Bulldozer vs Nehalem

[Harrisson]

From what we know Bulldozer seems nice new gen, first after K8 or even K7. No idea if AMD will manage to deliver monster cpu as promissed, especially against Nehalem.

Whats interesting, Bulldozer and Nehalem, R700 and Larabee will go lego style. Nehalem will mimic AMD's Fusion, but no details yet if Bulldozer will have graphics on cpu, probably depending on market.

For cores count - Bulldozer will have 8-16, while Nehalem up to 8 cores but will have HT up to 16.

Performance wise new core should have incredible IPC... the less threads you work on at any given time. Since cores will dynamicly adjust as "single core" if you work with single thread, and so on.

Provided cores linear scalling is true as promised, even if Bulldozer is slower than Nehalem in totally multithreaded app, it should be faster in <4 threaded apps. Its a wild guess but seems plausible.

 

[Major_Spittle]

There are some real morons on this forum.

 

[randomizer]

Same as anywhere.

 

[rwayne]

I think the ramp from 65nm to 45nm will be much kinder and easier on AMD than that of the 90nm to 65nm. So I think Bulldozer is going to be an amazing processor. The only problem is that I think Nehalem will be a little better; mainly because AMD saw large performance gains whilst putting an inbuilt memory controller. So I think that intel, who already has great core architecture, is gonna see some great gains from the inbuilt memory controller. Of course, we know almost nothing about these two processors, so until then, I am just going to have to pray that AMD can get enough cash to help their R&D team so they can be a competitor again.

I agree with this statement:

That the "ramp from 65nm to 45nm will be much kinder and easier on AMD than that of 90nm to 65nm".

The reason I believe that is because Fudzilla.com is reporting that the R700 ATI GPU that is coming out in probably early 2nd quarter 2008 is going to be built at 45nm. So AMD pretty much has the technology already to do this. The current HD3850 and 3870 ATI cards (which are relased products) are already at 55nm. Yes those are GPUs but I am sure that AMD has had 45nm CPU models in their lab for quite some time now. They just have to tweak it.

 

[Kamrooz]

I agree with this statement:

That the "ramp from 65nm to 45nm will be much kinder and easier on AMD than that of 90nm to 65nm".

The reason I believe that is because Fudzilla.com is reporting that the R700 ATI GPU that is coming out in probably early 2nd quarter 2008 is going to be built at 45nm. So AMD pretty much has the technology already to do this. The current HD3850 and 3870 ATI cards (which are relased products) are already at 55nm. Yes those are GPUs but I am sure that AMD has had 45nm CPU models in their lab for quite some time now. They just have to tweak it.

I hope you realize AMD/ATI doesn't fab their own graphics chips. They order them from TSMC...Just like Nvidia...So they don't have that 45nm technology yet.

 

[yomamafor1]

I agree with this statement:

That the "ramp from 65nm to 45nm will be much kinder and easier on AMD than that of 90nm to 65nm".

The reason I believe that is because Fudzilla.com is reporting that the R700 ATI GPU that is coming out in probably early 2nd quarter 2008 is going to be built at 45nm. So AMD pretty much has the technology already to do this. The current HD3850 and 3870 ATI cards (which are relased products) are already at 55nm. Yes those are GPUs but I am sure that AMD has had 45nm CPU models in their lab for quite some time now. They just have to tweak it.

Actually, all ATi chips are built on TSMC's process, not AMD's in house process. If I'm not mistaken, TSMC will soon implement HK/MG with its 45nm process.

TBH, I expect AMD to encounter greater problem in 45nm than 65nm. AMD already has to thicken the gate about 20% to counter the gate leakage at 65nm, which affected AMD's ability to clock high. (3.2Ghz 90nm vs 2.6Ghz 65nm). At 45nm, the leakage would become so severe that AMD would have to thicken the gate again to counter it, which again hamper their ability to clock high.

Without HK/MG, I really don't see how AMD's 45nm will help with Barcelona's thermal. It will probably increase yield by a slight amount, but I don't expect any miracle.

 

[rwayne]

You obviously know more than me. That is impressive knowledge.

 

[monsterrocks]

Actually, all ATi chips are built on TSMC's process, not AMD's in house process. If I'm not mistaken, TSMC will soon implement HK/MG with its 45nm process.

TBH, I expect AMD to encounter greater problem in 45nm than 65nm. AMD already has to thicken the gate about 20% to counter the gate leakage at 65nm, which affected AMD's ability to clock high. (3.2Ghz 90nm vs 2.6Ghz 65nm). At 45nm, the leakage would become so severe that AMD would have to thicken the gate again to counter it, which again hamper their ability to clock high.

