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

[GaiaHunter]

I not sure . But it would seem that some people are not being honest here .

Ya I understand that AMD modules are 2 core 2 threads.

But whats not being said is Intels modules and thats what intel calls them are 2 cores 4 threads . Now AMD may or may not end up with more modules than intel on 1 die. But intels single module is 2 cores 4 threads. Ya can break it up any way ya want . But thats the way it is . so if we take a 2 module BD and go up against 2 module SB . this is what we have .

AMD 2 modules= 4 cores 4 threads . Intels 2modules= 4 cores 8 threads .

AMD may choose to go with more modules per die but than its not apples to apples.

So basically what you are saying is that price doesn't matter for the consumer. Die size doesn't matter for the companies producing the silicon.

You are saying that if an AMD BD quad-core (a 2 modules) is half the price and die size of an Intel IB quad-core (4c/8t), you still need to compare it to that and not an IB dual core (2c/4t) that is the same die size and price?

So the only possible benchmark is core vs core - price, power consumption, die size, be damned?

I don't care about HT or CMT . You missed the whole point! 1 module AMD = 2 cores, 1 module SB = 2 cores. BP 1 module = 2 core 2 threads SB 1 module = 2 cores 4 threads

What is your obsession with the module?

Intel doesn't even call it modules - it is an AMD designation to show how they are organizing the resources inside their chip.

Just call it processors.

Compare processor to processor.
Those are the ones that do the stuff.
Processor is what you buy in stores. Processor is what you slot in your motherboard.

 

[itsmydamnation]

I don't care about HT or CMT . You missed the whole point! 1 module AMD = 2 cores, 1 module SB = 2 cores. BP 1 module = 2 core 2 threads SB 1 module = 2 cores 4 threads

You say I am twisting to suite my end . LOL! I am using the same metric for BD and SB as whats been traditional in hardware forums . It is AMD who is hyping a new method of performance based on their end. That even fails . We have JFAMD saying BD in 2 nd qt 2011 for BD yet yesterday he said its hard to make guesstaments based on NON final silly. Its going to be enen harder for AMD to have BD in 2nd qt if they haven't final silly at this date. It also tells me that the lastest waffers sucked bigtime with the management repositioning going on at AMD.

simple facts,

module size == SB core size
module total int resources == SB core total int resources
module cache hierarchy (almost)== SB cache hierachy
module FP resources ==(kinda) SB core FP resouces ( SB more AVX , bulldozer more SSE 1,2,3,4)

the difference is the front end setup, simple fact is, a bulldozer module is very comparable to a SB core.

how about you justifity your position from a technical standpoint . JF AMD has said millions of times, that modules are used to describe the architexture and are not going to be used in marketing in any way shape or form, but if you have to make stuff up to justify your position dont be suprised when people call it.

 

[Nemesis 1]

Above your doing the same thing as the decievers. But ya 1 upped them

module size == SB core size
Above what does core size have to do with it . Are you including the HD3000 in core size LOL.

module total int resources == SB core total int resources
I will wait befor posting what intel calls a SB core later This is to good to ruin.

1 amd module 2 cores = total int resources. = about 2x

1 intel module 2 cores= total int resources = 2x

Were did I make anything up . Are you saying that intel doesn't referr to 2 SB cores as a module. You will have to wait as I won't post to intel until latter.


You wrote this .

JF AMD has said millions of times, that modules are used to describe the architexture and are not going to be used in marketing in any way shape or form, but if you have to make stuff up to justify your position dont be suprised when people call it.

Than why is he using it to market BD. Ya I see what AMD has done with BD . shared resources . I preferr unshared resources for IPC and the win.

You yourself need to explain this goodie .

module FP resources ==(kinda) SB core FP resouces ( SB more AVX , bulldozer more SSE 1,2,3,4) I couldn't help but LOL . Not picking on ya . But my grandson knows more than this and he is just turning 4 and we fight over the gaming pc.

 

[Idontcare]

So basically what you are saying is that price doesn't matter for the consumer. Die size doesn't matter for the companies producing the silicon.

