Sign in with
Sign up | Sign in
Your question

amd bulldozer: will it bulldoze or will it be bulldozed?

Last response: in CPUs
Share
February 10, 2008 8:12:21 PM

During the past week, I've been checking out the latest articles on Intel's nehalem and amd's bulldozer. Since a lot of people are saying that AMD is unable to compete in the processor market in 2008, I wanted to ask you guys about your predictions for next year. Do you think that the "Bulldozer" will be as good as AMD says it will be? Or will it be another Quad FX?

Here is the article that I found about the bulldozer.

http://www.tech.co.uk/computing/upgrades-and-peripheral...
a c 126 à CPUs
a b À AMD
February 10, 2008 8:22:04 PM

Consideing how great Phenom was said to be and what it turned out to be I have little faith in Bulldozer. And considering what Intel stated Core2 would be abd what it turned out to be I have a lot of faith in Nehalem.

Back when Intel announced Core2 I was skepticle. Still loved Intel CPUs but not very sure of its real power. And Phenom I was expecting to be big since A64/K8 was really good.

Thing is that Intel has a lot fo the features AMD's Bulldozer is to have coming in Nehalem. Not to mention that QPI shows a very large memory bandwisth. That and the fact that they want to get to 16 cores in a jump from quad core doesn't sound promising.

Guess we will have to wait and see. I still have more faith that Nehalem will "bulldoze" Bulldozer.
February 10, 2008 8:25:18 PM

Most likely true. The 16 core design will probably still underperform the best nehalem cpu, not to mention produce to much heat.
Related resources
February 10, 2008 8:30:13 PM

With AMD's process technology, I think 8 cores will be more likely than 16 cores. They're already having problems with yielding monolithic quad cores on their 65nm technology, and I speculate that monolithic 16 cores will have to wait for the maturation of 22nm.
February 10, 2008 9:11:59 PM

yomamafor1 that might be the case but y cant 16 cores e possible. they dont say on the release date we will release a moster 16 core cpu. "The big news is that it will form the basis of AMD's first massively multi-core PC processor with up to 16 execution cores". so onrelease they could be 2, 4 or my 8 cores until die shrinks then there plans of 16 cores..... maybe
February 10, 2008 9:16:55 PM

My take on bulldozer ... ask again in 18 months.
a c 126 à CPUs
a b À AMD
February 10, 2008 9:22:25 PM

coret said:
My take on bulldozer ... ask again in 18 months.


LOL. Nice answer.

As for that, right now I doubt AMD will be able to release Bulldozer on time sinc now they have to team up with other companies to find a way to produce HK/MG transitior technolgies so their 45nm will be delayed. As for 22nm mono 16 core I doubt it will be any sooner than 2010 then since we all know that Intel will probably hit that process first.

I just wounder if Intel will have a 16 core. Or they could always be secretly getting their terascale chip ready and perfected for mainstream. That would just put Bulldozer to bed.
February 10, 2008 9:23:16 PM

sorry, just have to get this off before I die of laughter......

AMD4LIFE!
February 10, 2008 9:25:17 PM

bmadd said:
yomamafor1 that might be the case but y cant 16 cores e possible. they dont say on the release date we will release a moster 16 core cpu. "The big news is that it will form the basis of AMD's first massively multi-core PC processor with up to 16 execution cores". so onrelease they could be 2, 4 or my 8 cores until die shrinks then there plans of 16 cores..... maybe


I only specifically stated 16 cores, so whether Bulldozer comes in 16 cores, 8 cores, or 4 cores, the 16 core version of it will be economically viable, and mature only when AMD's 22nm is ready. As for other flavors, I think they'll have to wait for HK/MG implemented 32nm before Bulldozer is viable.

According to the link posted by OP, it seems like Bulldozer is like Prescott (long pipeline, high clockspeed), which relies on superior process technology to make it shine. In other words, if AMD can't successfully implement Bulldozer with 45nm (which seems very very likely), AMD will again have to shell out the resources and time to wait for IBM to finish development on the process. We can be talking about 2009~2010 at this point.

Again, everything is simply speculation at this point. I wouldn't be surprised if AMD postponed the delivery date again to accommodate with the re-design of Bulldozer if Nehalem outperformed their expectations.
a c 126 à CPUs
a b À AMD
February 10, 2008 9:30:00 PM

yomamafor1 said:
I only specifically stated 16 cores, so whether Bulldozer comes in 16 cores, 8 cores, or 4 cores, the 16 core version of it will be economically viable, and mature only when AMD's 22nm is ready. As for other flavors, I think they'll have to wait for HK/MG implemented 32nm before Bulldozer is viable.

