• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Rumour: Bulldozer 50% Faster than Core i7 and Phenom II.

Page 46 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.
You keep telling us not to get to excited about the release date news. Reason?
"Well, the customers would stop buying current offering if the Bullldozer proved to good."
???
Excuse me?
The customers that want/need performance buy Intel (already).

AMD is living from low budget system builds; and those customers will expand their budgets because AMD is going to put out a performance 8 core chip?
No joke; auu! I mean! So if a person is thinking about AMD often enough money from heavens rains down on him/her? :whiste: :whiste: :whiste:

Not sure if it was this forum (there are so many threads...) but people were talking like it was the launch or that we would be releasing the pricing and benchmarks. I think rumors get spun up. The reason I say don't get too excited is that things tend to get spun up to a fever pitch and then we get the "crap, they lied to us...." replies that start showing up.

I feel somewhat letdown by AMD's reluctance to share any info at all on BD. I'm a longterm AMD fan but they really need to work on their execution and marketing. The only exception to this is the former ATI who somehow is able to produce chip after chip on time and on budget that is able to perform very well and be manufactured for a decent cost. Can ATI please teach AMD how to make decently performing chips that are released on time and dont require massive transistor budgets to achieve comparatively lackluster performance?

I'm just getting a bit jaded I suppose, I'm a long time AMD fan but its been years and years since they made a CPU that competed well with Intel's products, in terms of the raw silicon. They just about have to sell them at a loss to sell anything. BD better blow the socks off everyone or else I fear we will return to a single x86 manufacturer.

I actually think we have released everything but pricing, SKUs, benchmarks and clock speeds. But those are all things that we release at launch (as does our competitor.)

Perhaps you should go back and read the bulldozer blogs, especially the 20 questions blogs (and especially all of the answers that I gave to follow-on questions.) I just don't get how people say we are not sharing enough. What is it that you would like to see?
 
I actually think we have released everything but pricing, SKUs, benchmarks and clock speeds. But those are all things that we release at launch (as does our competitor.)

Perhaps you should go back and read the bulldozer blogs, especially the 20 questions blogs (and especially all of the answers that I gave to follow-on questions.) I just don't get how people say we are not sharing enough. What is it that you would like to see?

Benchmarks? The earliest reference that I can see to Bulldozer is from 2008 - 3 years ago! Perhaps that is the source of the frustration - its not what information you have or have not released, its that the information has been spread over a 3 year period, making it look anemic compared to Intel, who announces a new product and then ships it less than 12 months later. We like knowing AMD's plans, but I guess I struggle to understand why Bulldozer has taken this long. Again, compare yourselves to ATI. I'm not sure exactly how much difference there is in complexity between GPUs and CPUs, but since modern GPUs run in at over 2 billion transistors, I'd be surprised if they are somehow vastly simpler. Yet despite that, ever since the HD 2900XT, AMD have had no problems releasing products on time.

You arent performing AMD! I'm sorry to say this, I championed AMD back when Intel was using its dirty tricks and everyone thought AMD wasnt doing well because the Core 2 Duo was just better. Now I realize perhaps there was a shred of truth to that.
 
Hi all.
Just registered on this forum so I could Answer JF-AMD question.
Please forgive my not so good english as it is not my native language, french is.

I actually think we have released everything but pricing, SKUs, benchmarks and clock speeds. But those are all things that we release at launch (as does our competitor.)
Wrong : we had SB info as soon as 8/27/2010, (that is 4 months before launch) on this very site
http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Granted exact pricing was not disclosed, but this is about the only thing was not released : my mistake
EDIT : re-read the article : even pricing WAS included (see page 4)

In short you just pointed almost everything needed to choose beetween SB and BD. And I suspect this is part of your marketing plan, so as to delay as long as possible this choice.
But, who ever said business had to be fair-play ????
 
Last edited:
I've been hearing about Ivy Bridge for 3 yrs now, and Sandy even longer before that, and Haswell for nearly as long by now.

I don't get this sense of entitlement that some people seem to have in regards to AMD, Bulldozer, and information from one about the other.

Some of you guys are acting like spurned or jilted lovers.

Frankly it is bizarre.
 
Dont't know what you expected. But if AMD shows Bulldozer at CeBit it would be clear that this would be a showing behind closed doors for limited audience. According to a rumor this will not be happen, but we actually do not know.
 
I've been hearing about Ivy Bridge for 3 yrs now, and Sandy even longer before that, and Haswell for nearly as long by now.

I don't get this sense of entitlement that some people seem to have in regards to AMD, Bulldozer, and information from one about the other.

Some of you guys are acting like spurned or jilted lovers.

