Intel engineer discusses their dual-core design

[David Schwartz]

Horse-hockey. "Two processors in a package" == dual core, must include
the IBM 3168 of thirty years ago, then. One package (as big as a
box-car), two processors. You Intel apologists are truely amazing.

Why would you waste everyone's time with that type of stupidity. No
definition is sufficiently precise to state, with no ambiguity, precisely
what is and isn't covered by that definition.

Rude? Perhaps. I prefer to think of it as terse. You *are* stupid, if
you buy into Intel's garbage.

I have no idea what you're talking about.

Oh, wow! You can post all sorts of Intel propaganda. I'm truely amazed.

Ahh yes, it must be a conspiracy. It can't be that you're wrong and
everyone else is right. It must be that you're right and everyone else is
stupid.

DS

 

[Felger Carbon]

amazed.

Ahh yes, it must be a conspiracy. It can't be that you're wrong and
everyone else is right. It must be that you're right and everyone else is
stupid.

It was just reported that the earth has a solid core that rotates
inside a second, molten core at a different rate. That means the
earth has a dual core in a single package, and (since our glorious
leader believes in creationism) has had two cores for several thousand
years before either Intel or AMD got around to it. ;-)

 

[jack]

: <snip>

:: Ahh yes, it must be a conspiracy. It can't be that you're
:: wrong and everyone else is right. It must be that you're right and
:: everyone else is stupid.
:
: It was just reported that the earth has a solid core that rotates
: inside a second, molten core at a different rate. That means the
: earth has a dual core in a single package, and (since our glorious
: leader believes in creationism) has had two cores for several
: thousand years before either Intel or AMD got around to it. ;-)

Oh, now that is just tooooo funny! LOL! :-O

j.

 

[Robert Redelmeier]

It was just reported that the earth has a solid core that
rotates inside a second, molten core at a different rate.

The dual core has been known for 30+ years. The differing
rate (degrees per year) is the new discovery inferred from
inner (solid) core non-uniformities.

That means the earth has a dual core in a single package,
and (since our glorious leader believes in creationism)

Hmm ... I think you're stretching the bounds of sarcasm
to breaking. Who you consider your leader and what you consider
glory are personal matters. You ought not to presume "our".

-- Robert

 

[Del Cecchi]

Why would you waste everyone's time with that type of stupidity. No
definition is sufficiently precise to state, with no ambiguity, precisely
what is and isn't covered by that definition.




I have no idea what you're talking about.




Ahh yes, it must be a conspiracy. It can't be that you're wrong and
everyone else is right. It must be that you're right and everyone else is
stupid.

DS

Well, I've been around for a pretty long time and I never heard "dual
core" apply to anything other than a single chip with two processor
cores on it. Dual Processor, sure. But "core" has always been a chip
concept.

del

 

[Rob Stow]

...and what is wrong with Opterons? Had one here for well over a year.
Amazingly the price hasn't dropped.

A year already? Jesus! Seems like just yesterday you were
talking about buying the parts and putting it all together.

 

[keith]

It was just reported that the earth has a solid core that rotates
inside a second, molten core at a different rate. That means the
earth has a dual core in a single package, and (since our glorious
leader believes in creationism) has had two cores for several thousand
years before either Intel or AMD got around to it. ;-)

Perhaps, but even the almighty Felger Carbon can't detect a .3-.5 degree
rise per year in a processor. Maybe you ought to go compare that with
your global-warming chickens. ;-)

 

[keith]

A year already? Jesus! Seems like just yesterday you were
talking about buying the parts and putting it all together.

Yep, Time flies when you're having fun! ;-) It was all together in June
of last year. I still haven't gotten the SATA controller working under
Linux though, so have a 160GB drive sitting there warming the room.

 

[Pankaj]

AMD's approach recently has been to target collaboration and sponsorship
with high tech activities, mostly in some form of sports. I wonder how
this is panning out: e.g. has the Lance Armstrong/Discovery Team connection
been worth anything to them? How about the Sauber & Ferrari F1 connection?
I don't watch NASCAR but that might be a more worthwhile avenue for their
collaboration/sponsorship efforts in the U.S. market??

For AMD, its a step in the right direction, but I dont think they
should spend too much of their energy (and money) at this stage to make
it a brand name. (As Tony has mentioned above, to an average consumer,
it doesnt matter whats inside the computer. They buy Dell or HP or ...)
They should concentrate on getting market share while making money. I
think Hector Ruiz is doing a good job at it. He may not have gained any
market share as yet, but he sure has realized that he cannot gain it by
reducing the price 25% below Intels's, as Jerry tried to do. Intel
wants them to be around 15% market share irresepctive of the price.
Intel desperately wants AMD to be around, as they dont want to be
labelled a monopoly. Hector has realized these facts and if you have
noticed, the average price of the lower end chips (money losing) has
gone up over the last few years. Intel has been very shrewd in their
pricing strategy. They want AMD at 15%, but dont want them to make any
money, which can be used to stage a comeback anytime. Every dollar of
the cpu price is set to make sure that in the equation , AMD's
profitability is just around 0. And what cannot be done with price, it
is done with strong-arming OEMs. (I really dont want to start another
argument here

Regards,
Pankaj

 

[Nate Edel]

Here's a more interesting question: Intel built the D/C chips on P4
rather than P-M, presumably so they could offer the ht model at a huge
premium.

