Intel brings forward schedule for Clarkdale CPUs to 4Q09

[IntelUser2000]

Originally posted by: 21stHermit

That's right out of the conspiracy theory book. Doesn't fly with me. Rather all those SOC derivatives of Atom are too low volume for Intel and would be in the 10's - 100's of thousands, whereas Intel has shipped 10's of millions of Atoms. I believe all Atom's are the same die and the different SKUs are via validation and binning. IIRC, Atom has the largest profit of any Intel CPU.

Intel isn't sending ENTIRE Atom SoC to TSMC though. Critical parts like the CPU core/graphics/memory controller is still going to be manufactured by Intel. They have their plans of SoC process versions all the way to 22nm.

For example, the Langwell/Pineview I/O controller is being manufactured by TSMC, while the Lincroft CPU core will be SoC version of Intel's 45nm process.

While I may have been incorrect in my original statement, I don't know why everyone jumped at me. The fact is that before Merom, Intel had a reputation for not meeting deadlines. The Tick Tock strategy was a response to this, and yes, up until now, it has worked quite well. All I was saying was that I don't want them to go back to the days before Core 2.

I think the conclusion is maybe the actual release date hasn't been really forwarded or delayed. Clarkdale and 32nm in general seems to being going very smoothly. I don't know why but people do exaggerate news of delays quite a bit.

By Drizek: Basically what they did was eliminate havendale under the guise of having a really mature 32nm process, then one month later they delay the 32nm parts back to their original release dates. I guess Havendale didn't work out for them and they didn't want to admit failure.

Let me make this clear. The ORIGINAL 32nm launch BEFORE Havendale got cancelled was Q2 for Gulftown/Arrandale/Clarkdale(back then the names weren't known yet).

Then Havendale got cancelled and 32nm dual cores went to Q1 2010 schedule.

http://theovalich.wordpress.co...havendale-fusion-cpus/

"Arandale was originally supposed to debut for Back to School season 2010, alongside 32nm quad-core and sexa-core Westmere processors (Core i7 die-shrinks). But now, Arandale core has been brought forward by six months to Q1'2010. The debut is set probably for March (can you say CeBIT?) timeframe."

No, you are just bashing. Maybe you should research a bit more before posting hmm?

 

[IntelUser2000]

Originally posted by: Idontcare

Uhm, really? I'd love to disagree with this forum's growing sentiment regarding your apparant bias against Intel, all you need do here is supply some links that show proof Intel was late with Banias, Dothan and Yonah.

(And late means it was released after Intel said it would be released, not after some d-bag on a blog said it was supposed to be released because he felt like it should have been released earlier than it really was. I await delivery of your links.)

Most of the rumors are generally pretty accurate. Purely for info let me link some information from the Intel presentation I have.

According to Intel, the delays were:

Banias-0.1 months early
Dothan-5.0 months late
Yonah-0.9 months late
Cedar Mill-0.5 months early

Willamette was the longest at 7.8 month delay and followup was Prescott with 6.0 months.

Prescott I think was the most hyped and publicized though. Dothan was also a widely known delay back then. Banias/Yonah/Cedar Mill was so-so.

It's kind of striking the differences the products made according to their delay/forward status. Prescott for example was one of the worst in their history.

 

[taltamir]

well, from some perspective it makes sense.
Development is a steady thing, evolutionary if you will. At some point you have a very good product much earlier than you expected, instead of still improving it more, just release as is as it already exceeds expectations.
On the other hand, if you are delaying it is because you just can't get it right, and eventually you will relent and release something that is just not quite right, but will have to do since you needed a product months ago.

 

[Fox5]

Originally posted by: IntelUser2000

Originally posted by: Idontcare

Uhm, really? I'd love to disagree with this forum's growing sentiment regarding your apparant bias against Intel, all you need do here is supply some links that show proof Intel was late with Banias, Dothan and Yonah.

(And late means it was released after Intel said it would be released, not after some d-bag on a blog said it was supposed to be released because he felt like it should have been released earlier than it really was. I await delivery of your links.)

Most of the rumors are generally pretty accurate. Purely for info let me link some information from the Intel presentation I have.

According to Intel, the delays were:

Banias-0.1 months early
Dothan-5.0 months late
Yonah-0.9 months late
Cedar Mill-0.5 months early

Willamette was the longest at 7.8 month delay and followup was Prescott with 6.0 months.