Frankly it is bizarre.
100% agree

I can understand the frustrated excitement of some enthusiasts, but the bitterness of others is alien to me.
 
Intel announces the names of their new cores in advance, but not much else. We really didnt know much about the differences until recently, and in the meantime, Intel's tick tock steadily produces good architecture after good architecture, on time.

I was a fan of the underdog for many years, in part because quite often it had a massive performance advantage over Intel while simultaneously being cheaper. Or at least there was performance parity and AMD was cheaper. These days, except for the new Bobcat chips, AMD's chips are all late and all lacking in performance, and hence they are sold cheaply. Its one thing to lack performance - Intel after all has far more of an R&D budget - but massively delayed projects look like poor project management and planning. Pretty much like Duke Nukem Forever.

I think AMD has lost a lot of fans over the last 4-5 years, ever since Intel released the Core 2 Duo and turned itself around. For whatever reason, its lost my sympathy and now I'm irritated with it for constantly f*cking stuff up and always coming late to the party with power hungry chips that dont perform as well as the competition. I have only so much goodwill for the underdog and its nearly all used up.
 
I read that some guys on SA looked at the die shots, and by useing the 30something mm^2 statement AMD made, guessed the 8core bulldozer to be around 294mm^2.

So the 8core will be rather big compaired to say Intels Sandy bridge thats only 216mm^2 or something.


AMD Thuban's Die Size~ 346 mm², which is for the phenom II x6 core cpu.

Doing a 346mm^2 6core --> bulldozer's ~294mm^2 8core, doesnt sound so bad, its still huge when compaired to the 216mm^2 sandy bridges, So Im hopeing they deliver some nice performance.
I totally doubt this 294/300/3xx die sizes which are around.
That is for just one simple reason: I do not think that AMD needs 50% of the die size for uncore. That would be just a new record in wasting die size.

All those high estimations are based on a die shot that was obfuscated by AMD using Photoshop or that sort. What some did was a try to "Photoshop" it back and then start pixel measurements from numbers they know. It is clear that this only works if the "Photoshop it back" was done flawless.

I on the other hand make estimations based on engineering know how, means with assumption that the uncore area does not explode with Bulldozer.

Then you get ~200 mm². If you assume some big additional uncore compared to Deneb you might end up with 240 mm² but I really do not believe we get 300 mm². If that would be really true it would be a real failure of AMD engineers.
 
Hi all.
Just registered on this forum so I could Answer JF-AMD question.
Please forgive my not so good english as it is not my native language, french is.

Wrong : we had SB info as soon as 8/27/2010, (that is 4 months before launch) on this very site
http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Granted exact pricing was not disclosed, but this is about the only thing was not released : my mistake
EDIT : re-read the article : even pricing WAS included (see page 4)

In short you just pointed almost everything needed to choose beetween SB and BD. And I suspect this is part of your marketing plan, so as to delay as long as possible this choice.
But, who ever said business had to be fair-play ????

Anandtech.com is not intel.com. Intel did not release that data.

Now, if you want to argue about leaks, that is another issue, but we do not leak data, it is not how we do business.
 
I totally doubt this 294/300/3xx die sizes which are around.
That is for just one simple reason: I do not think that AMD needs 50% of the die size for uncore. That would be just a new record in wasting die size.

All those high estimations are based on a die shot that was obfuscated by AMD using Photoshop or that sort. What some did was a try to "Photoshop" it back and then start pixel measurements from numbers they know. It is clear that this only works if the "Photoshop it back" was done flawless.

I on the other hand make estimations based on engineering know how, means with assumption that the uncore area does not explode with Bulldozer.

Then you get ~200 mm². If you assume some big additional uncore compared to Deneb you might end up with 240 mm² but I really do not believe we get 300 mm². If that would be really true it would be a real failure of AMD engineers.

I was under the impression that AMD has released both the module die shot as well as the actual Orochi die shot between ISSCC & Cebit.

We know the die aread of a module, simple matter of determining the Orochi die-size at that point, isn't it?
 
BD 4 Module 8 Core will probable compete against Intel 6-8 core SB-E
No chance for this one. If Intel brings a 8 core / 16 thread Sandy Bridge than it will be the top performer and no 4 module / 8 thread Bulldozer will compete with half of the threads. That could only work if Sandy Bridge is clocked very low (for TDP reasons).

However when is that Sandy Bridge part coming? At the time it might come then it is likly that AMD will ready a 6 module / 12 core which should be able to surpass.

Btw. I heard that for the high speed design the number of pipeline stages was increased to 15 for Bulldozer. Can anyone confirm that? That would be really good for a high speed design compared with e.g. the 31 stages of Prescott. Afaik Deneb has 12 and Sandy Bridge 14.
 