Or because they had all the existing cache/bus arbitration logic designed
for the Netburst Xeon/Nocona which was not far off in design from the
P4/Prescott, whereas no such logic had yet been designed for the P-M.

Given the low power and far better performance of the P-M in terms of
work/watt and work/clock, why not a dual core Pentium-M? Then when the
better P4 D/C chip is ready they could offer that?

Just curious as to the logic for the decision if anyone has any insight.

Because it was much less design work and validation to do it with a
P4/Netburst core, I'd imagine.

 

[Bill Davidsen]

My question then is:
(1) The readers of this group are probably more tech savvy than average,
so would you say that you have followed my downward opinion of Sony over
the past 10 yrs?

Having had the chance to try a fair number of digital cameras over tha
last few years, in the $700-1200 range, including the top rated
Konica-Minolta, I will stick with Sony. My daughter does professional
photography as part of her job, and gets to use equipment bought with a
government budget (30 inch monitor type budget). She bought a Sony with
her own money, too.

Did a lot of shopping and wound up with a Sony camcorder as well, after
deciding on something else from specs and then trying both.

(2) Now, going into the wider population, do you think Sony has fallen
from a premium brand to just one among many?

That's harder to say, I get the impression that most people still
percieve them as one of the better brands.

 

[Bill Davidsen]

Yep, Time flies when you're having fun! ;-) It was all together in June

of last year. I still haven't gotten the SATA controller working under
Linux though, so have a 160GB drive sitting there warming the room.

I assume you're running a recent kernel, but 2.6.13 probes the PCI bus
itself instead of trusting the BIOS to do it (may be an option, it's
just out yesterday). Whatever came with FC4 found the controller in my
ASUS board, but I haven't put a drive on it, having a stock of fast PATA
left.

To address the original point YK raised, Intel still has the clear lead
in low power for mobile. And they spend more on research, so the
technical lead may be due to choosing a better direction than having
better people. Intel is likely to play a top game of catchup.

As for Intel offering a big discount to vendors who sell only Intel
CPUs, how is that different from Microsoft doing the same thing. I can't
buy a brand name laptop with Linux, and worse I have to pay for Windows
I never use. Nice that there are a few vendors who don't follow that
path, but they do cater to a niche market.

My next system will almost definitely be Intel dualie, but I have used a
fair number of AMD CPUs in the last few years, so it's not a brand
loyalty decision, just where I think I should go first.

 

[Bill Davidsen]

"Thermal Design Power" (TDP) is defined as the maximum amount of heat a
cooler has to take away under defined conditions (given die temperature and
air temperature). I.e. you define a temperature difference delta T, and a
power P, and the cooler has to make sure that for a given P the delta T is
met.

AFAIK, AMD didn't change the TDP definition when going to 90nm, but they
lowered the maximum temperature rating for their dies. So in reality, you
have a lower delta T to reach, but with the given TDP (for cooler
designers), they simply overstate the TDP, with the same effect. AMD also
doesn't have thermal throttling, so the TDP really must fit into a
worst-case szenario, while Intel allows programs like CPUburn to exceed the
power - the idea is that in typical use, power bursts like that are short,
and you can prevent damage by throttling.

If you really need the CPU power to run long simulations/encodings/game
sessions/whatever, and you buy Intel, then you also need a cooler that can
handle more than Intel asks.

That's exceptionally well stated. It also is a clear explanation of
throttling benefits and performance issues, which could help people make
a brand decision.

 

[Bill Davidsen]

Spin? Like to load them words, eh? One reason to do SMP testing is that
it's "easy" to test cache coherence with two processors. Two processors
can bang the caches pretty hard against each other.

Not really, "spin" usually refers to how facts look from another
viewpoint, which is just what I was after. But if the chip has no SMP
logic, such testing would not be possible.

Do you really think Intel has no SMP verification capabilities? Do they
not "borrow" tools from one organization to another? There's more here
than meets the eye. Someone dropped the ball, big time!

I really think the desktop P4 has no SMP capabilities, because they
would take real estate and need to be tested. Not features you want in
production.

Absolutely. ...which makes the SMP verification issue even more
strange.




Search me. I've given up trying to explain Intel's moves; too bizare.

In rethinking my question, I can see from a marketing point both HT and
64 bit are needed.

 

[George Macdonald]

I assume you're running a recent kernel, but 2.6.13 probes the PCI bus
itself instead of trusting the BIOS to do it (may be an option, it's
just out yesterday). Whatever came with FC4 found the controller in my
ASUS board, but I haven't put a drive on it, having a stock of fast PATA
left.

To address the original point YK raised, Intel still has the clear lead
in low power for mobile. And they spend more on research, so the
technical lead may be due to choosing a better direction than having
better people. Intel is likely to play a top game of catchup.

