IBM vs. Intel 90nm power consumption

[Tony Hill]

Actually all I was really trying to point out is that the problems
Intel is having with the Prescott are NOT strictly process-related
problems as seemed to be the common consensus. For a chip with 125M
transistors running at 3+ GHz, it's power consumption [isn't?] really out of
whack with expectations.

Can it be? I've discovered a mistake in one of Tony Hill's posts?

Just one?! Geez, I make spealing an grammar mistake in nearly every
post I make! :> As Felger correctly pointed out, I once again wrote
"it's" instead of "its", a mistake that I make on a pretty regular
basis and would fix if I bothered reading over the message at all!

You are quite correct, I did mean to write "its power consumption
isn't really out of whack with expectations."

 

[Tony Hill]

Actually all I was really trying to point out is that the problems
Intel is having with the Prescott are NOT strictly process-related
problems as seemed to be the common consensus. For a chip with 125M
transistors running at 3+ GHz, it's power consumption [isn't?] really out of
whack with expectations.

Can it be? I've discovered a mistake in one of Tony Hill's posts?

And Prescott heat is not out of whack with /whose/ expectations ?

The expectations of anyone who's looked to design a chip with 125M
transistors running at 3+ GHz? Of course, since Intel is the only
company doing such a thing, I guess that's rather limiting.

It certainly is out of whack from the viewpoint of consumers -
Prescott can't compete with an Opteron/Athlon64/AthlonFX but uses

Whether or not the P4 "Prescott" can compete with an Athlon64 depends
heavily on what applications you are running. There are many areas
where the a 3.2GHz Prescott will beat out an Opteron/Athlon64/Athlon64
FX and many areas where opposite is true. Fortunately us consumers
have a choice so we can decide which chip to get based on what
applications we use.

almost twice as much juice and produces almost twice as much heat.

When the Athlon64/Opteron reach 3.2GHz, they will likely be consuming
a similar amount of power, even when shrunk to 90nm. The trick is
that at a 3.2GHz K8 chip SHOULD perform a lot better than a 3.2GHz P4
chip at almost all applications. Given that the two chips have a
similar number of transistors and the "Prescott" isn't cranking up to
high clock speeds like expected, this should suggest something about
the design of the chip.

Sure makes it easy for everyone to take no more than two seconds
to evaluate and discard the Prescott option - especially now that
air conditioning season is about to begin in the northern hemisphere.

I don't know about your little corner of the northern hemisphere, but
over here we had a snow storm pretty much all day today. It'll be a
little while yet before we start cranking up the air conditioner. The
air conditioning argument makes good sense for the southerners, but
for us Canucks (you live in Sask. if memory serves?), higher power
consumption reduces our heating/cooling cost for the year as a whole.
Where we usually only have air conditioning on for June -> August or
thereabouts, we've got the heat on from Oct. -> April. Plus, the
hottest day in the summer is only about 15C hotter than desired
ambient temps indoors, the coldest day in the winter can easily be
40-60C lower than indoor ambient.

 

[Rob Stow]

On Sun, 04 Apr 2004 10:40:20 -0400, Tony Hill

Actually all I was really trying to point out is that the problems
Intel is having with the Prescott are NOT strictly process-related
problems as seemed to be the common consensus. For a chip with 125M
transistors running at 3+ GHz, it's power consumption [isn't?] really out of
whack with expectations.

Can it be? I've discovered a mistake in one of Tony Hill's posts?

And Prescott heat is not out of whack with /whose/ expectations ?

The expectations of anyone who's looked to design a chip with 125M
transistors running at 3+ GHz? Of course, since Intel is the only
company doing such a thing, I guess that's rather limiting.

It certainly is out of whack from the viewpoint of consumers -
Prescott can't compete with an Opteron/Athlon64/AthlonFX but uses

Whether or not the P4 "Prescott" can compete with an Athlon64 depends
heavily on what applications you are running. There are many areas
where the a 3.2GHz Prescott will beat out an Opteron/Athlon64/Athlon64
FX and many areas where opposite is true. Fortunately us consumers
have a choice so we can decide which chip to get based on what
applications we use.

almost twice as much juice and produces almost twice as much heat.

When the Athlon64/Opteron reach 3.2GHz, they will likely be consuming
a similar amount of power, even when shrunk to 90nm.

And when AMD64 chips hit 3.2 GHz the Intel chips they are competing
against will be running at 4.x or 5 GHz. There is no point comparing
Intel's current chips against what AMD might be making next year:
today's products compete against today's products.

The trick is
that at a 3.2GHz K8 chip SHOULD perform a lot better than a 3.2GHz P4
chip at almost all applications. Given that the two chips have a
similar number of transistors and the "Prescott" isn't cranking up to
high clock speeds like expected, this should suggest something about
the design of the chip.




I don't know about your little corner of the northern hemisphere, but
over here we had a snow storm pretty much all day today. It'll be a
little while yet before we start cranking up the air conditioner. The
air conditioning argument makes good sense for the southerners, but
for us Canucks (you live in Sask. if memory serves?),

Yes, I live in SK. Irrelevant, however. When someone in California
or Texas cranks up the air conditioning, those are global resources
he is consuming. The air *Canadians* breathe is affected - coal gets
burned in SK to feed juice to the US. *Everyone* should worry about
the long term costs of 70 million Texans and Californians having to
use a few more GW of power because Intel dropped the ball.

higher power
consumption reduces our heating/cooling cost for the year as a whole.