According to the link posted by OP, it seems like Bulldozer is like Prescott (long pipeline, high clockspeed), which relies on superior process technology to make it shine. In other words, if AMD can't successfully implement Bulldozer with 45nm (which seems very very likely), AMD will again have to shell out the resources and time to wait for IBM to finish development on the process. We can be talking about 2009~2010 at this point.

Again, everything is simply speculation at this point. I wouldn't be surprised if AMD postponed the delivery date again to accommodate with the re-design of Bulldozer if Nehalem outperformed their expectations.


So from what you are saying AMD is trying 45nm SOI and waiting for HK/MG @ 32nm?

You know I can see Intel sitting and laughing at AMD when they show the specs of this. Its like AMD didn't learn from Prescott but Intel did. The long pipelines and higher clock speeds wont help anything. Not to mention it will kill their IPC and SOI 45nm will probably be a mess. Either it will run too hot and take too much power or it will just not work with all the leakage it will cause.

Then Intel will go "We told ya SOI past 65NM sucks!!!". AMD will probably be very unhappy with the results. This just goes to show ya that AMD is not being managed right. They should have seen this one coming.
February 10, 2008 11:19:33 PM

jimmysmitty said:
So from what you are saying AMD is trying 45nm SOI and waiting for HK/MG @ 32nm?

You know I can see Intel sitting and laughing at AMD when they show the specs of this. Its like AMD didn't learn from Prescott but Intel did. The long pipelines and higher clock speeds wont help anything. Not to mention it will kill their IPC and SOI 45nm will probably be a mess. Either it will run too hot and take too much power or it will just not work with all the leakage it will cause.

Then Intel will go "We told ya SOI past 65NM sucks!!!". AMD will probably be very unhappy with the results. This just goes to show ya that AMD is not being managed right. They should have seen this one coming.


Actually. this wouldn't be AMD's fault (at least not entirely). This is one of the downside to be buying process technology from IBM: you simply can't always get what you wanted. If IBM says SOI for 45nm, you can't say, "but I want HK/MG over SOI".

I'm sure AMD knew d@mn too well that SOI is a lost cause past 90nm. Heat dissipation is mounting up, and electrons are leaking like nuts. This is also one of the reason why AMD needs to thicken their gates by 25%, which in turns lower their ability to clock high. Turpit did a thorough analysis on this, and you can look it up. Then when you have an architecture that has lower IPC than your competition, you will be right in AMD's shoes.

IMO, Prescott was not a failed architecture. It was simply too early for its time, for the process node it was supposed to build on. Neither was Intel's 90nm and 65nm a failure. Intel's 90nm was already on par with AMD's 90nm, and far superior on 65nm node. The combination of Prescott and 90/65nm made it a bad processor. I'm sure if Prescott is built on the 45nm HK/MG today, or even 32nm, we'll be seeing 10Ghz Prescott at very reasonable power consumption and power dissipation.
a c 126 à CPUs
a b À AMD
February 11, 2008 4:21:27 AM

yomamafor1 said:
Actually. this wouldn't be AMD's fault (at least not entirely). This is one of the downside to be buying process technology from IBM: you simply can't always get what you wanted. If IBM says SOI for 45nm, you can't say, "but I want HK/MG over SOI".

I'm sure AMD knew d@mn too well that SOI is a lost cause past 90nm. Heat dissipation is mounting up, and electrons are leaking like nuts. This is also one of the reason why AMD needs to thicken their gates by 25%, which in turns lower their ability to clock high. Turpit did a thorough analysis on this, and you can look it up. Then when you have an architecture that has lower IPC than your competition, you will be right in AMD's shoes.

IMO, Prescott was not a failed architecture. It was simply too early for its time, for the process node it was supposed to build on. Neither was Intel's 90nm and 65nm a failure. Intel's 90nm was already on par with AMD's 90nm, and far superior on 65nm node. The combination of Prescott and 90/65nm made it a bad processor. I'm sure if Prescott is built on the 45nm HK/MG today, or even 32nm, we'll be seeing 10Ghz Prescott at very reasonable power consumption and power dissipation.


