Intel brings forward schedule for Clarkdale CPUs to 4Q09

[Sylvanas]

Digitimes

Intel has brought forward the mass-production schedule of its new 32nm processor codenamed Clarkdale to the fourth quarter of 2009 from the originally planned first quarter of 2010, according to sources from motherboard makers.

The company's 32nm Clarkdale CPUs will account for 10% of Intel's total OEM desktop CPU shipments in the fourth quarter, while 45nm Core i7 processors will account for 1%, Lynnfield-based processor 2%, Core 2 Quad processors 9%, Core 2 Duo E8000/E7000 series processors 35%, Pentium E5000/E6000 series 31%, Celeron E3000 and Atom series together 8%, and 65nm-based Pentium E2000 and Celeron 400 together 4%.

In the first quarter of 2010, the proportion of Clarkdale-based CPUs will rise to 20%, while 45nm-based processors will account for 78%, and 65nm-based Celeron 400 series will drop to only 2%, the sources noted.

Intel declined the opportunity to respond to this report saying it cannot comment on unannounced products.

 

[Idontcare]

So digitimes just finally caught up to publishing the information that Intel made public knowledge back in Feb?

http://www.anandtech.com/cpuch...howdoc.aspx?i=3513&p=5

They used to be so much better than this.

 

[Acanthus]

Give us 32nm Atoms with a north bridge that doesn't consume 15x the load power of the CPU please.

 

[IntelUser2000]

Originally posted by: Idontcare
So digitimes just finally caught up to publishing the information that Intel made public knowledge back in Feb?

http://www.anandtech.com/cpuch...howdoc.aspx?i=3513&p=5

They used to be so much better than this.

The info I understood about Clarkdale was always shipment in late Q4 along with official launch early Q1(probably during CES January). The recent news seems like a real CPU launch at Q4, which is significant. According to similar sources, Intel is expecting 10% of shipments for their CPUs as Clarkdale in Q4.

On a different info if anyone is interested. I tried to extrapolate die sizes of Clarkdale CPU core and the MCM GPU/PCI-Express/MC and I think I got it. On Anandtech's Lynnfield preview, they had comparison shots of Lynnfield and Core 2 CPUs. The packaging size of LGA1156 CPU and LGA775 CPUs are the exact same 37.5mm x 37.5mm.

I have previously estimated the Northbridge MCM and CPU core die size as 160mm and 100mm2 respectively. It had to be bit off as it was compared to Penryn's die then estimated its relative size. Now I revised its estimates.

Northbridge MCM=135mm2
CPU Core=85mm2

It's still hard to estimate the shrink of Northbridge portion as I/O portions have a different scaling ratio than CPU cores, but most of the time, things like memory controller and PCI-Express controller seems to have worse scaling.

I see the die size increases on Northbridge as an important factor whether Intel allowed more transistor budget on their IGPs which are directly related to performance and features.

Give us 32nm Atoms with a north bridge that doesn't consume 15x the load power of the CPU please.

N270=2.5W
945GSE=5W

2x...

Nettops(Desktop Atoms) don't count.

 

[21stHermit]

I would ask if any of this is really possible. I'd estimate Intel ships 60M CPU's per year. So in order to have Clarkdale be 10% of the mix in Q409, that would be 6M/4 or 1.5M Clarkdale's in Q4. Is that even possible?

New CPU, new MB's, new socket (LGA1156), new products by OEM partners. Seems a very tall order since we're in Q3 now.

In any case, I want a mini-ITX w/ a S1156 and a Clarkdale to replace my Atom 330. Go Go Intel!!!

 

[KingstonU]

This will put Intel how many months ahead of AMD/IBM's 32nm? 3, 6, 9?

 

[ilkhan]

9-12 probably.

 

[VirtualLarry]

I'm waiting for laptops with this chip in it. They should be powerhouses.

 

[Cookie Monster]

I wonder if the iGPU on Clarkdales are derivatives of larrabee or just a beefed up version of what they have out now today.