Natural gas heating costs about 2/3 as much as electrical
heating in Saskatchewan. An example I am quite familiar with:
keeping a detached double garage at 0'C all winter long will
average $80/month of power, but only $50/month for gas if you
convert to natural gas heating.

(I always get a kick out of listening to people whining about
having to heat their garages in the winter because they can't
let it get too cold for the fridge and/or deep freeze they have
in the garage.)

 

[Tony Hill]

And when AMD64 chips hit 3.2 GHz the Intel chips they are competing
against will be running at 4.x or 5 GHz. There is no point comparing
Intel's current chips against what AMD might be making next year:
today's products compete against today's products.

The point I was getting at though was that in the fairly near future
AMD (and IBM, and everyone else) is going to face a lot of the same
high power consumption issues that Intel is running into now.

Yes, I live in SK. Irrelevant, however. When someone in California
or Texas cranks up the air conditioning, those are global resources
he is consuming. The air *Canadians* breathe is affected - coal gets
burned in SK to feed juice to the US. *Everyone* should worry about
the long term costs of 70 million Texans and Californians having to
use a few more GW of power because Intel dropped the ball.

I'm *MUCH* more concerned about all the big pickup trucks and SUVs
that Texans and Californians (and all other North Americans) are
buying rather than what type of processor they are using! As a rough
estimate, running a Prescott full-out (100W) capacity for an entire
day would use roughly the same amount of fossil fuels as driving a
large SUV (15L/100km) for 5km. A lot of power for sure, but virtually
no one runs their chips full out 24/7, while LOTS of people drive an
average of at least 5km/day.

(I always get a kick out of listening to people whining about
having to heat their garages in the winter because they can't
let it get too cold for the fridge and/or deep freeze they have
in the garage.)

Haha! That's one of those situations where I wouldn't be sure if I
should smile and nod, laugh in the persons face or simply smack them
around! :>

 

[George Macdonald]

First P2 (deuthchech or something like that), ppga Celerons
(mendocinos), then fcpgas (coppermine) and tadaaa tualatins share the
same bus and work on the very same motherboards.

I think you mean DesChutes for the P2. Even the Coppermine, which had the
same pin-outs as the PII/PIII(Katmai) had some issues on some iBX mbrds -
Intel had some specific recommendations to do with split-plane mbrd power.
The Tualatin had a different arrangement of pins and needed at least a
Powerleap or other brand adapter to make the Celeron model work in a
PII/PIII mbrd. Even then there were BIOS issues since you were running a
CPU which the BIOS did not support.

Thats as close as one
can get, so in my opinion its the same core. Dohan is just a tweaked up
tualatin (I bet one could make them run on old iBX).

You'd need a pretty fancy adapter to match a Banias or Dothan FSB to the
old AGTL+ FSB... and it would be a total waste of time, which would likely
cost as much as a newer mbrd.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[RusH]

Even the Coppermine, which
had the same pin-outs as the PII/PIII(Katmai) had some issues on
some iBX mbrds - Intel had some specific recommendations to do
with split-plane mbrd power.

The pinout was different !

The Tualatin had a different
arrangement of pins and needed at least a Powerleap or other brand
adapter to make the Celeron model work in a PII/PIII mbrd.

Again the pinout was different, in both cases intel 'smart' ppl
exchanged gnd with _reset pins.

Even
then there were BIOS issues since you were running a CPU which the
BIOS did not support.

BiosPatcher http://rom.by/

You'd need a pretty fancy adapter to match a Banias or Dothan FSB
to the old AGTL+ FSB... and it would be a total waste of time,
which would likely cost as much as a newer mbrd.

AFAIK there are no newer mbrd's for those chips At least not
non_mobil.


Pozdrawiam.

 

[George Macdonald]

The pinout was different !

Oh it's getting hazy now and not worth arguing but on the desktop Katmai
was principally a Slot 1 and Coppermine was both Slot 1 and socket. The
Slot 1s were certainly compatible with *some* older mbrds... with the
proviso already mentioned.

Again the pinout was different, in both cases intel 'smart' ppl
exchanged gnd with _reset pins.

"Arrangement of pins"?

BiosPatcher http://rom.by/


AFAIK there are no newer mbrd's for those chips At least not
non_mobil.

I dunno what's being used in blades but apparently Intel will sell you the
P-M for non-mobile use if you cozy up to them and make promises. I've no
idea if it's a different package from the mobile version or not - maybe
someone else can comment.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Grumble]

"Banias" and "Dothan" share a lot of the same technology as the P6,
but they definitely aren't the same. There were some rather
important change made to the core, it most definitely is NOT just a
"PIII on a P4 bus" as many people have suggested.

I'll point out Hannibal's excellent overview.
http://arstechnica.com/cpu/004/pentium-m/pentium-m-1.html

 

[Grumble]

BiosPatcher http://rom.by/

Because everyone here can read Russian...

 

[RusH]

Because everyone here can read Russian...

or is so blind that missed "English" button


Pozdrawiam.