I was under the impression that AMD has released both the module die shot as well as the actual Orochi die shot between ISSCC & Cebit.

We know the die aread of a module, simple matter of determining the Orochi die-size at that point, isn't it?
Afaik AMD released no die shot at all, or do you have any link?

All these 300 mm² calculations have been undertaken with a obfuscated die shot taken from a AMD presentation slide and that was long before ISSCC. At ISSCC there was given exact module size numbers of 30.9 mm² (and L2 size afaik) which have been only estimated before.
 
Afaik AMD released no die shot at all, or do you have any link?

All these 300 mm² calculations have been undertaken with a obfuscated die shot taken from a AMD presentation slide and that was long before ISSCC. At ISSCC there was given exact module size numbers of 30.9 mm² (and L2 size afaik) which have been only estimated before.

Yeah they showed it at ISSCC this year:
5.jpg

Source: http://pc.watch.impress.co.jp/docs/column/kaigai/20110301_430044.html
 
Intel announces the names of their new cores in advance, but not much else. We really didnt know much about the differences until recently, and in the meantime, Intel's tick tock steadily produces good architecture after good architecture, on time.

I was a fan of the underdog for many years, in part because quite often it had a massive performance advantage over Intel while simultaneously being cheaper. Or at least there was performance parity and AMD was cheaper. These days, except for the new Bobcat chips, AMD's chips are all late and all lacking in performance, and hence they are sold cheaply. Its one thing to lack performance - Intel after all has far more of an R&D budget - but massively delayed projects look like poor project management and planning. Pretty much like Duke Nukem Forever.

I think AMD has lost a lot of fans over the last 4-5 years, ever since Intel released the Core 2 Duo and turned itself around. For whatever reason, its lost my sympathy and now I'm irritated with it for constantly f*cking stuff up and always coming late to the party with power hungry chips that dont perform as well as the competition. I have only so much goodwill for the underdog and its nearly all used up.


You're dead wrong about this, and have one foot in the grave about everything else. From a marketing perspective though, this doesn't sound out of place from the kind of sentiment intel would like to throw around. intel probably realizes they need to go all in for FUD because they see the writing on the wall for what Llano brings to the table, as well as the potential for Bulldozer. These are the kind of posts I was expecting would start to show up. Of course, there are also those pathetic intel investors (and TROLLS; read yahoo AMD investor boards) that are now playing on the crisis in the Middle East as a reason to kill investor confidence since they are running out of FUD.
 
You're dead wrong about this, and have one foot in the grave about everything else. From a marketing perspective though, this doesn't sound out of place from the kind of sentiment intel would like to throw around. intel probably realizes they need to go all in for FUD because they see the writing on the wall for what Llano brings to the table, as well as the potential for Bulldozer. These are the kind of posts I was expecting would start to show up. Of course, there are also those pathetic intel investors (and TROLLS; read yahoo AMD investor boards) that are now playing on the crisis in the Middle East as a reason to kill investor confidence since they are running out of FUD.

I'm sorry, how am I dead wrong? Years ago, there was no choice at all. You either bought an Athlon 64, or you bought a Pentium 4 Prescott and a fridge to put it in. The A64 ran rings around the P4 in every way, and so among gaming enthusiasts, it was the only chip that mattered.

Along came the C2D and changed that. It was more than competitive, clock for clock it was faster than the A64, and as I remember pricing wasnt too bad. Yes, some people will still buy AMD out of brand loyalty, but now at least Intel offers options that you wouldnt be embarrassed or stupid to buy. Intel hasnt made a bad processor in recent years.

AMD hasnt made any truly bad processors, probably except for the Phenom I which I think was absolutely terrible. It couldnt clock high, drew lots of power and ran hot. The Phenom II still draws lots of power, but at least it performs decently. Still, compared to Sandy Bridge, it just doesnt perform well.

Perhaps its the Phenom I and the Bulldozer delays that have jaded me so much. Phenom I was worse than the HD 2900 XT in terms of crapness. Bulldozer is running incredibly late, and AMD STILL wont let us know what its supposed to perform like, probably less than a quarter before its release. Or what its power usage is. That irritates me, and doesnt instill much confidence in me.
 
I'm sorry, how am I dead wrong? Years ago, there was no choice at all. You either bought an Athlon 64, or you bought a Pentium 4 Prescott and a fridge to put it in. The A64 ran rings around the P4 in every way, and so among gaming enthusiasts, it was the only chip that mattered.

