AMD Barcelona @ 2Ghz beating Intel Clovertown @ 2.33Ghz by 25% on SPEC CPU

[coldpower27]

Originally posted by: ViRGE

Originally posted by: Nemesis 1
Of course you know Intel is going to drop prices like crazy. right . Why shouldn't they . AMD did. Clear out inventory for penryns. Plus penryn is going to be much cheaper for intel to produce than K10 . Yields will be a lot higher also.

That's all conjecture at this point. Penryn will only be cheaper if Intel is getting higher yields and/or the core is smaller than Conroe. We don't know either of those, and it's actually rather unlikely Penryn will be significantly bigger/smaller than Conroe, seeing as Intel likes to stick with an average size for a CPU die. As for yields, Conroe has yielded very well, I wouldn't expect Penryn to be able to do much more than equal it.

Actually we already know that, the die size of Penryn is 107mm2 for the 6MB core so it would be much better for Intel to move over to it over Conroe.

http://www.firingsquad.com/har...tel_penryn_sneak_peek/

The pricing point is already known as well for Servers as DailyTech has already released a list of the pricing points for the Harpertown units.

http://www.dailytech.com/Intel...+Xeons/article8074.htm

We will have to see as we approach Wolfdale and Yorkfield's launch if Intel will lower the prices on existing stock of Conroe and Kentsfield, which is entirely possible.

 

[coldpower27]

Originally posted by: classy
Well it looks like barcy will be making some noise. Fud has a nice tidbit on the 30k 3dmark06 report. While it won't hit 30k, looks like someone has hit 26800, lol.

Fud link

26800 seems to be hit with a Core 2 Duo at 5.x GHZ and 2 extreme overclocked 8800 Ultra's.

 

[ViRGE]

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: Nemesis 1
Of course you know Intel is going to drop prices like crazy. right . Why shouldn't they . AMD did. Clear out inventory for penryns. Plus penryn is going to be much cheaper for intel to produce than K10 . Yields will be a lot higher also.

That's all conjecture at this point. Penryn will only be cheaper if Intel is getting higher yields and/or the core is smaller than Conroe. We don't know either of those, and it's actually rather unlikely Penryn will be significantly bigger/smaller than Conroe, seeing as Intel likes to stick with an average size for a CPU die. As for yields, Conroe has yielded very well, I wouldn't expect Penryn to be able to do much more than equal it.

Actually we already know that, the die size of Penryn is 107mm2 for the 6MB core so it would be much better for Intel to move over to it over Conroe.

http://www.firingsquad.com/har...tel_penryn_sneak_peek/

The pricing point is already known as well for Servers as DailyTech has already released a list of the pricing points for the Harpertown units.

http://www.dailytech.com/Intel...+Xeons/article8074.htm

We will have to see as we approach Wolfdale and Yorkfield's launch if Intel will lower the prices on existing stock of Conroe and Kentsfield, which is entirely possible.

Oh great, I didn't realize Intel had published that info yet. I'm a bit surprised TBH, I was expecting something bigger, 107mm2 is exceptionally small for their high end product (they haven't been this small since the P3 days), so they will be making good money off of that and even more off of the smaller low-end dice.

 

[ViRGE]

Eep, double post.

 

[coldpower27]

Originally posted by: ViRGE

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: Nemesis 1
Of course you know Intel is going to drop prices like crazy. right . Why shouldn't they . AMD did. Clear out inventory for penryns. Plus penryn is going to be much cheaper for intel to produce than K10 . Yields will be a lot higher also.

That's all conjecture at this point. Penryn will only be cheaper if Intel is getting higher yields and/or the core is smaller than Conroe. We don't know either of those, and it's actually rather unlikely Penryn will be significantly bigger/smaller than Conroe, seeing as Intel likes to stick with an average size for a CPU die. As for yields, Conroe has yielded very well, I wouldn't expect Penryn to be able to do much more than equal it.

Actually we already know that, the die size of Penryn is 107mm2 for the 6MB core so it would be much better for Intel to move over to it over Conroe.

http://www.firingsquad.com/har...tel_penryn_sneak_peek/

The pricing point is already known as well for Servers as DailyTech has already released a list of the pricing points for the Harpertown units.

http://www.dailytech.com/Intel...+Xeons/article8074.htm

We will have to see as we approach Wolfdale and Yorkfield's launch if Intel will lower the prices on existing stock of Conroe and Kentsfield, which is entirely possible.

Oh great, I didn't realize Intel had published that info yet. I'm a bit surprised TBH, I was expecting something bigger, 107mm2 is exceptionally small for their high end product (they haven't been this small since the P3 days), so they will be making good money off of that and even more off of the smaller low-end dice.