100% agreed. I never though Prescott was a failure just too early. But still AMD needs to start doing their own process technology or they will never keep up. Even if IBM says 45nm @ SOI its a huge mistake. And I am sure IBM knows that but doesn't care since AMD will buy it from them.

Right now I think that 65nm is SOI's limit as Intel has provided some nice ships based on SOI i.e. Core2. But now Intel has entered a whole new ball game with HK/MG and they will have an even better version with Nehalem or HK/MG 2. I for one would love to see a 32nm HK/MG Prescott. At 10GHz it would whomp anything :D 
February 11, 2008 4:52:55 AM

jimmysmitty said:
100% agreed. I never though Prescott was a failure just too early. But still AMD needs to start doing their own process technology or they will never keep up. Even if IBM says 45nm @ SOI its a huge mistake. And I am sure IBM knows that but doesn't care since AMD will buy it from them.

Right now I think that 65nm is SOI's limit as Intel has provided some nice ships based on SOI i.e. Core2. But now Intel has entered a whole new ball game with HK/MG and they will have an even better version with Nehalem or HK/MG 2. I for one would love to see a 32nm HK/MG Prescott. At 10GHz it would whomp anything :D 


Intel does not use SOI on any of their processors. They looked at it and decided the benefits did not out weights the problems associated with employing SOI.
February 11, 2008 4:55:33 AM

mikekazik1 said:
During the past week, I've been checking out the latest articles on Intel's nehalem and amd's bulldozer. Since a lot of people are saying that AMD is unable to compete in the processor market in 2008, I wanted to ask you guys about your predictions for next year. Do you think that the "Bulldozer" will be as good as AMD says it will be? Or will it be another Quad FX?

Here is the article that I found about the bulldozer.

http://www.tech.co.uk/computing/upgrades-and-peripheral...



Ummm...that article is kind of dated....from June 07, and appears to be....questionable. According to an Ars Technica article from last month, quoting an Infoworld interview, Bulldozer is on the back burner while AMD concentrates on Fusion

Ars Technica: AMD ditches 'Dozer, taps Phenom for CPU/GPU Fusion

Quote:
But in a financial analyst day meeting early this past December, AMD revealed that the company's first Fusion product would "use existing CPU core and GPU core technologies," suggesting that Bulldozer had gotten the boot in favor of a core from AMD's current lineup.

Here is the Infoworld article
Infoworld: AMD's Fusion chip will be based on Phenom processor



For yomama: If I recall correctly, AMD wasnt going to go monolithic with the 8 cores or higher, but use a MCM setup made of monolithic quads.....of course, that was before their quad failed to hit the gates they set....who knows what their real plans are now.
February 11, 2008 4:57:09 AM

I'd laugh my ass off if Intel announced 32nm Pentium V based on NetBurst 2.
a b à CPUs
February 11, 2008 8:01:21 AM

I don't think you will see nehalem either ... well ... not anytime soon.

No point in giving us any more horsepower since they don't have to do anything more to compete.

In fact Intel should rasie their prices by at least 30% in terms of the value you currently get. They are way too cheap.

That's what happens when you have a virtual monopoly in play ... nothing.

Have the new Penryn Quads shipped yet (Yorkdale) ??
February 11, 2008 8:09:28 AM

jbj190 said:
I'd laugh my ass off if Intel announced 32nm Pentium V based on NetBurst 2.


I'd cry.
a b à CPUs
February 11, 2008 8:12:44 AM

You know, 8 cores, 16 cores, 2048 cores... whatever, isn't going to make a difference for most people. I do everything fine with 2 cores, same as most people, and I don't care whether Intel thinks I need 4096x more cores than that or not, I say NO! to the empire. Of course, if you do video editing, having 65536 cores would probably help to some extent.
February 11, 2008 9:13:42 AM

Not another thread like this PLEASE! ********
a b à CPUs
February 11, 2008 9:24:51 AM

Haha, Intel vs AMD war, it's inevitable!
February 11, 2008 10:07:30 AM

Way, waaaaaay too early to tell...
You can't predict the future from the past!
Logically Phenom should have been pretty good (longer dev time over Core 2 quad, the A64 K8 architecture was pretty good) and look how it turned out, not bad, but certainly not phenomenal... (;))
AMD may really have a kick-ass 45nm Phenom and Intel's fabs may explode, you just can't work anything out.
Can't see either being on time either....
February 11, 2008 10:13:18 AM

Exactly LukeBird, nobody 2 and a half years ago no one would have believed you if you said, "Intel's new processors are going to be much better than AMD's, at lower clocks too!" I would have probably scoffed and said you were a moron, but look at it now.
February 11, 2008 10:21:16 AM

IndigoMoss said:
Exactly LukeBird, nobody 2 and a half years ago no one would have believed you if you said, "Intel's new processors are going to be much better than AMD's, at lower clocks too!" I would have probably scoffed and said you were a moron, but look at it now.

