Intel 32 nm Clarkdale Chip Brought Forward to Q4 2009

[beyond_amusia]

Thats small 32nm! I wonder where the limit of silicon is

Edit: So do anyone know where the limit it?

The limit? should be around 11 nm - theoretically, you can only get down to the size of an atom - after that, no room for improvement via die shrinks, so chips will start expanding in size... I assume by then, quantum computing might be viable, or DNA computing... it's too early to tell where it will go.

 

[a_ump]

If what I read was correct AMD may skip 32nm and go straight to 22nm if GLOBALFOUNDRIES has their way. This is more rumor than it is fact but I think it might be true... 45nm for AMD clocks real well and has decent efficiency so... why change it?

because if 32nm isn't goofed by AMD, wouldn't it mean even better clocks, lower power consumption, less die space(lower cost) compared to 45nm.

 

[Flyordie]

because if 32nm isn't goofed by AMD, wouldn't it mean even better clocks, lower power consumption, less die space(lower cost) compared to 45nm.

The idea is why waste money on 32nm when 22nm SOI is already viable?

 

[farlex85]

Where the hell are these fabled "i5" processors. They've been talking about them forever now and still no word on release. I'm starting to think Intel is just using them as a backup plan in case PII's catch up further, and they have no intention of releasing them unless absolutely necessary.

 

[[I.R.A]_FBi]

Where the hell are these fabled "i5" processors. They've been talking about them forever now and still no word on release. I'm starting to think Intel is just using them as a backup plan in case PII's catch up further, and they have no intention of releasing them unless absolutely necessary.

Why so concerned?

 

[farlex85]

Why so concerned?

It just seems odd to me they haven't released them yet. Not really odd I understand, but i7 was released in what November 08 I think. And now here we are 8 months later and the 3 original models released at launch are still the only 3 next gen models you can get. They keep talking about all these other procs (i7 replacements, i5) yet they haven't seemed to move any closer to release. Like I said, I think they're just lying in wait until AMD threatens them enough.

Anyhow I wouldn't say I'm concerned, I just think it's funny.

 

[FordGT90Concept]

Also, Intel has yet to be found guilty in the US because the cases are not over yet... Intel's final trial is set for August 09... the judge is very pissed off at Intel for screwing with him over and over again... so I doubt this will end well for Intel.

Lawsuits aren't about feelings, they are about possible violations of established laws. If the judge makes it about feelings, the lawsuit could easily move to a higher court.

The idea is why waste money on 32nm when 22nm SOI is already viable?

AMD wants to beat Intel to a process. Intel has historically been one process ahead of AMD so the only way they could beat Intel is if they skip a process. That could actually be the boost AMD needs...

Where the hell are these fabled "i5" processors. They've been talking about them forever now and still no word on release. I'm starting to think Intel is just using them as a backup plan in case PII's catch up further, and they have no intention of releasing them unless absolutely necessary.

They were scheduled for Q1 2010 but some of them have been moved up to Q4 2009. I think the reason why it is taking so long to make is because they are based on Netburst architecture which benefits greatly from a lot of tinkering; moreover, they are moving the northbridge to the chip which adds additional development time.

 

[Wile E]

Lawsuits aren't about feelings, they are about possible violations of established laws. If the judge makes it about feelings, the lawsuit could easily move to a higher court.



AMD wants to beat Intel to a process. Intel, has historically always been one process ahead of AMD so they only way they will beat Intel is if they skip a process. That could actually be the boost AMD needs...



They were scheduled for Q1 2010 but some of them have been moved up to Q4 2009. I think the reason why it is taking so long to make is because they are based on Netburst architecture which benefits greatly from a lot of tinkering; moreover, they are moving the northbridge to the chip which adds additional development time.

It was also posted a couple of months back that they purposly postponed the i5 launch, due to the lack of competition, or something like that. I think they just wanted to clear some 775 stuff from the shelves first.

 

[a_ump]

still doesn't make sense to me for intel to postpone it when, correct me if i'm wrong, AMD is doing very good in the mainstream market(possibly winning?). I think the only reason why AMD isn't dominating mainstream right now is that the majority of people are oblivious to the great performance per dollar amd's CPU's offer, especially the x3's.

 

[btarunr]

I was thinking the same thing actually. It wasn't as bad of a stumble as AMD's 65nm but a stumble nevertheless.

It depends on what your definition of "stumble" is.

If it means that the manufacturing process technology was bad, Intel had the worse stumble. AMD's 2.8 GHz 1MB L2 chip, the Athlon64 FX 57 ran cooler than Intel Pentium 4 510 (2.80 GHz, 1MB L2 cache). AMD's 90nm Toledo dual-core chips ran cooler than Intel's 65 nm Presler dual-core.