Well Conroe was small to begin with, truth be told and if you look at the Pentium 4 each successive generation has gotten smaller.

Willamette -> Northwood -> Prescott -> Cedar Mill

217mm2 -> 131mm2 -> 112mm2 -> 81mm2

Die sizes usually grow when an architectural overhaul occurs. And you got to keep in mind, Intel intended to be building Quad Core with these dice, so 2x107mm2 are going to be used in fairly large amounts so the small dice is just what the doctor ordered.

 

[ViRGE]

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: Nemesis 1
Of course you know Intel is going to drop prices like crazy. right . Why shouldn't they . AMD did. Clear out inventory for penryns. Plus penryn is going to be much cheaper for intel to produce than K10 . Yields will be a lot higher also.

That's all conjecture at this point. Penryn will only be cheaper if Intel is getting higher yields and/or the core is smaller than Conroe. We don't know either of those, and it's actually rather unlikely Penryn will be significantly bigger/smaller than Conroe, seeing as Intel likes to stick with an average size for a CPU die. As for yields, Conroe has yielded very well, I wouldn't expect Penryn to be able to do much more than equal it.

Actually we already know that, the die size of Penryn is 107mm2 for the 6MB core so it would be much better for Intel to move over to it over Conroe.

http://www.firingsquad.com/har...tel_penryn_sneak_peek/

The pricing point is already known as well for Servers as DailyTech has already released a list of the pricing points for the Harpertown units.

http://www.dailytech.com/Intel...+Xeons/article8074.htm

We will have to see as we approach Wolfdale and Yorkfield's launch if Intel will lower the prices on existing stock of Conroe and Kentsfield, which is entirely possible.

Oh great, I didn't realize Intel had published that info yet. I'm a bit surprised TBH, I was expecting something bigger, 107mm2 is exceptionally small for their high end product (they haven't been this small since the P3 days), so they will be making good money off of that and even more off of the smaller low-end dice.

Well Conroe was small to begin with, truth be told and if you look at the Pentium 4 each successive generation has gotten smaller.

Willamette -> Northwood -> Prescott -> Cedar Mill

217mm2 -> 131mm2 -> 112mm2 -> 81mm2

Die sizes usually grow when an architectural overhaul occurs. And you got to keep in mind, Intel intended to be building Quad Core with these dice, so 2x107mm2 are going to be used in fairly large amounts so the small dice is just what the doctor ordered.

Cedar Mill however wasn't really a high end part, that was Pressler with 2 cores(162mm2). And even the Prescott was quickly replaced with the Prescott2M(135mm2). And then there with Smithfield, which was massive. Intel's best parts generally are bigger rather than smaller, Northwood really was the smallest high end part that actually lasted a generation.*

* It's a pedantic definition, but the point I'm getting at is that Intel hasn't managed to ship a small part as a long-lasting high-end part for quite a while

 

[coldpower27]

Originally posted by: ViRGE

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: coldpower27

Originally posted by: ViRGE

Originally posted by: Nemesis 1
Of course you know Intel is going to drop prices like crazy. right . Why shouldn't they . AMD did. Clear out inventory for penryns. Plus penryn is going to be much cheaper for intel to produce than K10 . Yields will be a lot higher also.

That's all conjecture at this point. Penryn will only be cheaper if Intel is getting higher yields and/or the core is smaller than Conroe. We don't know either of those, and it's actually rather unlikely Penryn will be significantly bigger/smaller than Conroe, seeing as Intel likes to stick with an average size for a CPU die. As for yields, Conroe has yielded very well, I wouldn't expect Penryn to be able to do much more than equal it.

Actually we already know that, the die size of Penryn is 107mm2 for the 6MB core so it would be much better for Intel to move over to it over Conroe.

http://www.firingsquad.com/har...tel_penryn_sneak_peek/

The pricing point is already known as well for Servers as DailyTech has already released a list of the pricing points for the Harpertown units.

http://www.dailytech.com/Intel...+Xeons/article8074.htm

We will have to see as we approach Wolfdale and Yorkfield's launch if Intel will lower the prices on existing stock of Conroe and Kentsfield, which is entirely possible.

Oh great, I didn't realize Intel had published that info yet. I'm a bit surprised TBH, I was expecting something bigger, 107mm2 is exceptionally small for their high end product (they haven't been this small since the P3 days), so they will be making good money off of that and even more off of the smaller low-end dice.

Well Conroe was small to begin with, truth be told and if you look at the Pentium 4 each successive generation has gotten smaller.