Without HK/MG, I really don't see how AMD's 45nm will help with Barcelona's thermal. It will probably increase yield by a slight amount, but I don't expect any miracle.

I think they will have High-k (or whatever) ready by then. You said earlier that you didn't think they would have that for their new gates until later, but if what you say is true, they don't have a choice. AMD has made some bad decisions, but if thickening their gates leads to lower clock speed; they will not be using the same old gate material, they know they can't afford any lower of clock speeds. If IBM is helping they research the high-k stuff, they will have IBM's R&D team on it too, which I imagine is much larger than AMD's; I could be wrong though...

What I mean is that they should have greater yields; which you pretty much can't disagree with (unless they have manufacturing problems) because the die size will be smaller, meaning more cores per wafer, which leads to fewer defect percentages. (a.k.a. if you had 100 cores on a wafer and 250 cores on a wafer, and both wafers had the same number of defects, you would have much higher yields with the 250 core wafer. I know those may not be realistic numbers, but they demonstrate what I am trying to point out.)

 

[jimmysmitty]

I don't think that AMD will be able to hit 45nm on their own. I doubt they would use TSNC's process as that would cost them money they don't have.

The only way they could hit 45nm is with IBMs help which is how they got their 64Bits(which wasn't the first 64bit as Intel had the Itanium but was set for x64 instead of x86-64 or x86 with 64bit instructions) up and running so well. But back then they had more R&D and weren't bleeding so bad.

What AMD has to understand is that SOI is extinct. Its a 40+ year old process and has hit the limits. Intel also has a new version of High-K to use with Nehalem. And the HT is not the same as the old HT. Its a newer version that improves over the previous version. According to Intel it scales much better than the previous version.

I don't think AMD will be able to release Bulldozer as planned without cash. They really need something to sell their processors and its not going to be Phenom. Even with a CFX setup your Phenom will bottleneck you. And without being able to OC without creating too much heat you are stuck at that bottleneck.

I don't think AMD is out just down and hurt really bad. They are missing their legs and need something to help them but they just can't get back to the upper area. I just really hated all of their talk. I don't mind if you can talk the talk and walk the walk but if all you have is talk then its just useless.

 

[yomamafor1]

I think they will have High-k (or whatever) ready by then. You said earlier that you didn't think they would have that for their new gates until later, but if what you say is true, they don't have a choice. AMD has made some bad decisions, but if thickening their gates leads to lower clock speed; they will not be using the same old gate material, they know they can't afford any lower of clock speeds. If IBM is helping they research the high-k stuff, they will have IBM's R&D team on it too, which I imagine is much larger than AMD's; I could be wrong though...

The problem is, AMD got their process technology directly from IBM. AMD doesn't develop its own process technology, so they're practically being led wherever IBM wants to go. If IBM doesn't see the need to transition to HK/MG, and/or they're still in the process to develop an usable technology, AMD has no choice but to wait.

IMO, AMD is more anxious than anyone to get their hands on HK/MG, and abandon SOI. SOI at 90nm is already showing its limit, and at 65nm it was doing more harm than good. At 45nm, implementing SOI would probably mean higher thermal than bulk.. AMD knows that, but they just can't throw away SOI. They simply don't have the financial and brain power to research their own HK/MG. So if IBM says HK/MG in 2H09, or even 1H10k, AMD has to say the same thing.

What I mean is that they should have greater yields; which you pretty much can't disagree with (unless they have manufacturing problems) because the die size will be smaller, meaning more cores per wafer, which leads to fewer defect percentages. (a.k.a. if you had 100 cores on a wafer and 250 cores on a wafer, and both wafers had the same number of defects, you would have much higher yields with the 250 core wafer. I know those may not be realistic numbers, but they demonstrate what I am trying to point out.)

They should have higher yield. But what does yield mean? AMD can yield more functional chip per wafer, but how many of them will go off its designed TDP? Given AMD won't implement HK/MG until the 2nd generation of 45nm, I would imagine the first generation 45nm chips to be worse than their 65nm counterparts. From what we've seen from 90nm=>65nm, I wouldn't be surprised if AMD can only clock to 2.4Ghz (for Brisbane) with their first batch of 45nm chips.

All of these are just pure speculations from very scant data we've known, and it can change as time goes on. However, from what we've seen, although it appears IBM's 45nm process is a little better than their 65nm process, I don't know how successful AMD can implement it.

