone else has something
really interesting and has money to spend on making it happen, there is
nothing in the way. Maybe if it is really interesting and there is
significant IP and a market for more instances, some sort of joint deal
could be arranged, although ETS doesn't normally do that sort of thing.

As for Blue Gene, several have been sold to places that are not national
laboratories, if that's what you mean.

A compute server is something like a piece of a compute farm.

As for Intel surviving, it seems likely. Although there is a point of
view that the future video game consoles will do everything that folks
in the home now use a PC for. And that could have an adverse effect on
Intel.

 

[Robert Myers]

As for Blue Gene, several have been sold to places that are not national
laboratories, if that's what you mean.

I've been noting the Blue Gene sales. I've talked to people who like
it. I think it's obvious why I don't.

A compute server is something like a piece of a compute farm.

As for Intel surviving, it seems likely. Although there is a point of
view that the future video game consoles will do everything that folks
in the home now use a PC for. And that could have an adverse effect on
Intel.

Seems like somebody has to lose out big time here, but I sure wouldn't
want to be placing any bets.

Not placing a bet on Cell is easy, until there's actual hardware in the
marketplace. There's somebody nosing around IBM's message boards
asking just about the same questions I would ask and being promised
answers real soon now, as far as I can tell. I do appreciate the links
you posted on comp.arch.

RM