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

[Elixer]

Because it is company policy not to comment on rumors. I can only talk about facts.

Speaking about facts, I thought you were going on vacation?
Now, if you are on vacation, you can pretty much bet that there is going to be no launch without one of the bigwigs (that is you JF!) around.. so...
And yeah, I know, JF is the server guy, not desktop.

 

[drizek]

Yes, turbo is tied to TDP only so it will be much more consistent. Core count an clock speed are highly correlated. I do not have the frequencies in front of me to know how that carries through to boost frequencies. Every processor will have 3 speeds: base, all core boost and max turbo boost.

Can you clarify the number of cores that will be active on "max turbo boost"? Two cores on the same module? Just one core on one module?

 

[busydude]

Can you clarify the number of cores that will be active on "max turbo boost"? Two cores on the same module? Just one core on one module?

From John's blog entry:

Some of the benefits of AMD Turbo CORE include:

• Up to 500MHz of additional clock speed available with all cores active. This means even with 16 cores active with server workloads, all cores can boost at the same time. For those customers that want to maximize their performance, they now have the tools to do it.
• Even higher boost states available with half of the cores active. We’re not stating exactly how high processors can boost with AMD Turbo CORE, but obviously if there is room for up to 500MHz with all cores active, fewer active cores would obviously mean less power, and more headroom to recapture with AMD Turbo CORE. At launch you will see processors marketed with a base and a maximum frequency, base will reflect the actual clock speed on the processor and max will reflect the highest AMD Turbo CORE state.
• AMD Turbo CORE is deterministic, governed by power draw, not temperature as other competing products are. This means that even in warmer climates you’ll be able to take advantage of that extra headroom if you choose. This helps ensure a max frequency is workload dependent, making it more consistent and repeatable.

I think this is all we know for now.

 

[Cerb]

The CPU also has to make sure that there is data to process and branches correctly predicted. Otherwise it'll have enough bubbles here and there to catch up with decoding. Also limited x86-code ILP, suboptimally filled dispatch windows and some dispatch/execution constraints (1 branch, 2 loads, 1 store, 1 mul, 1 div per dispatch group etc.) will also cause a little stall here and there.

Then I think it's safe to say that what we've got here is failure to communicate. If real code could execute up to X IPC, over a short period, the front end should have >=X IPC (generally to the nearest whole number > X), else it could be a bottleneck to optimized high-ILP code; not that if there are Y execution pipelines, the front end should have Y IPC.

 

[Dresdenboy]

Then I think it's safe to say that what we've got here is failure to communicate. If real code could execute up to X IPC, over a short period, the front end should have >=X IPC (generally to the nearest whole number > X), else it could be a bottleneck to optimized high-ILP code; not that if there are Y execution pipelines, the front end should have Y IPC.

I didn't find one of the nice IPC over time diagrams I have somewhere on my HDD.

Do you have the module concept in mind here? Otherwise delivering more than X decoded ops per cycle while execution only processes them at a maximum rate of X would be no bottleneck.

But you need to consider buffering queues and the huge effect of a single mispredicted branch or a cache miss. For BD both are in the range between 15 and 20 cycles, which means up to 80 decoded x86 ops. In many designs even a correctly predicted branch incurs at least a one cycle bubble -> which could again mean some additional decoded ops if the decoder is decoupled.

 

[Phynaz]

Because it is company policy not to comment on rumors. I can only talk about facts.

Is it a fact that this info is only a unsubstantiated rumor?

Is it a fact the release date you personally stated in this thread still hold?

 

[bryanW1995]

They never said June.. but they said Q2 2011 for Desktop and Q3 for server parts.

Regardless, if we don't have cpus until sept many people will be disappointed. Not me, however, I'm looking at BF anyway. Hopefully they (and SB-E) will have been out long enough by then for me to get a good idea of which one will best handle my DC/gaming requirements.

Speaking about facts, I thought you were going on vacation?
Now, if you are on vacation, you can pretty much bet that there is going to be no launch without one of the bigwigs (that is you JF!) around.. so...
And yeah, I know, JF is the server guy, not desktop.

How many times does he have to say that he's in the server dept? Server has always been Q3, anyway. He's probably getting his vacation out of the way now b/c he knows he'll be very busy later this year.