Along came the C2D and changed that. It was more than competitive, clock for clock it was faster than the A64, and as I remember pricing wasnt too bad. Yes, some people will still buy AMD out of brand loyalty, but now at least Intel offers options that you wouldnt be embarrassed or stupid to buy. Intel hasnt made a bad processor in recent years.

AMD hasnt made any truly bad processors, probably except for the Phenom I which I think was absolutely terrible. It couldnt clock high, drew lots of power and ran hot. The Phenom II still draws lots of power, but at least it performs decently. Still, compared to Sandy Bridge, it just doesnt perform well.

Perhaps its the Phenom I and the Bulldozer delays that have jaded me so much. Phenom I was worse than the HD 2900 XT in terms of crapness. Bulldozer is running incredibly late, and AMD STILL wont let us know what its supposed to perform like, probably less than a quarter before its release. Or what its power usage is. That irritates me, and doesnt instill much confidence in me.

The P4 arch was Intel's only mistake, but a lot of posters are to young to remember what went before.

Phenom was a return to normal the way I see it.
 
No chance for this one. If Intel brings a 8 core / 16 thread Sandy Bridge than it will be the top performer and no 4 module / 8 thread Bulldozer will compete with half of the threads. That could only work if Sandy Bridge is clocked very low (for TDP reasons).
Or, if AMD were going to come out with a CPU that could run 16 threads, like Interlagos.

Lonbjerg said:
The P4 arch was Intel's only mistake,
Please tell me that was a typo.
 
I'm sorry, how am I dead wrong? Years ago, there was no choice at all. You either bought an Athlon 64, or you bought a Pentium 4 Prescott and a fridge to put it in. The A64 ran rings around the P4 in every way, and so among gaming enthusiasts, it was the only chip that mattered.

Along came the C2D and changed that. It was more than competitive, clock for clock it was faster than the A64, and as I remember pricing wasnt too bad. Yes, some people will still buy AMD out of brand loyalty, but now at least Intel offers options that you wouldnt be embarrassed or stupid to buy. Intel hasnt made a bad processor in recent years.

AMD hasnt made any truly bad processors, probably except for the Phenom I which I think was absolutely terrible. It couldnt clock high, drew lots of power and ran hot. The Phenom II still draws lots of power, but at least it performs decently. Still, compared to Sandy Bridge, it just doesnt perform well.

Perhaps its the Phenom I and the Bulldozer delays that have jaded me so much. Phenom I was worse than the HD 2900 XT in terms of crapness. Bulldozer is running incredibly late, and AMD STILL wont let us know what its supposed to perform like, probably less than a quarter before its release. Or what its power usage is. That irritates me, and doesnt instill much confidence in me.

Should have said IMO. I think you're dead wrong because I don't think AMD HAS lost a lot of support. You may not see a lot of supporters posting, because if you are old enough (or young enough if you are too old) then you might remember the complete ridiculing dished out by intel supporters and marketing. I think you'll see a quite different landscape of support when AMD's vision of the future, which they've been preparing for 5 years, really surface in a couple months time. Don't be surprised by the support that will swell in AMD's favor, no matter how much FUD is spread between now and then.

Also, I think you'll find that many supporters of AMD products aren't merely supporting them for loyalty, fanboi, etc reasons like you suggest, but out of principle and the desire for a healthy marketplace that has been held hostage by intel for 20 years. Obviously, anyone can support whoever they want and be on whichever side of the fence they choose. It's too bad the vast majority of consumers are uninformed on the practices of the x86 market over the last 2 decades of rule by intel. You'd think that the informed technically savvy crowd would want better though.
 
The P4 arch was Intel's only mistake, but a lot of posters are to young to remember what went before.

Phenom was a return to normal the way I see it.

The AMD 486 had both a higher clock speed and better integer performance than the Intel 486.

The AMD K6 processor had better integer performance than the Intel Pentium Pro.

Intel obviously had their floating point bug with the original Pentium release, which forced me to be without a computer for a while.

Just because other clones like Cyrix had crappy quality, doesn't really have much to do eith AMD. They were always the best of the clones back then, and would even be faster than the Intel counterpart on occasion.

Just as you feel others are taking a selective look at history, by avoiding obvious Intel wins in the past, I feel you are doing the same by forgetting AMD wins in the past, even before the Athlon.
 
Should have said IMO. I think you're dead wrong because I don't think AMD HAS lost a lot of support. You may not see a lot of supporters posting, because if you are old enough (or young enough if you are too old) then you might remember the complete ridiculing dished out by intel supporters and marketing. I think you'll see a quite different landscape of support when AMD's vision of the future, which they've been preparing for 5 years, really surface in a couple months time. Don't be surprised by the support that will swell in AMD's favor, no matter how much FUD is spread between now and then.