 

[wolverinero79]

This thread is funny. Well, I can't resist but throwing my 2 cents in.

1.) AMD's transition from 90nm to 65nm was not really that smooth. Going from 65nm to 45nm is....*drumroll*...harder! To think they'd do better is really asking for a miracle. And miracles for companies Beleaguered by debt are few and far between.

2.) Nehalem comes out in 2008...most likely middle of 2008. They're basically done - the chip has taped out and booted. It's only a matter of time before the process yields well and all the bugs are worked out. Its performance is basically completely unknown. Why? It's really a brand spanking new architecture. Yes, the Integrated Memory Controller is all snazzy, but aren't we also wondering what other aces Intel has hiding?

3.) Bulldozer will NOT be up against Nehalem. It will be up against Westmere (aka Nehalem C - Nehalem at 32nm). Once again, AMD will find itself one process generation behind - meaning no matter what performance comes in at, Intel's costs will be significantly lower. But then again, Intel has to actually pay for their factories and research, etc. Oh, and if Bulldozer is delayed (why would that EVER happen...I mean, AMD is the best, right?), it competes against Sandy Bridge. Hmm, so yet another new arch, but at the 32nm level. Ouch.

4.) Bulldozer and Nehalem are new architectures. Ground up. Either one could be awesome or terrible. We have no idea. And we know for a fact (yay K10!) that AMD's marketing fluff is not to be trusted. A lot of Intel's marketing is also suspect at times. So we don't know! :-D

5.) Fusion vs Larrabee will be interesting to say the least. I have no idea what Intel's doing, but I don't know if Larrabee is supposed to be an actual 3rd choice for discrete graphics (aka a potential tie for loser, as Nvidia's pretty way out ahead), or if Larrabee is designed to be a GPU that is one or more cores in the die for Westmere. We saw that the Core 2 transition to 45nm (Penryn) had awesome extra goodies...packaging Larrabee with CPUs would be one heck of a goody for Westmere.

6.) IBM does not have a demonstrated HK/MG product. IBM's announcement reaked of "a smaller (yes, go check it out - IBM has more than quadruple the employees of Intel) company can NOT have better advanced technology than us". I.e., corporate posturing. Ah, politics. If IBM had the tech, AMD could get it easily and cheaply (if not free). IBM can't buy AMD since then the x86 license goes away, so they enjoy propping up the (currently) struggling company to help it survive.

7.) 64-bit hasn't taken off. There's no indication it will take off anytime soon. Tell me a need you have for 64-bit in desktop or mobile that at least 20% of the population would benefit from. Ok, try 10%.

8.) AMD needs money. Their current liabilities (which are mostly debt that they regularly pay interest on) are almost equal to their assets. That's incredibly bad. At the even level, that means that if AMD could sell all assets, they can just barely pay all their debt. That's a potential Chapter 11 situation.

9.) Faster desktop/mobile chips are not where the future money is. Especially in North America or Europe. Either companies will focus more on small cheap parts EVERYWHERE ("why does my freezer have 'Intel Inside' on it?"), or on cheap last year model's parts in poor, untapped countries. I'm running a P4 Prescott...why? Because it does EVERYTHING I need it for from playing great games (Supreme Commander is just fine) to video compression. So unless it'll blow out soon (and really, I've never had a chip that's worn out before becoming completely and utterly obsolete), there's no reason for me to upgrade.

Now I'm not painting a picture of doom and gloom because I hate AMD. Far from it. But they're in a tight spot right now and it looks to just be getting tighter for the future. That's just reality. We can try to prop them up, cheer them on, but none of that's going to ultimately make them deliver on their promises and make new products we actually want to buy. I'm disappointed in the direction they've gone. I don't know if it's just Intel being awesome, or if AMD just really doesn't care anymore, but we have no processor race/competition at this point in time. I'm eager to see what happens in the future, but considering how much more cheaply technology will sell for, and yet how much more complex it's becoming to produce, it's going to become more like a commodity and less like an actual race between two equal partners.

End of wall of text. Have a great day!

 

[pausert20]

I don't think that AMD will be able to hit 45nm on their own. I doubt they would use TSNC's process as that would cost them money they don't have.

The only way they could hit 45nm is with IBMs help which is how they got their 64Bits(which wasn't the first 64bit as Intel had the Itanium but was set for x64 instead of x86-64 or x86 with 64bit instructions) up and running so well. But back then they had more R&D and weren't bleeding so bad.