You are saying that if an AMD BD quad-core (a 2 modules) is half the price and die size of an Intel IB quad-core (4c/8t), you still need to compare it to that and not an IB dual core (2c/4t) that is the same die size and price?

So the only possible benchmark is core vs core, price, power consumption, die size, be damned?



What is your obsession with the module?

Intel doesn't even call it modules - it is an AMD designation to show how they are organizing the resources inside their chip.

Just call it processors.

Compare processor to processor.
Those are the ones that do the stuff.
Processor is what you buy in stores. Processor is what you slot in your motherboard.

Nemesis clearly needs his meds, or less alcohol, or more of both, or less of either...my point is just that you guys are truly wasting your time slamming your heads against a rock.

Regardless, I wanted to say that from the perspective of a day-to-day programmer you/we really should NOT call them "processors" as processors in the context of windows is a measure of the thread processing capabilities of the CPU.

A quad-core CPU will be reported as having 4 processors.

And there are such things as logical processors for HT/SMT CPU's.

We really ought to be referring to the thing we plug into the socket as simply "the CPU".

Yes processor can be used in the context of referring to the CPU itself, but it is open to easy misinterpretation and fanboy battles royale if we as a CPU community adopt such terminology.

 

[maddie]

I don't care about HT or CMT . You missed the whole point! 1 module AMD = 2 cores, 1 module SB = 2 cores. BP 1 module = 2 core 2 threads SB 1 module = 2 cores 4 threads

You say I am twisting to suite my end . LOL! I am using the same metric for BD and SB as whats been traditional in hardware forums . It is AMD who is hyping a new method of performance based on their end. That even fails . We have JFAMD saying BD in 2 nd qt 2011 for BD yet yesterday he said its hard to make guesstaments based on NON final silly. Its going to be enen harder for AMD to have BD in 2nd qt if they haven't final silly at this date. It also tells me that the lastest waffers sucked bigtime with the management repositioning going on at AMD.

It's more accurate to say 1 module AMD = 2 threads, 1 core SB = 2 threads.

Do you see yourself as the sole arbiter of definitions?

 

[Nemesis 1]

So basically what you are saying is that price doesn't matter for the consumer. Die size doesn't matter for the companies producing the silicon.

You are saying that if an AMD BD quad-core (a 2 modules) is half the price and die size of an Intel IB quad-core (4c/8t), you still need to compare it to that and not an IB dual core (2c/4t) that is the same die size and price?

So the only possible benchmark is core vs core - price, power consumption, die size, be damned?



What is your obsession with the module?

Intel doesn't even call it modules - it is an AMD designation to show how they are organizing the resources inside their chip.

Just call it processors.

Compare processor to processor.
Those are the ones that do the stuff.
Processor is what you buy in stores. Processor is what you slot in your motherboard.

Never said no such thing . Price = usage. Were discussing performance . As of this moment AMD cpus are way overpriced for the performance . About $160 should be the price of topend AMD desktop cpus. Tablets are cool and there cheap but I only seen one I want and thats sometime off. It certainly isn't for myself as they are cool and all but I use a P4c as browser and its almost at a crawl I hate it now . Intel needs to release the Z chip yesterday. Point is my old intel p4c is about = to these tablets and that is a joke.

 

[Nemesis 1]

It's more accurate to say 1 module AMD = 2 threads, 1 core SB = 2 threads.

Are you see yourself as the sole arbiter of definitions?

No I am not . I understand the shared resources . As for definitions . If intel calls 1 2core SB a module is their definition not suitable for you . Ya there are no shared resources other than IGP. SB 2600K has 6 cores intel SB modules with a total of 2 SB modules 4 cores 8 threads 2x IGP cores that would be 2x6 shaders which is 1 module and 12 shader cores and they share resources. Software will determine what shared resources are not AMD or Intel hardware.

 

[Nemesis 1]

Nemesis clearly needs his meds, or less alcohol, or more of both, or less of either...my point is just that you guys are truly wasting your time slamming your heads against a rock.

Regardless, I wanted to say that from the perspective of a day-to-day programmer you/we really should NOT call them "processors" as processors in the context of windows is a measure of the thread processing capabilities of the CPU.

A quad-core CPU will be reported as having 4 processors.