Prescott I think was the most hyped and publicized though. Dothan was also a widely known delay back then. Banias/Yonah/Cedar Mill was so-so.

It's kind of striking the differences the products made according to their delay/forward status. Prescott for example was one of the worst in their history.

Besides the heat issues, I'd say Willamette was a far more disappointing launch than Prescott. At least Prescott was faster than the previous gen, brought about some new features (some new SSE, dual core, and 64 bit) and somewhat competitive, Willamette was slower than the Pentium 3, and also meant Intel did not have a performance competitive part with the DDR Athlon XP at the time. Lucky for Intel, no one took AMD seriously yet and their chipsets still sucked.

 

[Idontcare]

Originally posted by: Fox5
Besides the heat issues, I'd say Willamette was a far more disappointing launch than Prescott. At least Prescott was faster than the previous gen, brought about some new features (some new SSE, dual core, and 64 bit) and somewhat competitive, Willamette was slower than the Pentium 3, and also meant Intel did not have a performance competitive part with the DDR Athlon XP at the time. Lucky for Intel, no one took AMD seriously yet and their chipsets still sucked.

Prescott was not better than Northwood, infact it had lower IPC and debuted at the same shipping clocks of Northwood at the time (so even the faster shipping Prescott was slower than the fastest shipping Northwood at the time).

Intel's Pentium 4 E: Prescott Arrives with Luggage

Prescott is being launched today at four clock speeds, giving us the following:

Pentium 4 3.40E
Pentium 4 3.20E
Pentium 4 3.00E
Pentium 4 2.80E

But don?t get too excited, the 3.40E chip isn?t actually available yet and to make up for that fact Intel also released a Northwood based 3.40GHz Pentium 4. The Northwood based Pentium 4 3.40GHz is currently available, but within the coming months you?ll see them replaced with Prescott based 3.40Es.

http://www.anandtech.com/showdoc.aspx?i=1956&p=12

Maybe you were thinking of Northwood's launch versus Willamette?

 

[drizek]

First off, IntelUser, thanks for clearing all that up. I made my original statement from memory and I had a hard time finding any of this stuff on google. I also misread the announcement for westemere and assumed that Q4 09 meant "back to school 09". I didn't realize that they had pushed it forward as much as they did.

 

[MODEL3]

Originally posted by: Idontcare
It's pretty clear why Intel chose to leverage TSMC for SOC derivatives of Atom. There always has been, currently is, and most likely will always be a significant gap in the process technology capabilities of Intel versus TSMC (and most others in the high-performance MPU market).

By putting customers of their atom SOC into TSMC's fabs instead of their own fabs Intel discreetly manages to keep their own internal atom-based SOC designs that much farther advanced and at an advantage over any other atom-based SOC design that it might eventually encounter out in the marketplace.

(say Nvidia makes an atom-based SOC for ION2, with the TSMC deal the power/performance of atom-based SOC's coming out of TSMC for NV won't be able to touch Intel's with a AAA battery)

(PS - I know this because this was partially our motivation at TI when we qualified three foundries at a time - TSMC, UMC, and Chartered for example - for the same tech node and production of the same mobile wireless DSP...we kept the competitive stuff inhouse and only pushed to the foundries the stuff we weren't concerned about bleeding over and being used by qualcomm and the rest of our fabless competitors)

There really isn't such a thing as being stretched too thin in the foundry world, everything is prioritized and the top priorities get the resources. The only way atom-based SOC would consume TSMC resources is if a customer signed up and was driving the need for it at TSMC, then it would get the priority and resources commensurate with the expected sales volume, etc, that TSMC would forecast for supporting the product line.

Standard resource management in the foundry world where your business is judiciously resourcing the 100's of projects your potential customers want you to make be your number one priority every day.

Yes that's the right way to see the whole deal.

And probaby 21stHermit was right about the theory stuff (I was just teasing him)

 

[Idontcare]

Originally posted by: drizek
First off, IntelUser, thanks for clearing all that up. I made my original statement from memory and I had a hard time finding any of this stuff on google. I also misread the announcement for westemere and assumed that Q4 09 meant "back to school 09". I didn't realize that they had pushed it forward as much as they did.

Yeah its all cleared up now for me too. I think we all have a better understanding of both Intel's release history as well a better appreciation of where you were coming from when making the posts you did. Its all good.