Yup, exactly.
Maybe AMD really will kick ass, as Intel try and integrate native quad & IMC onto a 45nm/32nm process. Maybe it's just too much in one step?
It will certainly be interesting.
And I don't mind taking one for the team and staying with AMD for competitions sake! :D 
February 11, 2008 1:57:31 PM

turpit said:
For yomama: If I recall correctly, AMD wasnt going to go monolithic with the 8 cores or higher, but use a MCM setup made of monolithic quads.....of course, that was before their quad failed to hit the gates they set....who knows what their real plans are now.


If I remember correctly from the last AMD Analyst Day, they're only looking to build monolithic quad cores, then do MCM for 8 cores.

Quite a huge jump from monolithic 16 cores to monolithic 4 cores huh... :kaola: 
February 11, 2008 1:58:11 PM

Reynod said:
I don't think you will see nehalem either ... well ... not anytime soon.

No point in giving us any more horsepower since they don't have to do anything more to compete.

In fact Intel should rasie their prices by at least 30% in terms of the value you currently get. They are way too cheap.

That's what happens when you have a virtual monopoly in play ... nothing.

Have the new Penryn Quads shipped yet (Yorkdale) ??


I wonder where I heard that before... :sarcastic: 
February 11, 2008 2:11:36 PM

I think the future should be based on hetrogeneous cores.

Have one deep pipeline core, that can be clocked up for single thread apps that would use it. Then shallower cores with better IPC for the parallel work. Heck, have an encoding core, have a gaming core, have a low power core for cool 'n' quiet, have a GP-GPU core for parallel calcs.


Multi core offers multi options - there is no need to compromise so much IMO.
February 11, 2008 2:41:53 PM

Its really too early to tell,especially with all of the conflicting articles and the lack luster turn out of Phenom.

I do like the modular idea. If they can get the thread fusion working, There might be a mixbag of results when comparing AMD vs Intel at that point. Itel could do well with heavily multi-threaded apps while AMD might do better with older or single threaded apps, or with games that use fewer than 4 cores. If they can make 8 to 16 cores split 4 or less threads.

Who knows!
February 11, 2008 4:38:48 PM

30 Jul 2007

Article is pretty old, they even refered to QuickPath as CSI which was its original codename.

AMD's bulldozer could have an advantage (at least in bandwidth) in the low end since lower end nehalems will still use the FSB.

I highly doubt bulldozer will have 16 cores, I mean come on.

February 11, 2008 4:46:53 PM

Amiga500 said:
I think the future should be based on hetrogeneous cores.


Are you intollerant of homo cores?
a c 126 à CPUs
a b À AMD
February 11, 2008 5:27:13 PM

Amiga500 said:
I think the future should be based on hetrogeneous cores.

Have one deep pipeline core, that can be clocked up for single thread apps that would use it. Then shallower cores with better IPC for the parallel work. Heck, have an encoding core, have a gaming core, have a low power core for cool 'n' quiet, have a GP-GPU core for parallel calcs.


Multi core offers multi options - there is no need to compromise so much IMO.


Thats what Intels Terascale is. Basically what you want per core. You can mix and match them and have it setup your way. Kinda makes you think Intel should be releasing it instead.

I think Nehalem will be on time for Q4 08 but the mainstream may be a bit later than Q1 09 unless AMD releases a miracle and I doubt that will happen.
February 11, 2008 8:46:31 PM

Reynod said:
I don't think you will see nehalem either ... well ... not anytime soon.

No point in giving us any more horsepower since they don't have to do anything more to compete.

In fact Intel should rasie their prices by at least 30% in terms of the value you currently get. They are way too cheap.

That's what happens when you have a virtual monopoly in play ... nothing.

Have the new Penryn Quads shipped yet (Yorkdale) ??

