Bulldozer Clocks leaked

Hmm the 2nd from the top is 3.1 Ghz base, 4.1 Ghz "Turbo" frequencies. The top one on the chart is odd...3.2 base, 3.7 turbo? You have to wonder if AMD is pushing these hard to get to the turbo frequencies in hopes of beating out Sandy Bridge in the numbers game, but the OC numbers will tell the real tale.
 
I agree the model numbers are confusing. Also it is confusing that the model #s don't mention cores. And the same model number can have a different turbo.

I am disappointed with the low base clocks considering these chips will most likely have a lower per core IPC than SB but I am optimistically happy with the turbo.
 
Last edited:
Also it is confusing that the model #s don't mention cores.

AMD_FX_Zambezi_IHS_600.jpg


Right after the W, there is an 8, for 8 cores :p
 
Also, what I find interesting, is that the highest base clock, has a lower turbo clock...

Looks to me like they are concerned with either power use or heat dissipation, otherwise there is no reason the highest base clocked part should have a lower turbo speed.
 
I agree the model numbers are confusing. Also it is confusing that the model #s don't mention cores. And the same model number can have a different turbo.

I am disappointed with the low base clocks considering these chips will most likely have a lower per core IPC than SB but I am optimistically happy with the turbo.


those aren't the final model numbers, those are the ES model number codes. they are not what the final name will be, this information was already posted only this time it added some random persons opinion to the information..

http://hardforum.com/showthread.php?t=1603408
 
Zarathustra[H];1037170855 said:
AMD_FX_Zambezi_IHS_600.jpg


Right after the W, there is an 8, for 8 cores :p

I know that there is extra code for core count.

What I meant is will for example model 2420 be an 4, 6 and 8 core chip?

I guess 2420 x4 ...
 
Last I saw the FX series will have 8000, 6000, and 4000 denominations signifying how many cores are in each one
 
I know that there is extra code for core count.

What I meant is will for example model 2420 be an 4, 6 and 8 core chip?

I guess 2420 x4 ...
I think the only upcoming lines that will have model numbers that low (if this information is correct) are Llano-based products, even though there will be no Llano parts over 4 cores.

Maybe, whoever originally put that chart together was piecing together different bits of info, because most of that chart really doesn't mesh well with the information that is already out about the FX series.

EDIT:
In fact, if the 4-digit number corresponds to the actual model numbers, then those numbers are for the Llano A8, A6 and A4 series. According to this link from X-bit labs, the lowest model number is the E2-3250.
 
I think the only upcoming lines that will have model numbers that low (if this information is correct) are Llano-based products, even though there will be no Llano parts over 4 cores.

Maybe, whoever originally put that chart together was piecing together different bits of info, because most of that chart really doesn't mesh well with the information that is already out about the FX series.

Llano won't be a bulldozer part. Llano uses K10.5 cores, just like Phenom II.
 
I think the only upcoming lines that will have model numbers that low (if this information is correct) are Llano-based products, even though there will be no Llano parts over 4 cores.

Maybe, whoever originally put that chart together was piecing together different bits of info, because most of that chart really doesn't mesh well with the information that is already out about the FX series.


like i already said, they are ES model numbers not actual processor numbers. ES model numbers have no science or reasoning for them. they change with every processor generation, some use the clocks, some just use random numbers. they don't signify anything.

the actual model numbers will all be based on the FX 8xxx, FX 6xxx, FX 4xxx numbers. what they actually decide to use in place of the x's is anyone's guess.
 
Bulldozer is going to own.

It would "own" if AMD could match SB in IPC however that is highly doubtful. The only time I can remember Intel or AMD managed > 30% improvement in per core IPC on a new generation is when core2 was introduced and that was mainly because P4 was very inefficient.
 
Zarathustra[H];1037171022 said:
Llano won't be a bulldozer part. Llano uses K10.5 cores, just like Phenom II.

like i already said, they are ES model numbers not actual processor numbers. ES model numbers have no science or reasoning for them. they change with every processor generation, some use the clocks, some just use random numbers. they don't signify anything.

the actual model numbers will all be based on the FX 8xxx, FX 6xxx, FX 4xxx numbers. what they actually decide to use in place of the x's is anyone's guess.