What AMD has to understand is that SOI is extinct. Its a 40+ year old process and has hit the limits. Intel also has a new version of High-K to use with Nehalem. And the HT is not the same as the old HT. Its a newer version that improves over the previous version. According to Intel it scales much better than the previous version.

I don't think AMD will be able to release Bulldozer as planned without cash. They really need something to sell their processors and its not going to be Phenom. Even with a CFX setup your Phenom will bottleneck you. And without being able to OC without creating too much heat you are stuck at that bottleneck.

I don't think AMD is out just down and hurt really bad. They are missing their legs and need something to help them but they just can't get back to the upper area. I just really hated all of their talk. I don't mind if you can talk the talk and walk the walk but if all you have is talk then its just useless.

JimmySmitty, You are confusing 2 different things when you are talking about SOI. It is not a 40+ old technology. It actually is a very new technology. What you are talking about 40+ years is the use of Silicon Oxide SO2 as the insulator between the gate and PN channel of the basic gate.

SOI is actually part of the wafer. It is an embedded layer of silicon Oxide covered by silicon. The silicon Oxide is the insulator. JumpingJack discussed this in many threads earlier this year before he left in disgust dealing with BM. There are limitations to the use of SOI that get greater and greater the smaller the process technology that is used.

The use of these SOI wafers are an added cost to AMD. They cost more than standard 100 or 110 Silicon wafers that Intel uses. Intel specifically looked at SOI and did not see the benifit of the reduce bulk leakage overcame its other limitations. I believe they are being proved correct.

This is why I believe that AMD is going to have even more problems moving from 65nm to 45nm than they did from 90nm to 65nm.

 

[Harrisson]

@wolverinero79

1) How do you know transition to 45nm will be harder? One presumption it may be because of SOI, other than that AMD is going fast with new tech. process, for graphics they moved from 80nm > 65nm> 55nm in one year, and in six months - 45nm.

3) Did AMD mentioned they wont be releasing Bulldozer in 2008? Link please.

5) Larrabee is planed to be discreet graphics (including high-speed), we'll see if Intel manages it, at least they have far more resources than AMD/nVidia.

7) 64bit not so important atm for average consumers, but no serious gaming rig will have less than 4GB RAM within a year. Already new games ask for more than 2GB for best performance.

8) Situation is bad but worst period is over - graphics and chipsets are doing nicely, and Barce is released. To ramp up production is way wasier than to launch new core.

 

[yomamafor1]

@wolverinero79

1) How do you know transition to 45nm will be harder? One presumption it may be because of SOI, other than that AMD is going fast with new tech. process, for graphics they moved from 80nm > 65nm> 55nm in one year, and in six months - 45nm.

The transition to 45nm will be a lot more difficult for AMD is because of 45nm's physical limitation. At 65nm, AMD is showing difficulties in maintaining their clocks compared to their 90nm counterparts. When AMD reaches for 45nm, without HK/MG, the power leakage will be more severe than 65nm. The gate will be so thin, that electrons will literally "jump" through the shielding. AMD already used 20% thicker gate to counter this problem, and it is likely that AMD will have to again increase the gate thickness.

I don't have enough knowledge to answer this question in a more technical manner. I suggest you to read a review on HK/MG. It explains in detail why 45nm is a lot more difficult to implement than 65nm.

http://www.eetimes.com/news/design/showArticle.jhtml?articleID=202806020

3) Did AMD mentioned they wont be releasing Bulldozer in 2008? Link please.

http://www.behardware.com/news/9053/amd-sse5-the-bulldozer.html
http://blogs.zdnet.com/BTL/?p=5766
http://www.bit-tech.net/news/2007/07/28/amd_goes_modular/1

Also, although it is still to early to tell, but it seems like Bulldozer's main feature will be higher clocks (with same IPC), as well as modular capability.

So BM, no 1Ghz Bulldozer trumps any Nehalem this time.

5) Larrabee is planed to be discreet graphics (including high-speed), we'll see if Intel manages it, at least they have far more resources than AMD/nVidia.

It is also planed to be integrated into Nehalem core via modularity.

7) 64bit not so important atm for average consumers, but no serious gaming rig will have less than 4GB RAM within a year. Already new games ask for more than 2GB for best performance.

Agreed. I'm personally using 64bit and 4Gb, although I don't see any improvement in performance.

8) Situation is bad but worst period is over - graphics and chipsets are doing nicely, and Barce is released. To ramp up production is way wasier than to launch new core.