I think you mean yorkfield [yorkdale.....[lol, snickers] sorry, couldnt resist]..no rumours of US release just yet, but apparently, they are close to release in the UK.

yorkfield
Quote:
He... The brand new Intel 'Yorkfield' Quad-Core CPUs are close to launch and OcUK will be one of the first UK suppliers of this brand new technology so we're making them all available for pre-order today. With improved core improvements over the previous generation combined with being produced on the new 45nm-process technology it really is a worthwhile upgrade for the performance increase over the 'Kentsfield' generation and the bonus of being able to run both cooler and quieter not to mention their overclocking potential! They absolutely tear through CPU-intensive tasks whether it's gaming, multimedia applications, 3D modelling or any other task where CPU-power is king. We have the Q9300, the Q9450 and the daddy Q9550 available to order. So secure yours today and be one of the first in the queue.


If this is true, its a fair bet you will see them in the US soon. Havent the foggiest how long it will take them to ship down under.

Now, why, pray tell, should Intel raise its prices? Why are you basing the relative value with AMDs products as the foundation? Relativity is a two way street. One could just as easily say that AMD is overcharging for its products and needs to lower its prices. But frankly, either assesment is false when you compare the actual performance to the actual retail pricing. If anything, in strict terms of performance vs price, Intel is overcharging for some of its dual cores, (see E6600 vs X2 6400)and the quad pricing is fairly balanced, following the norm of a parabolic curve with price increasing disproportionally with increases in clockspeed.


BTW what was with that idiot down there who lit himself on fire and tried to jump a bicycle off a platform into a plie of mattresses?
February 11, 2008 8:53:18 PM

TechnologyCoordinator said:
Are you intollerant of homo cores?



Totally...


I don't even want to think of the way homo cores 'interface' with each other :ouch: 
February 11, 2008 8:54:40 PM

LukeBird said:
Way, waaaaaay too early to tell...
You can't predict the future from the past!
Logically Phenom should have been pretty good (longer dev time over Core 2 quad, the A64 K8 architecture was pretty good) and look how it turned out, not bad, but certainly not phenomenal... (;))
AMD may really have a kick-ass 45nm Phenom and Intel's fabs may explode, you just can't work anything out.
Can't see either being on time either....


Given the loss in performance AMD experianced in going from 90nm SOI to 65nm SOI, and the known problems of SOI, why would anyone expect a performance improvement in going to 45nmSOI? Yes, the smaller node will require less power, but the limitations of SOI at decreasing nodes indicate AMD will see a further loss of clockspeed when they transition to 45nm. The real advantage of 45nm for AMD, if they do get there, and go there using SOI, and can make SOI work successfully at that node, is not in performance, but increased yields and lowering TDP.
a c 126 à CPUs
a b À AMD
February 12, 2008 2:39:52 AM

turpit said:
Given the loss in performance AMD experianced in going from 90nm SOI to 65nm SOI, and the known problems of SOI, why would anyone expect a performance improvement in going to 45nmSOI? Yes, the smaller node will require less power, but the limitations of SOI at decreasing nodes indicate AMD will see a further loss of clockspeed when they transition to 45nm. The real advantage of 45nm for AMD, if they do get there, and go there using SOI, and can make SOI work successfully at that node, is not in performance, but increased yields and lowering TDP.


But if they want to "compete" they will need to have high enough clock speeds which will not benefit them at all since at the same clock speeds Intel saw that 45nm SOI would produce much more leakage(power dissipation and heat) and less performance compared to a 65nm same clocked product. Hence their change to HK/MG.

I feel that if AMD allows themselves to try to go SOI @ 45nm then it will harm them more than help them. They wont be able to clock high enough and will lose performance thus puting them in the budget PC arena such as eMachines which we all know suck. They wont make enough money to cover the costs of transitioning to 45nm and will hurt their entire company.

So this should be a rather interesting year, no?
a b à CPUs
February 12, 2008 2:56:59 AM

Amiga500 said:
Totally...


I don't even want to think of the way homo cores 'interface' with each other :ouch: 

Last I checked they interfaced at least 2000 times per second.
February 12, 2008 4:19:18 AM

It sounds promising, I hope for their sake it's all its cracked up to be. If they are going for a complete redesign, chances are they will come up with something that is at the very least competitive with Intel. If they master thread fusing it could be a big leap forward in processor technology.
February 12, 2008 4:57:00 AM

yomamafor1 said:
If I remember correctly from the last AMD Analyst Day, they're only looking to build monolithic quad cores, then do MCM for 8 cores.