I know that, but what I was saying that maybe the person that made this chart was using different bits of info, because that wasn't an AMD-made OPN chart. It wouldn't be the first time someone has made a chart using information from completely different products.
 
It would "own" if AMD could match SB in IPC however that is highly doubtful. The only time I can remember Intel or AMD managed > 30% improvement in per core IPC on a new generation is when core2 was introduced and that was mainly because P4 was very inefficient.

Bulldozer is a new ground up design so you can't really tell.
 
Hmm the 2nd from the top is 3.1 Ghz base, 4.1 Ghz "Turbo" frequencies. The top one on the chart is odd...3.2 base, 3.7 turbo? You have to wonder if AMD is pushing these hard to get to the turbo frequencies in hopes of beating out Sandy Bridge in the numbers game, but the OC numbers will tell the real tale.

Its hard to say. These may be representative of shipping parts but that doesn't mean that they will be the shipping parts (I mean outside being engineering parts in the first place). For example the have multiple models named 2820 3020 2620. Two of them have 3 different turbo values. My guess is for full evaluation of power requirements for a given range. Also notice the name of one used as the sample. Its based on the A01 revision and not the B0. What isn't shown is on the table of models, what CPU revision they are. It could be that the 3220, was revision A1 chip they had sent out before (maybe when they didn't fit AM3 sockets) and the 3120 was based off the Rev B0 chip. Probably allowing the 4.1GHz top end to have a reduced load power usage. Considering that models with 3 versions tend to have a 500Mhz version to go with the 900MHz and 1GHz versions, I think that would be highly likely.

Also keep in mind that AMD also intends to have BE versions of some of the FX's. It could be that the 500Mhz turbo's are either the non BE ones for easy to recognize reasons or in a wierd twist the 500MHz versions are actually the BE models. So that Turbo can be left on without it trying crazy clocks. Remember just because they are trying to run closer to TDP doesn't mean that TDP is the only limiting factor of running faster. So the 3.2 GHz version since their is only one of them might actually be the BE version, implying that they feel decently safe in having you run it as high as 3.7 for a base clock.
 
Bulldozer is a new ground up design so you can't really tell.

Agreed, but AMD themselves stated that 8 core server bulldozers will be 50% faster that hexacore k10 models (presumably at the same clock)

This translates to only a 12.5% IPC increase per core.

Based on that, its doubtful that bulldozer will even catch up with last gen Core i7's (or i5's for that matter).

Bulldozer DOES have more cores, (shared FPU instead of hyperthreading ought to be faster in highly multithreaded applications) but the truth is for most desktop use today we see limiting returns from more cores already somewhere between 2 and 4, sop boosting to 8 cores is likely useless for desktop use (except for those rendering / encoding a lot)

So, while - as an AMD fan of the past - am hopeful that AMD will catch up to Intel in desktop performance, being realistic, I see it as highly unlikely in this generation of CPU's. At least not until multithreaded code becomes more pervasive in desktop applications than it is today.

This doesn't mean that Bulldozer will suck. On the contrary, it ought to be a great CPU, able to power through any game, even with quad SLI/ XFire on the highest end GPU's without being the bottleneck, and run any desktop application really well for years to come, but Sandy Bridge will likely still be faster when comparing the heavy CPU loads...
 
Its hard to say. These may be representative of shipping parts but that doesn't mean that they will be the shipping parts (I mean outside being engineering parts in the first place). For example the have multiple models named 2820 3020 2620. Two of them have 3 different turbo values. My guess is for full evaluation of power requirements for a given range. Also notice the name of one used as the sample. Its based on the A01 revision and not the B0. What isn't shown is on the table of models, what CPU revision they are. It could be that the 3220, was revision A1 chip they had sent out before (maybe when they didn't fit AM3 sockets) and the 3120 was based off the Rev B0 chip. Probably allowing the 4.1GHz top end to have a reduced load power usage. Considering that models with 3 versions tend to have a 500Mhz version to go with the 900MHz and 1GHz versions, I think that would be highly likely.