 

[Nemesis 1]

because you are looking at it from the present point of view. now what happens if and when Bulldozer's design puts ht to shame? it'll pail in comparison, and look like a relatively very poor design decision, that's how. You know, just like how the intel fanboy's claim ht is a better design decision than cores, even though ht shows -5% - +30%. Suppose, AMD's new FlexFP and dual core module design puts ht to shame, which by all accounts appears to be the case with +80% performance boost from the second integer core, will you still claim ht is a better design decision?

no doubt the viral marketing team will

Whats wrong with You? You hate AT and would never link to an AT article. Thats just crazy . AT isn't perfect but he is the best in this business.

Really HT was a bad decision . Man you do not belong in a hardware forum . HT has been around since P4 and has helped Intel maintain its 80% market share threw hammer up to the present. You really need a reality check. Without HT AMD would have been Even further ahead of Intel when hammer was released, Ht saved intels rear. Yet you come into this forum on hardware talking 100% trash .

 

[garagisti]

Whats wrong with You? You hate AT and would never link to an AT article. Thats just crazy . AT isn't perfect but he is the best in this business.

Erm, everyone has their failings... Recently there's more of a lean towards Intel, and that guy wasn't the first one in the world to say that... I've read countless reviews since the days of Athlon over here and i see somewhat of a leaning over to the les blues... I'd not fault Anandtech for that. Everybody has their preferences and so does he. Only human!

...HT has been around since P4 and has helped Intel maintain its 80% market share threw hammer up to the present. You really need a reality check. Without HT AMD would have been Even further ahead of Intel when hammer was released, HT saved Intels rear. Yet you come into this forum on hardware talking 100% trash .

I'll bet that paying Dell and other OEM manufacturers to not sell as many AMD machines when Athlon was much stronger had a lot more to do with marketshare than HT. There is also this issue of Intel Compilers artificially crippling other processors as well. Face it... both companies are going to try and maximize their profits. None gives a crap about what you and i say about them here.

 

[piesquared]

Whats wrong with You? You hate AT and would never link to an AT article. Thats just crazy . AT isn't perfect but he is the best in this business.

Really HT was a bad decision . Man you do not belong in a hardware forum . HT has been around since P4 and has helped Intel maintain its 80% market share threw hammer up to the present. You really need a reality check. Without HT AMD would have been Even further ahead of Intel when hammer was released, Ht saved intels rear. Yet you come into this forum on hardware talking 100% trash .

You seem to be ignoring the real reason intel has maintained it's 80% marketshare for 20+ years. It sure as hell can't be from performance since it has only been the last 5-6 years that they have had the absolute performance crown. Your post is just pure flamebait, since there is no doubt you know the real reason intel maintained that marketshare. And it's not about volume alone, but also name recognition. intel's unlawful schemes made darn sure that wasn't about to happen. But to the point, if SMT scales much worse than CMT, it will indeed look inferior. I can't believe you are using P4's HT as a means to explain how intel maintained it's monopoly...

[edit]

BTW, how many AMD article do you see on this page, and how many intel articles do you see? Before you say, well intel is releasing more stuff! Or, intel has more important stuff to cover, i'll have my list ready of pretty important events and announcments from AMD that Anandtech never bothered to cover.

http://www.anandtech.com/tag/cpus

 

[Nemesis 1]

Its not I who doesn't understand that AMD was FAB constrained . They simply didn't have the Fabs to supply more than 20% of the market at that node. This is a well known fact that AMD fanbois skip over . Everyone on this board is aware of this . Besides if it only cost Intel 1.25 billion Intel will do business as usual.

 

[OCGuy]

BTW, how many AMD article do you see on this page, and how many intel articles do you see? Before you say, well intel is releasing more stuff! Or, intel has more important stuff to cover, i'll have my list ready of pretty important events and announcments from AMD that Anandtech never bothered to cover.

http://www.anandtech.com/tag/cpus

I'm still not sure why our forums are allowed to be brought down to 9/11 Truther level with these conspiracy accusations.....

Why can't you argue the merits of AMD CPUs against the competition, instead of having to turn to conspiracy theories to account for its dismal marketshare?