Also, I think you'll find that many supporters of AMD products aren't merely supporting them for loyalty, fanboi, etc reasons like you suggest, but out of principle and the desire for a healthy marketplace that has been held hostage by intel for 20 years. Obviously, anyone can support whoever they want and be on whichever side of the fence they choose. It's too bad the vast majority of consumers are uninformed on the practices of the x86 market over the last 2 decades of rule by intel. You'd think that the informed technically savvy crowd would want better though.

Didn't you say you would never visit this site again? Have you had a change of heart?
 
Should have said IMO. I think you're dead wrong because I don't think AMD HAS lost a lot of support. You may not see a lot of supporters posting, because if you are old enough (or young enough if you are too old) then you might remember the complete ridiculing dished out by intel supporters and marketing. I think you'll see a quite different landscape of support when AMD's vision of the future, which they've been preparing for 5 years, really surface in a couple months time. Don't be surprised by the support that will swell in AMD's favor, no matter how much FUD is spread between now and then.

Also, I think you'll find that many supporters of AMD products aren't merely supporting them for loyalty, fanboi, etc reasons like you suggest, but out of principle and the desire for a healthy marketplace that has been held hostage by intel for 20 years. Obviously, anyone can support whoever they want and be on whichever side of the fence they choose. It's too bad the vast majority of consumers are uninformed on the practices of the x86 market over the last 2 decades of rule by intel. You'd think that the informed technically savvy crowd would want better though.

I've seen enough posts to indicate that they have, in two forms.

The first and most common is financial support. People just couldnt justify buying AMD products when they knew Intel products were superior. Yes, price points and all that, but lets face, Intel has made some very competitive products.

The second, and less common but still occurring, is, err, moral support, like me. I had an Athlon K7-700, then an Athlon XP 2400+, then an Athlon X2 5600+, and now an Athlon II X4 620. I might buy Bulldozer if its good, but I think I'm done rooting for AMD in the moral sense. What Intel did to the market years ago was terrible, and no doubt AMD still feels the hurt today. However, as far as we know, those practices have stopped, and Intel not only paid money to both Nvidia and AMD, and a lot of it, they have also had to pay fines to the FTC and the EU, as well as enter into technology sharing agreements that they probably didnt want. You might say, they've faced the music. Since then, they make great products, they release them on time. AMD..... hasnt done anything exciting since the Phenom II. That was years ago, and even that wasnt enough to properly stick it to Intel.

No doubt Bulldozer will perform well and sell well - from what I've read its awesome. And, in today's landscape, AMD might actually be able to sell a few, unlike in earlier years where it was impossible. I feel that the moral duty of supporting AMD because of Intel's foul play has pretty much now passed, and now I face the choice of a company that cant seem to organize a pissup at a brewery, or a company that, well, just works.
 
I don't get this sense of entitlement that some people seem to have in regards to AMD, Bulldozer, and information from one about the other.

Some of you guys are acting like spurned or jilted lovers.

Frankly it is bizarre.

No doubt. It's kind of sad and really weird how some people feel they are entitled to so much these days. I guess some people have to have something to QQ about. If AMD wants to release all of their info at launch, then that's their prerogative. No need in getting butt hurt about it.
 
Not really, this is the old source, see the link to the origin from 2010-10-19 if you look at the original source:
http://pc.watch.impress.co.jp/img/pcw/docs/408/107/html/02.jpg.html
So there is nothing new from ISSCC as far as I can see.

With that picture you get 275-285 mm² by measuring pixel count (I used gimp for that).

We do not know if it was obfuscated by AMD.
In addition this is a Opteron 4-server part with 4 hypertransport links. Don't know if they use that also for the Zambezi client parts.

This would be a really bad engineering. Such a waste of die space in the uncore area.

I still do believe that the Zambezi will be well below that 280 mm². And if not it is only because they use a 4 hypertransport links die for Zambezi.

At least it shows that the first 300+ calculations (300-360) with the other obfuscated die picture (before 2010-10-19) were wrong because they took the smaller cores as a base but as this picture from october last year suggests the larger cores are real. However it is very complicated if not impossible to determine the obfuscation in the uncore area.

Why should it be that much larger than Sandy Bridge when the size of a Bulldozer module (18.0 mm²) is even smaller than that of a Sandy Bridge Core (18.9 mm²). Yes Bulldozer has 6 MB more cache (+ 38.7 mm²) but on the other hand Sandy Bride has a GPU (+ xx mm²) on it.
 
Status
Not open for further replies.
Back
Top