Quite a huge jump from monolithic 16 cores to monolithic 4 cores huh... :kaola: 



Oh yeah...now I remember where I saw that, and now I remember why I couldnt remember (read: didnt want to remember).....it was from the Inquirer :sarcastic: 
February 12, 2008 5:00:39 AM

jimmysmitty said:
But if they want to "compete" they will need to have high enough clock speeds which will not benefit them at all since at the same clock speeds Intel saw that 45nm SOI would produce much more leakage(power dissipation and heat) and less performance compared to a 65nm same clocked product. Hence their change to HK/MG.

I feel that if AMD allows themselves to try to go SOI @ 45nm then it will harm them more than help them. They wont be able to clock high enough and will lose performance thus puting them in the budget PC arena such as eMachines which we all know suck. They wont make enough money to cover the costs of transitioning to 45nm and will hurt their entire company.

So this should be a rather interesting year, no?


And there in layeth the problem of sticking to SOI. AMD can continue to try and compete with slower clockspeeds, if they get good yeilds at the lower node, but it will force them further down that road where they really cant compete...low margin, high volume "budget" processors.
February 12, 2008 5:05:07 AM

turpit said:
Given the loss in performance AMD experianced in going from 90nm SOI to 65nm SOI, and the known problems of SOI, why would anyone expect a performance improvement in going to 45nmSOI? Yes, the smaller node will require less power, but the limitations of SOI at decreasing nodes indicate AMD will see a further loss of clockspeed when they transition to 45nm. The real advantage of 45nm for AMD, if they do get there, and go there using SOI, and can make SOI work successfully at that node, is not in performance, but increased yields and lowering TDP.

Could you explain why you think having a layer of silicon oxide in the center of the wafer would cause loss of performance?
February 12, 2008 5:08:21 AM

With the "luck" AMD has been having lately, I wouldn't be too suprized if thier attempt at using hafnium resulted in them accidentaly creating Hf-178-m2.
Should be a lot of bang in one of those chips.
February 12, 2008 5:10:51 AM

endyen said:
Could you explain why you think having a layer of silicon oxide in the center of the wafer would cause loss of performance?



Again???? NO....go find the thread. Im not writing another book :kaola: 
February 12, 2008 8:14:21 AM

Amiga500 said:
I think the future should be based on hetrogeneous cores.

Have one deep pipeline core, that can be clocked up for single thread apps that would use it. Then shallower cores with better IPC for the parallel work. Heck, have an encoding core, have a gaming core, have a low power core for cool 'n' quiet, have a GP-GPU core for parallel calcs.


Multi core offers multi options - there is no need to compromise so much IMO.


isn't that similar to the Cell Processor??
One main cell w/ 8 tiny useless cells...
a b à CPUs
February 12, 2008 8:35:58 AM

Useless? Fold man, fold!
a c 126 à CPUs
a b À AMD
February 12, 2008 12:35:18 PM

endyen said:
Could you explain why you think having a layer of silicon oxide in the center of the wafer would cause loss of performance?



I'll handle this one turpit

Easy way to put it...

Its not silicon thats causing the performance loss per say but the fact that after 90nm it starts to reach its limits. Basically it starts to thin out and creates small holes that get larger at smaller processes, such as 65nm, causing more leakage thus you have to clock it lower to be able to release a decent non solar heater CPU. Also more leakage = less performance and lower clock speeds to battle leakage = less performance.

Think of it this way. SOI is cream cheese and the smaller it gets the thinner it gets. After a while of spreading it thiner and thiner it creates holes allowing the electrons to roam free.
February 13, 2008 1:20:58 AM

randomizer said:
Useless? Fold man, fold!



Done. Now what?
a b à CPUs
February 13, 2008 1:24:24 AM

Get some more work :lol: 
February 13, 2008 1:25:56 AM

randomizer said:
Get some more work :lol: 



:kaola: 
February 13, 2008 2:11:33 AM

turpit said:
And there in layeth the problem of sticking to SOI. AMD can continue to try and compete with slower clockspeeds, if they get good yeilds at the lower node, but it will force them further down that road where they really cant compete...low margin, high volume "budget" processors.

I thought AMD cpu's were already budget processors?
February 13, 2008 2:24:56 AM

yup
!