If it means that the processor architecture was bad, AMD had the worse stumble with 65 nm, because by the time they went 65nm, Intel already had the better architecture, which forced AMD to step up clock-speeds, and TDP, beyond what the architecture was originally capable of. This is why you've seen a 2.6 GHz quad-core AMD chip come with a TDP of 125W.

 

[FordGT90Concept]

Changing process means lower TDP which in turn should mean higher clockspeeds (unless there is a major architectural change). Going from 90nm to 65nm, AMD had to adapt half multipliers because they found they couldn't push the speeds at all.

The fastest AMD 90nm processor was the Athlon 64 X2 6400+ (Windsor) at 3.2 GHz.
The fastest AMD 65nm processor was the Athlon 64 X2 6000+ (Brisbane) at 3.1 GHz.

AMD gained nothing except lower power consumption (the architectures are virtually the same).


The fastest Intel 180nm was the Pentium 4 2.0 (Willamette) at 2.0 GHz.
The fastest Intel 130nm was the Pentium 4 EE 3.46 (Gallatin) at 3.46 GHz
The fastest Intel 90nm was the Pentium 4 HT 571 and Pentium 4 HT 672 (Prescott, Prescott 2M) at 3.8 GHz.
The fastest Intel 65nm was the Pentium EE 965 (Presler dual-core) at 3.73 GHz.

The architectures actually got a lot more complex throughout Pentium 4 with the addtion of SSE3, Hyper-Threading, Intel 64, XD bit, and Enhanced Intel SpeedStep Technology. Dispite this, Intel's clockspeeds, for the most part, climbed. It looks like, if anything, 180nm and 65nm were flawed.

If we look at the dates, 180nm didn't get very high because there was less than a year between its launch and the launch of 130nm processors. Perhaps it was flawed, perhaps not.

65nm didn't get very high because it launched just 6 months before Core 2. Perhaps it, too, was flawed but there is no way to be certain.

 

[Tatty_Two]

Lawsuits aren't about feelings, they are about possible violations of established laws. If the judge makes it about feelings, the lawsuit could easily move to a higher court.



AMD wants to beat Intel to a process. Intel has historically been one process ahead of AMD so the only way they could beat Intel is if they skip a process. That could actually be the boost AMD needs...



They were scheduled for Q1 2010 but some of them have been moved up to Q4 2009. I think the reason why it is taking so long to make is because they are based on Netburst architecture which benefits greatly from a lot of tinkering; moreover, they are moving the northbridge to the chip which adds additional development time.

AMD however, even if they skip a process need to up their game, process improvement is good but cant substitute for architectural development and efficiency....they have shown before that their newer processes have had little performance advantage over intels older process.

 

[btarunr]

Changing process means lower TDP which in turn should mean higher clockspeeds (unless there is a major architectural change). Going from 90nm to 65nm, AMD had to adapt half multipliers because they found they couldn't push the speeds at all.

The fastest AMD 90nm processor was the Athlon 64 X2 6400+ (Windsor) at 3.2 GHz.
The fastest AMD 65nm processor was the Athlon 64 X2 6000+ (Brisbane) at 3.1 GHz.

AMD gained nothing except lower power consumption (the architectures are virtually the same).


The fastest Intel 180nm was the Pentium 4 2.0 (Willamette) at 2.0 GHz.
The fastest Intel 130nm was the Pentium 4 EE 3.46 (Gallatin) at 3.46 GHz
The fastest Intel 90nm was the Pentium 4 HT 571 and Pentium 4 HT 672 (Prescott, Prescott 2M) at 3.8 GHz.
The fastest Intel 65nm was the Pentium EE 965 (Presler dual-core) at 3.73 GHz.

The architectures actually got a lot more complex throughout Pentium 4 with the addtion of SSE3, Hyper-Threading, Intel 64, XD bit, and Enhanced Intel SpeedStep Technology. Dispite this, Intel's clockspeeds, for the most part, climbed. It looks like, if anything, 180nm and 65nm were flawed.

If we look at the dates, 180nm didn't get very high because there was less than a year between its launch and the launch of 130nm processors. Perhaps it was flawed, perhaps not.

65nm didn't get very high because it launched just 6 months before Core 2. Perhaps it, too, was flawed but there is no way to be certain.

For convenience sake, use "the highest clock speed" instead of "the fastest" The fastest 65 nm Intel CPU was the Core 2 Extreme QX6850.
Right, and the highest clock speed for a Core 2 65 nm CPU was 3.00 GHz (E6850 or QX6850), and that for a 45 nm Core 2 was 3.33 GHz (E8600). Nothing spectacular.