And there are such things as logical processors for HT/SMT CPU's.

We really ought to be referring to the thing we plug into the socket as simply "the CPU".

Yes processor can be used in the context of referring to the CPU itself, but it is open to easy misinterpretation and fanboy battles royale if we as a CPU community adopt such terminology.

Come on IDC . alcohol, I don't drink or smoke pot anymore . I even stopped taking the illegal drugs I got from Mexico.

You haven't noticed the sharred resources thing . If Intel uses HT is that not sharred resources. Yep AMDs CMT seems to perform better than HT . So what? Have you not noticed that when ya open Task manager that 4 cores are present but 8 threads is that not sharred resources. Be it not 2 cores its shared resources in a single core.

I didn't see ya say anything to this poster about his drinking problems whatever. The one who said AMD coined the word CPU module . Idare ya to put cpu module into google search . AMD coined the word module LOL!

 

[Anexate]

Nemesis clearly needs his meds, or less alcohol, or more of both, or less of either..

:ninja:
a vacation in Tunisia should do wonders ^_^

my point is just that you guys are truly wasting your time slamming your heads against a rock.

this thread could compete for Nobel prize for Greek philosophy

 

[Idontcare]

Come on IDC . alcohol, I don't drink or smoke pot anymore . I even stopped taking the illegal drugs I got from Mexico.

You haven't noticed the sharred resources thing . If Intel uses HT is that not sharred resources. Yep AMDs CMT seems to perform better than HT . So what? Have you not noticed that when ya open Task manager that 4 cores are present but 8 threads is that not sharred resources. Be it not 2 cores its shared resources in a single core.

I didn't see ya say anything to this poster about his drinking problems whatever. The one who said AMD coined the word CPU module . Idare ya to put cpu module into google search . AMD coined the word module LOL!

I believe you, but I fail to comprehend the relevance of your statements in relation to the topic that the other 99.9% of us are trying to discuss here.

You of all people know that it is not enough to merely be "correct" when it comes to matters of communicating the truth.

Presentation is key, as is relevance.

And right now you are failing, clearly, as you haven't convinced anyone here that your message has relevance.

Without establishing relevance, not many of us care to invest the efforts needed to decipher your message to determine if there is any truth to it.

Just being pragmatic about it. We all have to filter out the cranks in our lives or else we'd never get anything done aside from listening to cranks.

Right now your posts in here come across as that of a crank, we are inclined and predisposed to dismiss your arguments.

If you want to change the balance of things then you need to compose yourself and your thoughts, be less like the drunk on the street corner who is claiming the end is near and we better all repent, and present a compelling argument and position that underscores whatever point it is that you think you need to be making here.

Until then we are all going to just be beating our heads on the rock because we haven't a fapping clue what it is you are really driving at, and all of our collective guessing on the matter are not well received by you which suggests to me we have yet to "get" your point.

 

[Nemesis 1]

What in the world are you talking about?

Anyone else ever heard of this?

Sure they have . If they read ATs articles . The one on SB were ananda talks about the ring bus and the stops . and how intel says . They can add or subtract any kind of module they like . read the article

 

[Nemesis 1]

I believe you, but I fail to comprehend the relevance of your statements in relation to the topic that the other 99.9% of us are trying to discuss here.

You of all people know that it is not enough to merely be "correct" when it comes to matters of communicating the truth.

Presentation is key, as is relevance.

And right now you are failing, clearly, as you haven't convinced anyone here that your message has relevance.

Without establishing relevance, not many of us care to invest the efforts needed to decipher your message to determine if there is any truth to it.

Just being pragmatic about it. We all have to filter out the cranks in our lives or else we'd never get anything done aside from listening to cranks.

Right now your posts in here come across as that of a crank, we are inclined and predisposed to dismiss your arguments.

If you want to change the balance of things then you need to compose yourself and your thoughts, be less like the drunk on the street corner who is claiming the end is near and we better all repent, and present a compelling argument and position that underscores whatever point it is that you think you need to be making here.

Until then we are all going to just be beating our heads on the rock because we haven't a fapping clue what it is you are really driving at, and all of our collective guessing on the matter are not well received by you which suggests to me we have yet to "get" your point.