Also keep in mind that AMD also intends to have BE versions of some of the FX's. It could be that the 500Mhz turbo's are either the non BE ones for easy to recognize reasons or in a wierd twist the 500MHz versions are actually the BE models. So that Turbo can be left on without it trying crazy clocks. Remember just because they are trying to run closer to TDP doesn't mean that TDP is the only limiting factor of running faster. So the 3.2 GHz version since their is only one of them might actually be the BE version, implying that they feel decently safe in having you run it as high as 3.7 for a base clock.

http://www.xbitlabs.com/news/cpu/di...ulldozer_Chips_Incoming_Details_Revealed.html

All BE

FX = Automatic BE
 

That might be initial products out the door. Might even be the only retail versions available. But considering the lineup of ships they have sent for testing, why are there only two 8 core versions there? 1 6 core? 1 4 core?

Specially if the goal of keeping quiet is to sell through PhII and Athlon II stock.

Maybe those three are listed not because they are the only BD chips selling soon, but those are the only ones available in BE?

Maybe the 3020 3GHz-3.6GHz is the 8110 and the 3220 3.2GHz-3.7GHz is the 8130?

Maybe the non BE versions will have 80*0 numbers and where not included.

Maybe the 3 2620 versions are same binned chips for testing, 1 representative of an 8 core, another a 6 core, and the third a 4 core chip. Maybe the 2620 6 core might be the clock speed of what will be the 6 core BE (6110) or it could be the third 3020. The ones with a 1GHz turbo might just be 6 or 4 core versions. We don't know because these chips are not what we will be receiving.
 
That might be initial products out the door. Might even be the only retail versions available. But considering the lineup of ships they have sent for testing, why are there only two 8 core versions there? 1 6 core? 1 4 core?

Specially if the goal of keeping quiet is to sell through PhII and Athlon II stock.

Maybe those three are listed not because they are the only BD chips selling soon, but those are the only ones available in BE?

Maybe the 3020 3GHz-3.6GHz is the 8110 and the 3220 3.2GHz-3.7GHz is the 8130?

Maybe the non BE versions will have 80*0 numbers and where not included.

Maybe the 3 2620 versions are same binned chips for testing, 1 representative of an 8 core, another a 6 core, and the third a 4 core chip. Maybe the 2620 6 core might be the clock speed of what will be the 6 core BE (6110) or it could be the third 3020. The ones with a 1GHz turbo might just be 6 or 4 core versions. We don't know because these chips are not what we will be receiving.

Those might actually be the future Trinity(<-Llano) Bulldozer clocks not just the current clocks as anything with the FX = Black Edition + Turbo Core 2.0

FX name is a high end name so all of the CPUs have to be unlocked and have potential to OC

The 8130P is also clocked at 3.5GHz which is not shown
 
Those might actually be the future Trinity(<-Llano) Bulldozer clocks not just the current clocks as anything with the FX = Black Edition + Turbo Core 2.0

FX name is a high end name so all of the CPUs have to be unlocked and have potential to OC

The 8130P is also clocked at 3.5GHz which is not shown

I doubt they are seeing to many Trinity options. I wouldn't put a whole lot of trust into a rumor posting, specially with the limited options. I think you are readying waaaaay to much into a name. Lets just think clearly and slowly about the names.

First is the E series CPU's (known as Zacate, but there is at least 1 Llano). What service do they provide? They are economy chips sold in low power usage thin and light machines making them also Eco friendly. So they gave it an E designation. Much as Mecedes would do with the E class cars.

Okay A. Above even Zacate that shipped earlier AMD has done nothing but push that these chips are an absolute new class of CPU's and the driving force of new generation in CPU's called APUs. Makes sense that the letter designation would be an A for APU.

So knowing that. Knowing that those are two new markets for AMD. What would you call your performance chips. Since you are losing names (and frankly they destroyed the Athlon name and the Phenom name wasn't worth much) but you want people, specially fans of their products to recognize that this is the Performance series and part of their long line of performance targeted chips. FX as by your post implies that it still holds weight and something people feel familiar with. Its a perfect two letter assignment for that series.

What I think is going on is to many people are looking at that and trying to out think themselves. Is a FX-4K chip really going to cost $1000? I doubt it, yet before now AMD has never sold a FX chip for less then $800. FX just means performance in a time where they are more market diverse then they have ever been before. That doesn't mean every performance chip has to be unlocked. Doesn't mean that A-**** can be more expensive then a FX-****, just that they are aimed at different markets.