Actually, AMD's test has only begun. Barcelona is falling short of clockspeed, and Phenom is falling short of AMD's expectation. Penryn will further decrease AMD's performance/watt, while Nehalem will strip AMD of all of its architectural advantage (IPC/HT). This means AMD's core business will be invaded by high performing, modular, 45nm, native quad/hexa/octa cores. 2008-2009 will be AMD's bloodiest year.

 

[monsterrocks]

This thread is funny. Well, I can't resist but throwing my 2 cents in.

1.) AMD's transition from 90nm to 65nm was not really that smooth. Going from 65nm to 45nm is....*drumroll*...harder! To think they'd do better is really asking for a miracle. And miracles for companies Beleaguered by debt are few and far between.

2.) Nehalem comes out in 2008...most likely middle of 2008. They're basically done - the chip has taped out and booted. It's only a matter of time before the process yields well and all the bugs are worked out. Its performance is basically completely unknown. Why? It's really a brand spanking new architecture. Yes, the Integrated Memory Controller is all snazzy, but aren't we also wondering what other aces Intel has hiding?

3.) Bulldozer will NOT be up against Nehalem. It will be up against Westmere (aka Nehalem C - Nehalem at 32nm). Once again, AMD will find itself one process generation behind - meaning no matter what performance comes in at, Intel's costs will be significantly lower. But then again, Intel has to actually pay for their factories and research, etc. Oh, and if Bulldozer is delayed (why would that EVER happen...I mean, AMD is the best, right?), it competes against Sandy Bridge. Hmm, so yet another new arch, but at the 32nm level. Ouch.

4.) Bulldozer and Nehalem are new architectures. Ground up. Either one could be awesome or terrible. We have no idea. And we know for a fact (yay K10!) that AMD's marketing fluff is not to be trusted. A lot of Intel's marketing is also suspect at times. So we don't know! :-D

5.) Fusion vs Larrabee will be interesting to say the least. I have no idea what Intel's doing, but I don't know if Larrabee is supposed to be an actual 3rd choice for discrete graphics (aka a potential tie for loser, as Nvidia's pretty way out ahead), or if Larrabee is designed to be a GPU that is one or more cores in the die for Westmere. We saw that the Core 2 transition to 45nm (Penryn) had awesome extra goodies...packaging Larrabee with CPUs would be one heck of a goody for Westmere.

6.) IBM does not have a demonstrated HK/MG product. IBM's announcement reaked of "a smaller (yes, go check it out - IBM has more than quadruple the employees of Intel) company can NOT have better advanced technology than us". I.e., corporate posturing. Ah, politics. If IBM had the tech, AMD could get it easily and cheaply (if not free). IBM can't buy AMD since then the x86 license goes away, so they enjoy propping up the (currently) struggling company to help it survive.

7.) 64-bit hasn't taken off. There's no indication it will take off anytime soon. Tell me a need you have for 64-bit in desktop or mobile that at least 20% of the population would benefit from. Ok, try 10%.

8.) AMD needs money. Their current liabilities (which are mostly debt that they regularly pay interest on) are almost equal to their assets. That's incredibly bad. At the even level, that means that if AMD could sell all assets, they can just barely pay all their debt. That's a potential Chapter 11 situation.

9.) Faster desktop/mobile chips are not where the future money is. Especially in North America or Europe. Either companies will focus more on small cheap parts EVERYWHERE ("why does my freezer have 'Intel Inside' on it?"), or on cheap last year model's parts in poor, untapped countries. I'm running a P4 Prescott...why? Because it does EVERYTHING I need it for from playing great games (Supreme Commander is just fine) to video compression. So unless it'll blow out soon (and really, I've never had a chip that's worn out before becoming completely and utterly obsolete), there's no reason for me to upgrade.

Now I'm not painting a picture of doom and gloom because I hate AMD. Far from it. But they're in a tight spot right now and it looks to just be getting tighter for the future. That's just reality. We can try to prop them up, cheer them on, but none of that's going to ultimately make them deliver on their promises and make new products we actually want to buy. I'm disappointed in the direction they've gone. I don't know if it's just Intel being awesome, or if AMD just really doesn't care anymore, but we have no processor race/competition at this point in time. I'm eager to see what happens in the future, but considering how much more cheaply technology will sell for, and yet how much more complex it's becoming to produce, it's going to become more like a commodity and less like an actual race between two equal partners.

End of wall of text. Have a great day!