 

[FordGT90Concept]

I was only looking at the linage of a single architecture: Netburst for Intel and K8 for AMD. That minimizes changes in the architecture which means most of the changes in clockspeed are the result of a process change. When the architecture is changed, clockspeeds can't be used for reference unless the instructions per cycle are about equal (between Core 2 and Netburst, they aren't).

Core 2 clockspeeds scaled well between processes (more than we can say about AMD's 65nm). In fact, the percentage of change between Gallatin and Prescott was +9.8%, the percentage of change between Conroe and Penryn was +11% (virtually the same).

 

[mdm-adph]

Considering USA is about the only country that still conducts wars, most countries have little use for landmines anymore. We do. Landmines are excellent area denial weapons and much cheaper than $100,000 bombs.

Yep -- spoken like someone who doesn't have to live in a country with ancient landmines sitting around still blowing up people every month or so -- people who usually had nothing to do with the war in which those landmines were even used. :shadedshu

It just seems odd to me they haven't released them yet. Not really odd I understand, but i7 was released in what November 08 I think. And now here we are 8 months later and the 3 original models released at launch are still the only 3 next gen models you can get. They keep talking about all these other procs (i7 replacements, i5) yet they haven't seemed to move any closer to release. Like I said, I think they're just lying in wait until AMD threatens them enough.

Anyhow I wouldn't say I'm concerned, I just think it's funny.

THIS is what I've been talking about. Intel probably has these chips sitting around -- they could already sell them if they wanted to. You could already be enjoying your zippy, low-cost i5 goodness, but Intel doesn't give a crap about what you want.

 

[FordGT90Concept]

Yep -- spoken like someone who doesn't have to live in a country with ancient landmines sitting around still blowing up people every month or so -- people who usually had nothing to do with the war in which those landmines were even used. :shadedshu

Most of those are Russian/Soviet. USA almost always does a clean up afterwards unless it is something like North Korea or Vietnam where we had to beat a hasty retreat.


There's no indication Core i5s are ready to ship. That's a rumor.

 

[Deleted member 67555]

Remember there was the Athlon XP vs the Pentium 4, AMD had the better architecture.

There was the Athlon MP vs the P4 Xeon, AMD had the better architecture.

There was the AMD Duron vs the Celeron, AMD had the better architecture.

There was the Athlon FX vs the Pentium 4 Extreme Edition, AMD had the better architecture

There was the Celeron D vs empron, AMD had the better architecture.

There was the Athlon 64 vs the Pentium D, AMD had the better architecture

There was the Athlon X2 64 vs the Pentium D, AMD had the better architecture.

RIGHT! but timing is everything And Intel was first a lot

 

[mdm-adph]

Most of those are Russian/Soviet. USA almost always does a clean up afterwards unless it is something like North Korea or Vietnam where we had to beat a hasty retreat.

Then why doesn't the US sign the friggin' treaty...

There's no indication Core i5s are ready to ship. That's a rumor.

I remember back when we were using PentiumII 266's.

That was as fast as it could go, Intel said. 300MHz was impossible, Intel said. We should just be happy with 266, Intel said.

Then AMD came out with a 266MHz K2 (or whatever it was, can't remember), and Intel has a 300MHz chip out in a week.

I don't believe what Intel says for a second.

 

[a_ump]

i'd have to say i def think intel has i5 ready and could launch now if they wanted. With their money there's no reason for them not to have completed i5, especially when there were rumors of it being released during spring which didn't happen, and i think it's because intel doesn't feel it needs to. If core2 are making profits still for them in mainstream and i7 is paying for itself then they're in the plus and from a marketing standpoint why release something that need not be released when current products are doing the job fine. So i think intel will wait till Q1 2010 unless AMD release something by then that intel feels could be a threat. Anyone know when AMD's next architecture is supposed to arrive or around it's expected release date?

EDIT: i think AMD is doing good though as they're ahead of schedule according to their roadmap. Regor or Athlon II x2 wasn't planned till july-aug time but it was released what? like beginning of june
http://www.tomshardware.com/news/amd-deneb-heka-propus,6364.html

 

[FordGT90Concept]

Then why doesn't the US sign the friggin' treaty...

Because Kim Jung Il is a one-way street: take, take, take, no give. You do know that we are heading towards Korean War Pt. II, right? Why sign a treaty and go back on it a year later? The relationship is tenuous at best right now...

I remember back when we were using PentiumII 266's.

That was as fast as it could go, Intel said. 300MHz was impossible, Intel said. We should just be happy with 266, Intel said.

Then AMD came out with a 266MHz K2 (or whatever it was, can't remember), and Intel has a 300MHz chip out in a week.

I don't believe what Intel says for a second.

I still have a working Pentium II.

You have to realize we are talking about a corporation here. Not everyone knows everything there is about a line of processors. Just recently, there was a spokesman for Intel that said (Intel Developer Forum I believe, both are on TPU) that no "Core i5" exists. A few days later, Intel announced future Core i7, Core i5, and Xeon processors.