Also "maybe be as high as 3.5GHz" doesn't mean that it will be. Considering that they are starting out with one at 3.2 and Turbo clocks over 4GHz. I wouldn't be surprised if we see a 3.5GHz in September or October if BD sells well, without any changes, just by building up stock on their best chips and waiting.
 
I doubt they are seeing to many Trinity options. I wouldn't put a whole lot of trust into a rumor posting, specially with the limited options. I think you are readying waaaaay to much into a name. Lets just think clearly and slowly about the names.

First is the E series CPU's (known as Zacate, but there is at least 1 Llano). What service do they provide? They are economy chips sold in low power usage thin and light machines making them also Eco friendly. So they gave it an E designation. Much as Mecedes would do with the E class cars.

Okay A. Above even Zacate that shipped earlier AMD has done nothing but push that these chips are an absolute new class of CPU's and the driving force of new generation in CPU's called APUs. Makes sense that the letter designation would be an A for APU.

So knowing that. Knowing that those are two new markets for AMD. What would you call your performance chips. Since you are losing names (and frankly they destroyed the Athlon name and the Phenom name wasn't worth much) but you want people, specially fans of their products to recognize that this is the Performance series and part of their long line of performance targeted chips. FX as by your post implies that it still holds weight and something people feel familiar with. Its a perfect two letter assignment for that series.

What I think is going on is to many people are looking at that and trying to out think themselves. Is a FX-4K chip really going to cost $1000? I doubt it, yet before now AMD has never sold a FX chip for less then $800. FX just means performance in a time where they are more market diverse then they have ever been before. That doesn't mean every performance chip has to be unlocked. Doesn't mean that A-**** can be more expensive then a FX-****, just that they are aimed at different markets.

Also "maybe be as high as 3.5GHz" doesn't mean that it will be. Considering that they are starting out with one at 3.2 and Turbo clocks over 4GHz. I wouldn't be surprised if we see a 3.5GHz in September or October if BD sells well, without any changes, just by building up stock on their best chips and waiting.

desktop.jpg


notebook.jpg


Then again this post can actually be talking about the Server Bulldozer cpus(The first post)

FX-8130P will be 3.5GHz with a 500(Stock)-XX0MHz(User Defined) turbocore

amd-buldozer-positioning_x550.png


It needs to be to even fight i7-2600k ;)
 
Last edited:
I saw the system price, interpreted it as CPU price, and had a little shock moment. I wonder what they consider a system. I already have a monitor, would that knock $350 off the total?
 
Then again this post can actually be talking about the Server Bulldozer cpus(The first post)

FX-8130P will be 3.5GHz with a 500(Stock)-XX0MHz(User Defined) turbocore
It needs to be to even fight i7-2600k ;)

Not sure what you want me to see in the road map. Or where you get definitive clock speeds. I don't know where you get the 500MHz turbo or the idea that we can change the turbo. I really am not sure what you are debating me on.

We have 4 supposedly known shipping BE parts, that's one posting. In that they suggested that it "may be clocked as high as 3.5GHz. We have an outline of parts used for testing that range in over 800Mhz in defualt clock and 700Mhz of Turbo clock, with 18 different variations. Which is another.

I don't know why you believe some of those are Trinity since they either just have or haven't yet started shipping "first gen" BD, so its unlikely they are already doing platform tests on Komodo or Trinity. Specially since again the highest default speed CPU in the testing chart is running slower then the supposed initial default speed of first gen BD. Secondly "Trinity" will be either FS1/FS2 (or both).

I would stop using definitive statements in regards to BD, as even the info you have linked have not said anything near definitive.
 
Curious to see how they perform. I feel a new build might be due again this year.
 
That system price got me too haha, $700 CPU to compete with 2600k at $300? Then I saw the system part, and saw that the 2600k is in the same price bracket.

All in all I am excited to see what happens. My next upgrade is comming soon, and the way it looks bulldozer may land right in that window.

8 cores sounds so awesome for folding.
 