You don't see 64-bit being mainstream anytime soon, eh? Are you aware that windows Vienna is coming out 2009-2010 and that Vista was the last (yes, you read correctly) OS to support 32-bit computing? It's true, look it up. Secondly, you talk like a company cannot catch up once they fall behind. Remember intel in the pentium 4 days, and their first dual cores? They sucked compared to AMD's offerings. But look what they did with core 2. Yes they had (and still have) a better funded R&D team, but you speak like if once behind, always behind. As to mobile computing not being where money is, that is one of the fastest growing computing market today. As business grows for corprate companies, they need computers and the like to increase employees productivity. Have you seen the number of company issued laptops? It is a fast growing market. It is also AMD's second largest market (Server is their first, then comes desktop at 3rd). It is only a matter of time before intel's process yields for 45nm? What if it doesn't yield on time? Yes, they have had yields, but from what you say, not enough to put in production. You can't just say when you should have yields by, it doesn't work like that. You can guess when you will have them, but getting yields on wafers is not as easy as it sounds. One of the reasons AMD has had so many troubles.

 

[nruo]

Intel's power measurement is taken when we turn on every transistor on the chip and measure how much heat you would ever have to remove. Positively worst case scenario.

Granted, no one will ever use 100% of a chips transistors at any one time. Hell, if you use 50% at once, you're doin pretty good. But, better safe than sorry. If you actually measure the power draw of a 125W rated 45nm processor, you're likely to discover that you're only using .25 to .5 of that at full load.

this is not fully accurate. it is not even close to 100% of the transitors.
this is defined as the worst case possible in "normal "system and aplications.
it is not the worst possible theoretic scenario, but is is defined such that you will never need to remove more heat than that

 

[exit2dos]

You don't see 64-bit being mainstream anytime soon, eh? Are you aware that windows Vienna is coming out 2009-2010 and that Vista was the last (yes, you read correctly) OS to support 32-bit computing? It's true, look it up.

Vienna will have 32-bit support.
http://www.xbitlabs.com/news/other/display/20070721070729.html
Vista was initially supposed to be the 64-bit transition, it's rollout was "botched", and it's looking like Vienna may have to play that role.

 

[turpit]

Vienna will have 32-bit support.
http://www.xbitlabs.com/news/other/display/20070721070729.html
Vista was initially supposed to be the 64-bit transition, it's rollout was "botched", and it's looking like Vienna may have to play that role.

M$, Botch an OS? Never.....sheesh...some people

 

[monsterrocks]

M$, Botch an OS? Never.....sheesh...some people

:lol:

 

[jaguarskx]

Phenom has lost to C2D in this round but AMD's next gen Bulldozer is waiting in the wings. Upon it's release (2009-10), it'll probably go up against Nehalem.

Actually, Phenom will probably go head to head with Intel's third-generation Core microarchitecture codenamed Gesher which is expected in 2010 and will be 32nm instead of 45nm.

 

[monsterrocks]

Vienna will have 32-bit support.
http://www.xbitlabs.com/news/other/display/20070721070729.html
Vista was initially supposed to be the 64-bit transition, it's rollout was "botched", and it's looking like Vienna may have to play that role.

That article is dated 7/21/07. The stuff I read about Vienna was in PC magazine (I am pretty sure) a month ago (I am sure about the month ago part). It said Vienna would be 64-bit only. If vista was originally that transition, it only follows that since that failed they would try on the next OS. Interesting article, but since what I read is newer; I'll believe the newer stuff instead. I probably get accused of chronological snobbery for that though....

 

[quantumsheep]

Bulldozer SHOULD beat Nehalem, seeing as it's scheduled for 2 years after Nehalem is released.

As far as i can remember Intel is forecasting 32 core CPUs, on an architeture that succeeds Nehalem, by 2010 on a 32nm process. I'll try and find their VERY agressive roadmap as it makes an interesting read, saying that they'll release a brand new architecture every two years! If Intel keep going at their current development rate (actually meeting their deadlines) AMD will be dead by 2012. Well, that's my predicition.

 

[Grimmy]

I think this BullDozer will trash anything in its path:

😳. o O (okay.. its not a CPU)

 

[yomamafor1]

I think this BullDozer will trash anything in its path:

http://www.dkimages.com/discover/previews/797/45068195.JPG

😳. o O (okay.. its not a CPU)

Yes, but can it trash Nehalem, which is a river in WA?

Behold.. the mighty battle, between a bulldozer, and a river

 

[monsterrocks]

Yes, but can it trash Nehalem, which is a river in WA?

Behold.. the mighty battle, between a bulldozer, and a river

The bulldozer could dam the river...bulldozer wins!