 

[Nemesis 1]

Erm, everyone has their failings... Recently there's more of a lean towards Intel, and that guy wasn't the first one in the world to say that... I've read countless reviews since the days of Athlon over here and i see somewhat of a leaning over to the les blues... I'd not fault Anandtech for that. Everybody has their preferences and so does he. Only human!



I'll bet that paying Dell and other OEM manufacturers to not sell as many AMD machines when Athlon was much stronger had a lot more to do with marketshare than HT. There is also this issue of Intel Compilers artificially crippling other processors as well. Face it... both companies are going to try and maximize their profits. None gives a crap about what you and i say about them here.

1) One thing in life we was must all learn or fail . That is not to repeat the same mistakes . I thought the same as the person that post adderessed . Back when hammer was King. I to accussed AT of bias. Because he was constantly praising AMD and slamming intel and the performance gap was much closer than present Intel won multi media all the time . BUT lost gaming . So I screamed bias. When conroe appeared AT talked up Intel and hammered a bit on AMD . I seen I was wrong about AT . Its a lesson learned and not forgotten .

Your point 2 is meaningless. AMD was FAB CONSTRAINED if intel didn't give breaks for their cpus the only thing that would have changed would have been higher pricies for intel products . AMD was already selling every cpu the fabs could produce . These fines were levied on intel even tho the people setting in judgement knew AMD couldn't sell anymore Cpus than they could supply . Fabb 2 was a day late and a dollar short.

 

[jimbo75]

If AMD was fab constrained don't you think they'd have stopped selling everything else so they could concentrate with the much higher margin server cpu's?

And if they were that fab constrained why did Otellini spend so many $billions bribing MD? Surely he could just have waited until AMD ran out of production for the same result and saved himself a fortune?

 

[Nemesis 1]

I added this part seperately . Intels compilers are Intels . Intel does not have to make AMD cpus run = on there compilers its intels investment its intels property .

That has been resolved now . Intel has to pay for any recompile that companies choose to do for AMD . Thats it . Now intel simplely has to to tell those who use their compilers that they may or may not perform as well on a non intel cpu. Problem resolved . NOW intel has 80% market share . So intel knows full well that their compilers will still be the first choice . Nothing changed other than the warning of non intel cpu usage . Ya its sad . Do ya need a cring towel . I can send ya one.

 

[Nemesis 1]

If AMD was fab constrained don't you think they'd have stopped selling everything else so they could concentrate with the much higher margin server cpu's?

And if they were that fab constrained why did Otellini spend so many $billions bribing MD? Surely he could just have waited until AMD ran out of production for the same result and saved himself a fortune?

This is not even debatable. AMD was selling everthing they could fab . History homework for you . What year did the second german fab come online. EU charges were pure bs. They wanted and got their 1.5 billion because thats what the EU does . Not to worry tho the EU is almost overwith . Praise God.

 

[Schmide]

I added this part seperately . Intels compilers are Intels . Intel does not have to make AMD cpus run = on there compilers its intels investment its intels property .

Duh? Intel licenses the x86,SSE,2,3 etc to AMD. To make them pay for a license then subvert the benefits that come with that license was at the heart of that case. In many opinions that led up to the resolution, what you say about property is not a fair practice.

 

[out.of.order]

Yes, turbo is tied to TDP only so it will be much more consistent. Core count an clock speed are highly correlated. I do not have the frequencies in front of me to know how that carries through to boost frequencies. Every processor will have 3 speeds: base, all core boost and max turbo boost.

sorry jfamd but i don't understand one thing:

the turbo is fixed to the max frequency or is variable at intermediate multiplier settings?

for example:

defaulf frequency: 3ghz
all core boost: 3.5ghz

the all core boost can auto-adjust also at 3.1-3.2-.3.3-3.4ghz for match the tdp or is fixed to 3.5?