I saw the system price, interpreted it as CPU price, and had a little shock moment. I wonder what they consider a system. I already have a monitor, would that knock $350 off the total?

I know you made a simple and easy to make mistake in what you saw and I am not jumping on you at all but your post made me think about something that for some reason just irritates me.

This might be a nitpick and I could be wrong about the following of course but I am constantly amazed that people actually think that Bulldozer CPUs or builds based on Bulldozer could be priced much higher than the Intel 2x00K CPUs and Intel builds. I mean, WTF would give anyone that impression? That the laws of supply and demand and competition are going to somehow work in reverse once Bulldozer is released? AMD/Bulldozer is going to have to be competitive. Bulldozer would have to be far and away, demonstrably better for them to be able to charge significantly more than what Intel is charging for their CPUs. And who here really believes that will be the case? I like AMD. I am pulling for the underdog but I don't think there is a snowballs chance in hell that Bulldozer will be so significantly better that we will be looking at significantly higher than $300 for any Bulldozer CPU. I think it is kind of crazy to even think that given recent history, AMDs place in the market and the current competition.
 
I know you made a simple and easy to make mistake in what you saw and I am not jumping on you at all but your post made me think about something that for some reason just irritates me.

This might be a nitpick and I could be wrong about the following of course but I am constantly amazed that people actually think that Bulldozer CPUs or builds based on Bulldozer could be priced much higher than the Intel 2x00K CPUs and Intel builds. I mean, WTF would give anyone that impression? That the laws of supply and demand and competition are going to somehow work in reverse once Bulldozer is released? AMD/Bulldozer is going to have to be competitive. Bulldozer would have to be far and away, demonstrably better for them to be able to charge significantly more than what Intel is charging for their CPUs. And who here really believes that will be the case? I like AMD. I am pulling for the underdog but I don't think there is a snowballs chance in hell that Bulldozer will be so significantly better that we will be looking at significantly higher than $300 for any Bulldozer CPU. I think it is kind of crazy to even think that given recent history, AMDs place in the market and the current competition.

It's entirely reasonable to make that initial assumption that they meant CPU price (even though it would be ridiculous). I did, because the dollar price is in huge font, and "SYSTEM PRICE" is in tiny font. It's not people being stupid it's a stupid layout on the chart (interesting though that chart may be).
 
I don't know why you believe some of those are Trinity since they either just have or haven't yet started shipping "first gen" BD, so its unlikely they are already doing platform tests on Komodo or Trinity. Specially since again the highest default speed CPU in the testing chart is running slower then the supposed initial default speed of first gen BD. Secondly "Trinity" will be either FS1/FS2 (or both).

Llano->Trinity
FS1+FM1->FS2+FM2
if they need a new socket for the FM/FS socket
 
I know you made a simple and easy to make mistake in what you saw and I am not jumping on you at all but your post made me think about something that for some reason just irritates me.

I hear you. I expect any AMD chip, at their current level of competitiveness, to offer a high price per performance value in gaming.
 
This might be a nitpick and I could be wrong about the following of course but I am constantly amazed that people actually think that Bulldozer CPUs or builds based on Bulldozer could be priced much higher than the Intel 2x00K CPUs and Intel builds. I mean, WTF would give anyone that impression? That the laws of supply and demand and competition are going to somehow work in reverse once Bulldozer is released? AMD/Bulldozer is going to have to be competitive. Bulldozer would have to be far and away, demonstrably better for them to be able to charge significantly more than what Intel is charging for their CPUs. And who here really believes that will be the case? I like AMD. I am pulling for the underdog but I don't think there is a snowballs chance in hell that Bulldozer will be so significantly better that we will be looking at significantly higher than $300 for any Bulldozer CPU. I think it is kind of crazy to even think that given recent history, AMDs place in the market and the current competition.

A little wordy arn't wee...

All I can say is exucse f'in me but here is the reason.

2500K http://www.microcenter.com/single_product_results.phtml?product_id=0354589

is $179.00

Unlocked blah blah blah AMD chip is about $230

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103913&cm_re=1100T-_-19-103-913-_-Product

So fifty bucks or a 30% premium is what we are currently paying for a top of the line AMD chip that doesn't compete.