Willamette -> Northwood -> Prescott -> Cedar Mill

217mm2 -> 131mm2 -> 112mm2 -> 81mm2

Die sizes usually grow when an architectural overhaul occurs. And you got to keep in mind, Intel intended to be building Quad Core with these dice, so 2x107mm2 are going to be used in fairly large amounts so the small dice is just what the doctor ordered.

Cedar Mill however wasn't really a high end part, that was Pressler with 2 cores(162mm2). And even the Prescott was quickly replaced with the Prescott2M(135mm2). And then there with Smithfield, which was massive. Intel's best parts generally are bigger rather than smaller, Northwood really was the smallest high end part that actually lasted a generation.*

* It's a pedantic definition, but the point I'm getting at is that Intel hasn't managed to ship a small part as a long-lasting high-end part for quite a while

Well if your going to go with what was "high end at the time", you have an up and down scale if you include all high end products.

Coppermine -> Willamette -> Northwood -> Gallatin-2M -> Prescott 2M -> Smithfield -> Presler -> Conroe -> Kentsfield -> Yorkfield

Up -> Down -> Up -> Down -> Up -> Down -> Down -> Up - Down.

Suffice it to say though, any shrink/derivative generation is going to be smaller then it previous successor and this is a shrink/derivative generation, so die sized shrunk 🙂

That definition doesn't hold then as 107mm2 only refers to Wolfdale/Penryn which are the Dual Core variants which aren't high end, if your going to combine the Presler MCM item into 162mm2 from the 2x81mm2 unit then Yorkfield MCM is 2x107mm2 or 214mm2.

Smithfield is massive? Not really, compared to Gallatin, Kentsfield, or AMD's Clawhammer or Toledo.

Does Intel need to make small parts for the high end arena, anyway considering all their capacity?

 

[T2k]

Originally posted by: Viditor

Originally posted by: T2k

Originally posted by: Accord99
It's also quite telling that after 9 months, SPECfp_rate is still the only benchmark AMD is willing to talk about.

I agree, it's pretty pathetic, paints a pretty dark picture about AMD's next 6-8 months (until they can ramp up Barcelona's clock above ~2.5GHz where it begins to be competitive).

They weren't showing off the spec benches, what they were showing was the power/performance...
You should also keep in mind that that was an older stepping than will be released. The reason for the delay on Barcelona is that they had to wait until the newer steppings were done and ready for shipping.

I got that part but you're daydreaming if you think that could make more than 5% difference...

Edit: BTW, the 2.5 GHz+ is supposed to be released within 2 months, not 6-8.

Errr, no, you're wrong. Since they don't even have a 2.4GHz unit stable NOW then it's impossible they woul;d have a 2.6GHz 2 months from now -I BET it will be 2008 when you will see 2.6GHz en masse.

 

[Mr Vain]

Originally posted by: T2k

Originally posted by: Viditor

Originally posted by: T2k

Originally posted by: Accord99
It's also quite telling that after 9 months, SPECfp_rate is still the only benchmark AMD is willing to talk about.

I agree, it's pretty pathetic, paints a pretty dark picture about AMD's next 6-8 months (until they can ramp up Barcelona's clock above ~2.5GHz where it begins to be competitive).

They weren't showing off the spec benches, what they were showing was the power/performance...
You should also keep in mind that that was an older stepping than will be released. The reason for the delay on Barcelona is that they had to wait until the newer steppings were done and ready for shipping.

I got that part but you're daydreaming if you think that could make more than 5% difference...

Edit: BTW, the 2.5 GHz+ is supposed to be released within 2 months, not 6-8.

Errr, no, you're wrong. Since they don't even have a 2.4GHz unit stable NOW then it's impossible they woul;d have a 2.6GHz 2 months from now -I BET it will be 2008 when you will see 2.6GHz en masse.

AMD?s upcoming Phenom processor clocked at impressive 3GHz.

The CPU was not that hot and it was cooled with a standard cooler.

http://www.fudzilla.com/index....view&id=2585&Itemid=51

http://www.ocworkbench.com/200...AMD-Akiba-Event/g1.htm

The 3 GHz processor that was being used was stable enough that AMD displayed it publicly.

🙂

 

[swtethan]

Originally posted by: Mr Vain

Originally posted by: T2k

Originally posted by: Viditor

Originally posted by: T2k

Originally posted by: Accord99
It's also quite telling that after 9 months, SPECfp_rate is still the only benchmark AMD is willing to talk about.

I agree, it's pretty pathetic, paints a pretty dark picture about AMD's next 6-8 months (until they can ramp up Barcelona's clock above ~2.5GHz where it begins to be competitive).

They weren't showing off the spec benches, what they were showing was the power/performance...
You should also keep in mind that that was an older stepping than will be released. The reason for the delay on Barcelona is that they had to wait until the newer steppings were done and ready for shipping.