Relevance! I fail to see any proof in this topic that fits the topic title . I see claims being made and comparisons of AMD cmt and intel ht. The title also puts AMD PHII as = to i7.

The only thing that comes close to being honest here was JFAMDS statement yesterday that its hard to discuss performance without final silly . IPC isn't important according to this thread only sharred resources and x number of modules. I suppose that AMD could get BD out in second Qt . But beings how there is no final silly yet that is looking unlikely. Of course I understand AMD doesn't need to produce cpus on the scale of intel at launch but that being the case either AMD sells to OEMs or to the private sector but not both on ramp up. If BD is = or greater than baby SB at release oems will snatch everthing up. Truth is intel will pull another C2D. And show papa SB results when BD is released.

 

[Mopetar]

The title also puts AMD PHII as = to i7.

Both Phenom II and i7 are fairly wide targets and can refer to parts having four cores all the way to parts having six. Based on the slides associated with those claims, the two chips that BD is being compared against were the Phenom II X6 1100T and the i7 950.

I think you're taking the discussion out of the context in which it was framed and making assumptions based on your own understanding which differs from what everyone else is this thread knows from earlier parts of it.

 

[Dresdenboy]

Never said no such thing . Price = usage. Were discussing performance . As of this moment AMD cpus are way overpriced for the performance . About $160 should be the price of topend AMD desktop cpus.

Somehow this doesn't fit to what I see:

http://techreport.com/articles.x/18799/15

 

[GaiaHunter]

We really ought to be referring to the thing we plug into the socket as simply "the CPU".

Yes processor can be used in the context of referring to the CPU itself, but it is open to easy misinterpretation and fanboy battles royale if we as a CPU community adopt such terminology.

Sure.

 

[AtenRa]

this thread could compete for Nobel prize for Greek philosophy

I would say the last two pages could be from a Greek Parody/Tragedy.

I want the ISSCC presentations for Bulldozer Integer Execution units and OoO Schedulers (Session 4.5 and 4.6), could someone tell me where to find them ???

 

[Nemesis 1]

Both Phenom II and i7 are fairly wide targets and can refer to parts having four cores all the way to parts having six. Based on the slides associated with those claims, the two chips that BD is being compared against were the Phenom II X6 1100T and the i7 950.

I think you're taking the discussion out of the context in which it was framed and making assumptions based on your own understanding which differs from what everyone else is this thread knows from earlier parts of it.

So your comparring AMDs top desktop to not top i7 I assumed based on price . Nice metric this price comparison . So i7 950 is= to AMD1100 T . So SB has been out awhile now hows 1100T do against core II i7 2600K I bet its neck and neck right.

 

[Anexate]

I would say the last two pages could be from a Greek Parody/Tragedy.

I want the ISSCC presentations for Bulldozer Integer Execution units and OoO Schedulers (Session 4.5 and 4.6), could someone tell me where to find them ???

At Pcper (pc perspective) they've made an assessment of Bulldozer from all the available data till now.
It's clear however, that planned(supposed) and final performance can differ from wonder to under-performing once in the hands. (AMD and Intel both had their fair share of "wonders")
It's therefore of little help to wrinkle the last drop out of theory, especially if the available data are more like horoscope than a technical report.

http://www.pcper.com/article.php?aid=1083

 

[Nemesis 1]

Somehow this doesn't fit to what I see:

http://techreport.com/articles.x/18799/15

Ya . I see your chart pretty thing . But wait . OH no . weres the core II i7 2600k results in that chart. BD isn't going against anything on that chart . Its up against baby SB than followed by papa SB . Unless I got it wrong . I believe desktop BD comes befor server is that not so . So when BD is released its goes against baby SB . Than papa BD comes out just befor papa SB . This thread is nice but talking about papa SB being on intels 22nm is more interesting .

It is a nice thread. But someone forgot to add a chart comparring BD to P4s because that all AMD people want to do is apple to grapes . They want to compar zacata to SB now that thats a fail. They keep referring to intel atom instead of Oak trail . Its close but zacate barly wins performance but losses big to intels 3 watts.

The best tho is comparring 6 core AMDs to 4 core intels . Love that price metric as the new halo of what performance is.