 

[taltamir]

Originally posted by: Acanthus
Give us 32nm Atoms with a north bridge that doesn't consume 15x the load power of the CPU please.

you mean like this:
http://www.anandtech.com/cpuch...ts/showdoc.aspx?i=3565

 

[aigomorla]

i'll believe it when i start seeing 32nm samples.

So far i have yet heard JACK about 32nm... and those of you guys who know me, know i tend to get stuff 2-3 months earily.

So im still in doubt intel will roll out 32nm this year. Next year yes, but Q4 to the public?
Excuse me while i go laugh in a corner somewhere.

The only people getting 32nm Q4 will most likely be people like me.

 

[Atechie]

Originally posted by: aigomorla
i'll believe it when i start seeing 32nm samples.

So far i have yet heard JACK about 32nm... and those of you guys who know me, know i tend to get stuff 2-3 months earily.

So im still in doubt intel will roll out 32nm this year. Next year yes, but Q4 to the public?
Excuse me while i go laugh in a corner somewhere.

The only people getting 32nm Q4 will most likely be people like me.

You havn't been listing very well then, Intel's 32nm is ahead of schedule...

 

[taltamir]

usually if there are delays, then a company notifies of setbacks in the last minute... it does notify people that it is ahead of schedule, and is moving the release date to an earlier point in time.
I see no reason for intel to lie about that.

 

[Denithor]

Funny of the day.

I read the title as "Intel brings forward schedule for Clarkdale CPUs to 4009" instead of 4Q09.

 

[aigomorla]

Originally posted by: Atechie

Originally posted by: aigomorla
i'll believe it when i start seeing 32nm samples.

So far i have yet heard JACK about 32nm... and those of you guys who know me, know i tend to get stuff 2-3 months earily.

So im still in doubt intel will roll out 32nm this year. Next year yes, but Q4 to the public?
Excuse me while i go laugh in a corner somewhere.

The only people getting 32nm Q4 will most likely be people like me.

You havn't been listing very well then, Intel's 32nm is ahead of schedule...

and u dont know how fast i get my samples do you?

a list to me doesnt mean jack because its a list. A sample on the other hand means it will come out in production unless my sponsor has me a on a gag order on NDA.

Even then i still tend to leak out some stuff to my close forum friends, and you can ask IDC if he's heard anything from me on 32nm.

Infact the only thing IDC hears from me is GOD DAYAM INTEL WITH STUPID DELAYS.

So until i see the physical samples... no.. intel is bsing though there teeth, and they dont need to rush because PHII only has the same name and no performance behind it for intel to bring out 32nm.

There is no point in a faster car when your car is the fastest. Why would you want to cripple your fastest car already with another faster car?

This has been intel's method, otherwise we would of had i5's how many months ago? Oh yeah they were suposed to be rolled out on MARCH. ermm.. its july already and where the FAQ is my i5?

 

[taltamir]

those are some very good points aigomorla... if something was coming in the next few months there would be engineering samples already...

Anandtech says they got their hands on an i5 engineering sample a little while back. and tested it.
I haven't seen any 32nm engineering samples yet...

Than again. Q4 could be the very last day of the year.

 

[dmens]

and they dont need to rush because PHII only has the same name and no performance behind it for intel to bring out 32nm.

my sleep schedule begs to differ. 32nm is cheaper.

 

[taltamir]

yes... there is no reason for intel to give us CLOCKSPEED increases with 32nm (which they COULD, but they are not PLANNING ON; heck they could do so with current 45nm tech). But intel itself benefits from 32bit in the form of cheaper to produce (while costing the same or more to sell)

 

[Idontcare]

Originally posted by: aigomorla
Even then i still tend to leak out some stuff to my close forum friends, and you can ask IDC if he's heard anything from me on 32nm.

Infact the only thing IDC hears from me is GOD DAYAM INTEL WITH STUPID DELAYS.