I got that part but you're daydreaming if you think that could make more than 5% difference...

Edit: BTW, the 2.5 GHz+ is supposed to be released within 2 months, not 6-8.

Errr, no, you're wrong. Since they don't even have a 2.4GHz unit stable NOW then it's impossible they woul;d have a 2.6GHz 2 months from now -I BET it will be 2008 when you will see 2.6GHz en masse.

AMD?s upcoming Phenom processor clocked at impressive 3GHz.

The CPU was not that hot and it was cooled with a standard cooler.

http://www.fudzilla.com/index....view&id=2585&Itemid=51

http://www.ocworkbench.com/200...AMD-Akiba-Event/g1.htm

The 3 GHz processor that was being used was stable enough that AMD displayed it publicly.

🙂

3ghz is impressive eh? lol

 

[Regs]

Originally posted by: T2k

Errr, no, you're wrong. Since they don't even have a 2.4GHz unit stable NOW then it's impossible they woul;d have a 2.6GHz 2 months from now -I BET it will be 2008 when you will see 2.6GHz en masse.

Server segment requires more features to satisfy the customers.

Desktop customers don't need as many features.

AMD disables a few power saving/cooling ide modes and some CPUS can reach up to 3.4 GHz if not more.

 

[nyker96]

But I think we need to keep in mind, even if AMD is faster it's for server applications. And if their pricing is an indication then maybe true that they are faster by about 20% on server apps. This however, can't translate into desktop apps like games which is very sensitive to cache etc and Intel has upper hand on that. But I think AMD as an architecture scales better as you OC quad core since HTT keeps the processor fed nicely while Intel chips has some problem with scaling.

 

[Mr Vain]

Dell Revealed Barcelona Performance

Dell has no intention of reversing its decision last year to start using AMD processors in its product line. The company wanted to give customers a variety of choices and also because it wanted to guarantee for itself a steady supply of electronic components. Dell expects AMD to introduce its quad-core chip, code-named Barcelona, next week and, comparing it with Intel's Clovertown quad-core processor, the company would like to offer customers a choice of either product. "If you look at floating point instructions, Barcelona is about 30 percent faster than Clovertown. However, if you look at integer instructions, Clovertown is about 30 percent faster than Barcelona," Dell said.


http://www.vr-zone.com/article..._Performance/5239.html

 

[T2k]

Originally posted by: Mr Vain

Originally posted by: T2k

Originally posted by: Viditor

Originally posted by: T2k

Originally posted by: Accord99
It's also quite telling that after 9 months, SPECfp_rate is still the only benchmark AMD is willing to talk about.

I agree, it's pretty pathetic, paints a pretty dark picture about AMD's next 6-8 months (until they can ramp up Barcelona's clock above ~2.5GHz where it begins to be competitive).

They weren't showing off the spec benches, what they were showing was the power/performance...
You should also keep in mind that that was an older stepping than will be released. The reason for the delay on Barcelona is that they had to wait until the newer steppings were done and ready for shipping.

I got that part but you're daydreaming if you think that could make more than 5% difference...

Edit: BTW, the 2.5 GHz+ is supposed to be released within 2 months, not 6-8.

Errr, no, you're wrong. Since they don't even have a 2.4GHz unit stable NOW then it's impossible they woul;d have a 2.6GHz 2 months from now -I BET it will be 2008 when you will see 2.6GHz en masse.

AMD?s upcoming Phenom processor clocked at impressive 3GHz.

The CPU was not that hot and it was cooled with a standard cooler.

http://www.fudzilla.com/index....view&id=2585&Itemid=51

http://www.ocworkbench.com/200...AMD-Akiba-Event/g1.htm

The 3 GHz processor that was being used was stable enough that AMD displayed it publicly.

🙂

Display publicly != masss manufacturing

Besides there are words like yield, wattage etc.

Again: since they do not even have a 2.4GHz to launch they won't have a 2.6GHz before the end of this year. I'm talking about mass shipments, remember.

 

[JEDIYoda]

Originally posted by: Mr Vain
AMD Barcelona @ 2 GHz beating Intel Clovertown @ 2.33 GHz by 25% on SPEC CPU while sucking less power.

http://www.vnunet.com/tv/?chan...lipid=1386_vnunet_0096

VNUnet.com came up with a video showing AMD Barcelona floating point scores when compared to Intel quad-core Clovertown at sub-90W Thermal envelope. In case of Intel, processor worked at 2.33 GHz (Xeon 5345), while Barcelona worked at 2.0 GHz (Opteron 2350).

A positive start IMO.

Anyone else got some more positive info on K10?


🙂

One benchmark out of how many...lol