 

[AtenRa]

At Pcper (pc perspective) they've made an assessment of Bulldozer from all the available data till now.
It's clear however, that planned(supposed) and final performance can differ from wonder to under-performing once in the hands. (AMD and Intel both had their fair share of "wonders")
It's therefore of little help to wrinkle the last drop out of theory, especially if the available data are more like horoscope than a technical report.

http://www.pcper.com/article.php?aid=1083

thx for the link

 

[busydude]

thx for the link

I read that article yesterday.. and the author still says that CMT takes up about 5% of the die space.

CMT maximizes die space and threading performance seemingly much better than SMT (it scales around 1.8x that of a single core, as compared to 1.3x that using SMT), and CMP (chip multi-processor- each core may not be entirely utilized, and the die cost of replicating entire cores is much higher than in CMP). This balance of performance and die savings is the hallmark of the Bulldozer architecture. AMD has gone through and determined what structures can be shared, and what structures need to be replicated in each module. CMT apparently only increases overall die space by around 5% in a four module unit.

 

[Nemesis 1]

At Pcper (pc perspective) they've made an assessment of Bulldozer from all the available data till now.
It's clear however, that planned(supposed) and final performance can differ from wonder to under-performing once in the hands. (AMD and Intel both had their fair share of "wonders")
It's therefore of little help to wrinkle the last drop out of theory, especially if the available data are more like horoscope than a technical report.

http://www.pcper.com/article.php?aid=1083

Thats was an interesting read. He had me going for a while when talking about standard cells vs. custom . But he did mention at the end that intel is doing the same . But who did it first . He writes it as if it was AMD . AMDs product isn't even released. The power gating was interesting but SB has shown us those results again intel first. The turbo modes are cool for consumers both intel and AMD.

Thanks for the link . good read

 

[HW2050Plus]

You did read the topic title did you not? Which is a joke to start with. BD is 50% faster than core i7 and PH II. I didn't know phII and i7 were = I thought i7 was 25%+ faster than PH II . Hell core II i7 slaps magna cores around thats only 2 intel modules using HT.

It depends on which part of one side you compare to what part of other side. So of course there are always roughly equal performing parts of Phenom II and Core i7. Afaik the Core i7 part which was taken as a base was the Core i7 950. It was not the top Core i7 part. And from Phenom II they took afaik the top part.

In addition it is also dependend on what you use to measure performance. If you use anything else than SPECint/fp numbers your results will be quite inaccurate as a measurement of CPU performance.

 

[Mopetar]

So your comparring AMDs top desktop to not top i7 I assumed based on price . Nice metric this price comparison . So i7 950 is= to AMD1100 T . So SB has been out awhile now hows 1100T do against core II i7 2600K I bet its neck and neck right.

I'm not the one making the comparison. That's what the supposedly leaked performance slides indicated. Also, the two processors are fairly similar in price so it's not a bad comparison.

For what it's worth, there are other slides that show it performing better than a 2600K, but we have no way of knowing if they're legitimate or not. It's all speculation at this point and people attempting to determine how Bulldozer might be able to hit those performance figures based on design, etc.

 

[Nemesis 1]

Both Phenom II and i7 are fairly wide targets and can refer to parts having four cores all the way to parts having six. Based on the slides associated with those claims, the two chips that BD is being compared against were the Phenom II X6 1100T and the i7 950.

I think you're taking the discussion out of the context in which it was framed and making assumptions based on your own understanding which differs from what everyone else is this thread knows from earlier parts of it.

I have read the whole topic. I am not taking anything out of context. BD isn't even on final silly yet. I do recall as should many here befor PH I was released the pretty slides showing PH I beating intel by 40% .
So no I am taking nothing out of context . Did you not see tegra III is faster than C2D according to NV . They showed a benchmark proving it so . But what happened ? Others did an apple to apple comparison were both platiforms ran recompiled bench Mark . What was proven . NV lies as a matter of fact . C2D easily wins.

I have taken nothing out of context . Now show real final silly . If and when baby BD bitch slaps baby SB. and is priced Accordingly Than I will jump on the price performance bandwagon and not until.