I think that point has been made ad nauseum here: Intel has allowed
marketroids to dictate technical directions & policy. Recent events would
indicate that Intel is determined to continue with "spin" as their
strategy.

As for Intel offering a big discount to vendors who sell only Intel
CPUs, how is that different from Microsoft doing the same thing.

Have you not read the AMD complaint? It's much more than volume discounts:
demands to OEMs for exclusivity with threats of non-delivery of product
already contracted; same thing for large distributors; stealing AMD systems
from a promotional event; bribing retailers, with whom they have no direct
business relationship, to exclude any AMD systems from shelf space.
There's a long list - sounds more like the Mob than an honest business.

 

[George Macdonald]

That's exceptionally well stated. It also is a clear explanation of
throttling benefits and performance issues, which could help people make
a brand decision.

A little empirical observation will, however, reveal that AMD does not need
thermal throttling - an even better criterion for "brand decision".

 

[Tony Hill]

A little empirical observation will, however, reveal that AMD does not need
thermal throttling - an even better criterion for "brand decision".

Under anything reassembling a typical situation, Intel processors
don't need thermal throttling either, even if you are running some
application that has CPU use pegged at 100% for long periods of time.
Intel's TDP numbers are rather pessimistic and there are VERY few
situations where actual power consumption ever exceeds that number,
even for just a short peak.

What's more, it's a sort of worst-case from the process standpoint.
As you probably know, processors of a giving speed grade and stepping
can and do have different power consumption figures for the same load.
Minor differences from the process side of things dictate this. And
finally, any normal cooler sold with Intel systems doesn't JUST meet
the specs with no room to spare, there is always at least a small
margin of error.

Long story short, if your processor is throttling, regardless of what
you're running on the PC, there's probably something wrong with your
setup. Either that or ambient temp is well above normal.

 

[keith]

I assume you're running a recent kernel, > but 2.6.13 probes the PCI bus

2.65-7.155.29, according to "uname -a"

itself instead of trusting the BIOS to do it (may be an option, it's
just out yesterday). Whatever came with FC4 found the controller in my
ASUS board, but I haven't put a drive on it, having a stock of fast PATA
left.

I was quite dissapointed by SIL's lack of *working* SATA drivers, not to
mention Tyan's disinterest. I gave up and bought another 160GB PATA
drive.

To address the original point YK raised, Intel still has the clear lead
in low power for mobile. And they spend more on research, so the
technical lead may be due to choosing a better direction than having
better people. Intel is likely to play a top game of catchup.

They're focusing on mobile, there's no doubt there. At the same time
Intel is leaving the high performance market to AMD. I could kinda
figure that strategy, until it was quite clear taht Itanic hit the berg.
They've forfeited that market. ...to AMD from below, and IBM from above.

As for Intel offering a big discount to vendors who sell only Intel
CPUs, how is that different from Microsoft doing the same thing. I can't
buy a brand name laptop with Linux, and worse I have to pay for Windows
I never use. Nice that there are a few vendors who don't follow that
path, but they do cater to a niche market.

1) It's not a discount. It's an exclusive deal. "Either you sell mine
*exclusively* or you get no deal." "Even if you sell only mine, you take
second banana to Mikey."

2) It's no different, in fact. That doesn't excuse Intel from strong-arm
tactics. BTW, the major reason I only went one generation (Win2K) with M$.

My next system will almost definitely be Intel dualie, but I have used a
fair number of AMD CPUs in the last few years, so it's not a brand
loyalty decision, just where I think I should go first.

This makes no sense at all.

 

[keith]

Not really, "spin" usually refers to how facts look from another
viewpoint, which is just what I was after. But if the chip has no SMP
logic, such testing would not be possible.

The logic *is* there, though perhaps not brought out to the user.
Regardless, Intel *has* SMP verification capability. The only
excuse for such a stupid statement is that Intel has NIH burried so far
up their ass that even the exec's can't wiff the nonsense. They *have*
the test-cases.

I really think the desktop P4 has no SMP capabilities, because they
would take real estate and need to be tested. Not features you want in
production.

Irrelevant. They have SMP test and verification capability. If they
needed it for dual-core, they didn't have to invent new material. To
suggest such is simply silly.

In rethinking my question, I can see from a marketing point both HT and
64 bit are needed.

HT is a bust. Everyone knew 64bits was needed five years ago. Intel
tried to derail x86 and go with Itanic only. AMD decided otherwise.
Where is Itanic?

 

[Ketil Malde]

To address the original point YK raised, Intel still has the clear lead
in low power for mobile. [...]

They're focusing on mobile, there's no doubt there. At the same time
Intel is leaving the high performance market to AMD.

I think one important change is that the laptop market (like high
performance) used to be a high margin area. Now that laptop prices
have really plummeted, and Intel's laptop lead isn't as exclusive as
its Xeon business used to be (sure, Centrino is nice, but you can
always put together something similar based on AMD), it seems highly
unlikely that it will give anything near the same margins.

-k