So until i see the physical samples... no.. intel is bsing though there teeth, and they dont need to rush because PHII only has the same name and no performance behind it for intel to bring out 32nm.

Yeah I haven't heard anything about 32nm other than the press release marketing stuff.

Still though from my perspective its not unlike how Kentsfield, Yorkfield, and Nehalem were released in mid Nov.

Engineering samples weren't really out until Aug, and even then the mobo support (BIOS, microcode, etc) totally sucked so the benches you got off it (you know this, just saying for others) were not representative of the ultimate platform's capability anyways.

The only thing I will be surprised by this fall is if 32nm does not debut in mid-Nov.

 

[21stHermit]

This is not about engineering samples, rather about quantities, very large quantities shipped in Q4 2009. If you reread the article, it claims Clarkdale will make up 10% of Intel's Q409 shipments and 20% of Q1-10 shipments.

I would ask if any of this is really possible. I'd estimate Intel ships 60M CPU's per year. So in order to have Clarkdale be 10% of the mix in Q409, that would be 6M/4 or 1.5M Clarkdale's in Q4. Is that even possible?

New CPU, new MB's, new socket (LGA1156), new products by OEM partners. Seems a very tall order since we're in Q3 now.

 

[Viditor]

Originally posted by: 21stHermit
I would ask if any of this is really possible. I'd estimate Intel ships 60M CPU's per year. So in order to have Clarkdale be 10% of the mix in Q409, that would be 6M/4 or 1.5M Clarkdale's in Q4. Is that even possible?

New CPU, new MB's, new socket (LGA1156), new products by OEM partners. Seems a very tall order since we're in Q3 now.

In any case, I want a mini-ITX w/ a S1156 and a Clarkdale to replace my Atom 330. Go Go Intel!!!

I would say that your estimate is about a third of what they actually ship...

http://www.etforecasts.com/products/ES_pcww1203.htm

In 2005, total x86 sales were ~190 Million, and for 2010 they are estimated at just under 270 million units...
With Intel owning about %80 of that, they are shipping between 152 and 216 million units/year.

 

[21stHermit]

Originally posted by: Viditor

Originally posted by: 21stHermit
I would ask if any of this is really possible. I'd estimate Intel ships 60M CPU's per year. So in order to have Clarkdale be 10% of the mix in Q409, that would be 6M/4 or 1.5M Clarkdale's in Q4. Is that even possible?

New CPU, new MB's, new socket (LGA1156), new products by OEM partners. Seems a very tall order since we're in Q3 now.

I would say that your estimate is about a third of what they actually ship...

http://www.etforecasts.com/products/ES_pcww1203.htm

In 2005, total x86 sales were ~190 Million, and for 2010 they are estimated at just under 270 million units...
With Intel owning about %80 of that, they are shipping between 152 and 216 million units/year.

That's really beside the point, what really matters is: Is it possible for Intel to ship 10% of their total volume in Clarkdale's in 3Q09?

 

[aigomorla]

Originally posted by: dmens

and they dont need to rush because PHII only has the same name and no performance behind it for intel to bring out 32nm.

my sleep schedule begs to differ. 32nm is cheaper.

im sorry but the only 32nm sample i saw was the one dr.who, aka fransicos was flashing around on a post card at the giggybyte competition.
http://i219.photobucket.com/al...s_/Taipei/DSC04117.jpg

Note its a freaken POST CARD.

unless u want to tell us something we dont know dmens.

But chances are i'll get my samples near the same time you will, and i havent anything about them.

Otherwise id be bitting my tongue every post in this statement.

 

[Atechie]

Clicky...

*chough*

 

[aigomorla]

and what is that to prove?

i dont see a chip, i dont see a nice shiney ihs which says intel confidental, i dont see a board outside, nor do i see any benchmarks.

Sorry your link is pointless and invalid.

Until i see something that looks remotely like this:

http://i125.photobucket.com/al...a/Personal/Mychips.jpg

No, im gonna walk around with a needle poping all your dreams on 32nm, until i see a sample like 1 of the 5 above.