And don't give me any 6 core nonsence, I'm not an AMD fan boy but due own stock, and have to defend myself against Intel fan boys so...
 
A little wordy arn't wee...

All I can say is exucse f'in me but here is the reason.

2500K http://www.microcenter.com/single_product_results.phtml?product_id=0354589

is $179.00

Unlocked blah blah blah AMD chip is about $230

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103913&cm_re=1100T-_-19-103-913-_-Product

So fifty bucks or a 30% premium is what we are currently paying for a top of the line AMD chip that doesn't compete.

And don't give me any 6 core nonsence, I'm not an AMD fan boy but due own stock, and have to defend myself against Intel fan boys so...

http://www.newegg.com/Product/Product.aspx?Item=N82E16819115072

Compare from the same retailer ;)

Microcenter doesn't ship to everywhere
 
What I would say is that the 8-core is probably going to come in around the $400 range regardless of that it does. 398 or 389 etc...

I preface that say saying if it has devin intervention; and that for some reason it's 300% better than the 12 core magna or 200% better than the 2600K AMD will charge $700 for it.

In reality AMD is try to snooker us into thinking it's a SB equal or better and is going to be cheaper. Or equal in price but 10% better. It's all in the yard stick measure it based on what 10% better in energy efficiency or multibuyraygameingsoundoutlockwhilesurfingtheinternetandgaming benchmark cool.

AMD PLEASE RELEASE SOME REAL BENCHMARKS BEFORE Z68 OR ILL SELL SOME STOCK TO BUY A Z68 2500K COMBO.

Sorry to rant but just wanted to make myself clear. :(
 
What I would say is that the 8-core is probably going to come in around the $400 range regardless of that it does. 398 or 389 etc...

I preface that say saying if it has devin intervention; and that for some reason it's 300% better than the 12 core magna or 200% better than the 2600K AMD will charge $700 for it.

In reality AMD is try to snooker us into thinking it's a SB equal or better and is going to be cheaper. Or equal in price but 10% better. It's all in the yard stick measure it based on what 10% better in energy efficiency or multibuyraygameingsoundoutlockwhilesurfingtheinternetandgaming benchmark cool.

AMD PLEASE RELEASE SOME REAL BENCHMARKS BEFORE Z68 OR ILL SELL SOME STOCK TO BUY A Z68 2500K COMBO.

Sorry to rant but just wanted to make myself clear. :(

http://www.hardwaresecrets.com/article/Inside-the-AMD-Bulldozer-Architecture/1078/2

might want to read this tid bit....AMD screwed them selves by not evolving the Instruction sets

SSE3 is not enuff SSE4A is BS

But at least Bulldozer will have everything supported
 
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115072

Compare from the same retailer ;)

Microcenter doesn't ship to everywhere

Sure...
http://www.microcenter.com/single_product_results.phtml?product_id=0334811
$199.00 it is multiplier unlocked and basically the same chip

And I get a free motherboard
http://www.microcenter.com/single_product_results.phtml?product_id=0351634
a 50 dollar value ????

But then again you have to be next to a microcenter and the board suxs...
 
It's entirely reasonable to make that initial assumption that they meant CPU price (even though it would be ridiculous). I did, because the dollar price is in huge font, and "SYSTEM PRICE" is in tiny font. It's not people being stupid it's a stupid layout on the chart (interesting though that chart may be).

I didn't say it was unreasonable and I certainly never said anyone was stupid for making this particular mistake regarding this screenshot. In fact, I said specifically that I understand people seeing that chart that way. Corvette seemed to understand what I was saying.

My point was that I read from time to time how the new Bulldozer "FX" chips are going to be $700+ OMGz!1!1!! I think that is not very well thought out, "crazy" thinking. They would have to be some insane amount better than Intel's offerings and given that AMD has been behind for an eternity now, I have zero faith that Bulldozer will separate itself in that way. Not to mention that there will be very few people who would actually buy them when they could have a 5GHz Intel CPU for under $200. AMD would be cutting their own throat by pricing Bulldozer significantly higher than the Intel offerings - almost no matter how much better it is (if it is).
 
if these clocks are true, I'm exited about thew OC potential of these things!
 
Back
Top