I think what both instances show is Intel needs to train their people better in dealing with the press. That is, you can release this information and anything beyond that, "no comment." It doesn't necessarily mean Intel lied, it just means the people that spoke didn't have the information asked of them and therefore shouldn't say anything at all...but did.

 

[mdm-adph]

Because Kim Jung Il is a one-way street: take, take, take, no give. You do know that we are heading towards Korean War Pt. II, right? Why sign a treaty and go back on it a year later? The relationship is tenuous at best right now...

Don't make me laugh. The US will do nothing to piss of China -- they sell it all its cheap stuff, not to mention hold all its debt. (Let's not try to pretend that NK's existence isn't useful to China, either.) Hell, I'd be willing to bet that NK could actually lob a ICBM in its direction, and the US reaction would be to do nothing (can't say I blame it).

And this is really getting off-topic, but AFAIK, there are no use for landmines in modern warfare. :shadedshu You honestly think you're going to be fighting in trenches again? Or even fighting on your own soil? The only wars the US fights are wars on other peoples' land. There's no need to put landmines somewhere that you don't own.

You have ICBM's -- you have nukes. The only wars that a country fights when it has those things are wars that it starts itself. No one's going to invade you. You don't need to defend anything.

I still have a working Pentium II.

You have to realize we are talking about a corporation here. Not everyone knows everything there is about a line of processors. Just recently, there was a spokesman for Intel that said (Intel Developer Forum I believe, both are on TPU) that no "Core i5" exists. A few days later, Intel announced future Core i7, Core i5, and Xeon processors.

I think what both instances show is Intel needs to train their people better in dealing with the press. That is, you can release this information and anything beyond that, "no comment." It doesn't necessarily mean Intel lied, it just means the people that spoke didn't have the information asked of them and therefore shouldn't say anything at all...but did.

Now, don't get me wrong -- I'm not accusing Intel of outright lying. Just stretching the hell out of the truth.

 

[Wile E]

Then why doesn't the US sign the friggin' treaty...



I remember back when we were using PentiumII 266's.

That was as fast as it could go, Intel said. 300MHz was impossible, Intel said. We should just be happy with 266, Intel said.

Then AMD came out with a 266MHz K2 (or whatever it was, can't remember), and Intel has a 300MHz chip out in a week.

I don't believe what Intel says for a second.

Because Kim Jung Il is a one-way street: take, take, take, no give. You do know that we are heading towards Korean War Pt. II, right? Why sign a treaty and go back on it a year later? The relationship is tenuous at best right now...



I still have a working Pentium II.

You have to realize we are talking about a corporation here. Not everyone knows everything there is about a line of processors. Just recently, there was a spokesman for Intel that said (Intel Developer Forum I believe, both are on TPU) that no "Core i5" exists. A few days later, Intel announced future Core i7, Core i5, and Xeon processors.

I think what both instances show is Intel needs to train their people better in dealing with the press. That is, you can release this information and anything beyond that, "no comment." It doesn't necessarily mean Intel lied, it just means the people that spoke didn't have the information asked of them and therefore shouldn't say anything at all...but did.

Don't make me laugh. The US will do nothing to piss of China -- they sell it all its cheap stuff, not to mention hold all its debt. (Let's not try to pretend that NK's existence isn't useful to China, either.) Hell, I'd be willing to bet that NK could actually lob a ICBM in its direction, and the US reaction would be to do nothing (can't say I blame it).

And this is really getting off-topic, but AFAIK, there are no use for landmines in modern warfare. :shadedshu You honestly think you're going to be fighting in trenches again? Or even fighting on your own soil? The only wars the US fights are wars on other peoples' land. There's no need to put landmines somewhere that you don't own.

You have ICBM's -- you have nukes. The only wars that a country fights when it has those things are wars that it starts itself. No one's going to invade you. You don't need to defend anything.



Now, don't get me wrong -- I'm not accusing Intel of outright lying. Just stretching the hell out of the truth.

OK guys, drop the political debate. Go to generalnonsense.net for that.


And both companies "lie". All major corporations do, especially about products under NDA. Get over it.

 

[FordGT90Concept]

there are no use for landmines in modern warfare.

You capture a bridge and you don't want anyone crossing it. At the same time, you don't want to station troops there to guard it either. What do you do? Mine both sides of it with big ass signs saying "minefield." Not too likely anyone is going to attempt to cross it. If someone does and they lose their legs, that's reason enough for no one else to try. Again, mines are "area denial weapons."

And both companies "lie". All major corporations do, especially about products under NDA. Get over it.

If the product is under NDA, it could be intentional misdirection. Still lying and they still shouldn't say anything at all. XD