 

[Nemesis 1]

Intel doesn't recieve money from AMD for sse 2,3 4 . Amd is free to use sse freely per court decision . Same as intel doesn't pay AMD for AMD64 . even tho it was MS that made that happen . Thats right . AMD/MS is the reason X86 is still around . It hurt intel big . So ya know what intel did . They said screw this. They want X86 OK will give them x86 everywere.Schmide Intel didn't want AMD copying their cpus that what the 1999 lawsuite was about . The court forced intel to share X86 a cpu Intel invented . The only thing intel won in that suite was AMD could not use intels sockets any longer. That was a real winner for the masses . LOL! and it makes the EU decision look like a bunch of retards Intel brought us metal gates and now tri gate and the EU was saying intel was stifling innovation LOL. Intel constantly raising the bar on innovation was somehow ruled stifling. As for the compilers Intel does not have to invest for AMD profit period . AMD had 5 billion dollars to buy ATI . Let them build there own compilers.

 

[drizek]

sorry jfamd but i don't understand one thing:

the turbo is fixed to the max frequency or is variable at intermediate multiplier settings?

for example:

defaulf frequency: 3ghz
all core boost: 3.5ghz

the all core boost can auto-adjust also at 3.1-3.2-.3.3-3.4ghz for match the tdp or is fixed to 3.5?

I think it is fixed, but thats not really how it works since the CPU is going to clock down constantly as well. But when you are maxing one core, it is at 3.5GHz. If you are doing nothing, I'm assuming three modules will be powered down and the remaining one will be clocked/volted down.

 

[Schmide]

Intel doesn't recieve money from AMD for sse 2,3 4 . Amd is free to use sse freely per court decision . Same as intel doesn't pay AMD for AMD64 . even tho it was MS that made that happen . Thats right . AMD/MS is the reason X86 is still around . It hurt intel big . So ya know what intel did . They said screw this. They want X86 OK will give them x86 everywere.

Whether they pay by fungible assets or cross licensing agreements the end result is the same.

We really need to stop derailing this thread.

 

[bryanW1995]

I'm still not sure why our forums are allowed to be brought down to 9/11 Truther level with these conspiracy accusations.....

Why can't you argue the merits of AMD CPUs against the competition, instead of having to turn to conspiracy theories to account for its dismal marketshare?

Are you trying to say that 9/11 wasn't really a plot that abraham lincoln hatched in 1867??? The only "truth" that I've seen lately from AMD is that they are working on a cpu with a radically different design than previouw amd/intel processors. All the recent speculation and lack of info from amd leads me to believe that it won't be as fast as our silent army of longtime amd supporters wishes. Hopefully JF reads this and is allowed to offer us something to dispel this belief.

 

[piesquared]

Its not I who doesn't understand that AMD was FAB constrained . They simply didn't have the Fabs to supply more than 20% of the market at that node. This is a well known fact that AMD fanbois skip over . Everyone on this board is aware of this . Besides if it only cost Intel 1.25 billion Intel will do business as usual.

It was obvious where you were leading the dialogue, which is why I added this:

And it's not about volume alone, but also name recognition. intel's unlawful schemes made darn sure that wasn't about to happen

Of course intel had no control over retail PIB, and why AMD had great channel business. intel's goal was to block AMD from large OEM business' like Dell. Brand recognition was obviously intel's biggest concern, because if nobody here's of them, they are no threat to business. If consumers knew there was a better alternative, you can bet demand would increase, and funding made available for Fab expansion.

last OT from me

 

[Nemesis 1]

I agree . Derailment bad . OK so what do we know to be fact . Is it a fact that AMDs BD(non existant) is 50% faster than intels 6 core unit on the market . Is there proof of this anywere.

Is it true That AMDs BD per graphs, the highend uses 125watts comparred to intels 95watts in the same price range . AMD loves there fan base so they decided to give their Ultra high performance BD to their base at the same price point as 2600K(mid range), . If true I may buy AMD BD.Does anyone recall the price of AMDs lowest priced X2 befor the Conroe release. I believe the lowest priced one was $300 dollars +. But AMD has had a change of heart now and will give away 4 free cores at the same price of intels 4 cores . Truely a wonderful company . I am beginning to see the light now . AMD is truely A Holy Company righting wrong and standing strong with the people . Thank you guys for saving me. God bless you each and everyone in your sincerity

 

[Abwx]

Of course intel had no control over retail PIB

Not at all...
Here, in Europe , Intel went as far as making contracts with
retailers that stipulated that half the advertising comprising
an Intel CPU would be paid by Intel itself , and that the retailer
should no more sell products having another CPÜ brand..

The contract was of course illegal, but at the time the EU
wasnt aware of the matter...