AMD Bulldozer Eng. Sample leaked, benched

[cadaveca]

You think X6 CPUs are good?

Read my sig.

 

[Thatguy]

You think X6 CPUs are good?

Read my sig.

Work fine for me on my protools rigs, works as good as my friends sandbridge, Do you belive everything you say ? my results don't add up to your results. Either I have a better optimized system, or your benchs favor intel products. All I know is I can runn 100 tracks with effects on each track just like my buddys brand new sandybridge machine and we both get the same performance.

So, to me dsp loads are kind of where the rubber meets the road. I also use a multithreaded daw and I have run rendering benchmarks. while the SB does outrun the thuban, its not as dramatic as one would expect. A good 5.1 surround mix with alot of effects and alot of instruments, tracks etc can take 30-45 minutes to render to wav. My thuban can build in 37min my buddys SB in about 35 using the same project files setting plugins etc. To me thats a pretty moot point.

 

[HTC]

JF-AMD posted this @ XS:

Based on this conversation I have a feeling nobody here is in the semiconductor business, so let me try to explain it to you.

Engineering samples are designed to validate the design and for partners to validate their systems. They are not meant for benchmarking.

There are multiple steppings of samples, comparing one stepping to another tells you nothing unless you know WHY the stepping changed.

When we make engineering samples, we target the best yield, not the highest performance. Then we have more chips to work with. Typically you get a couple of the highest clock speed anyway, those typically stay in our labs, so it is unlikley that you see those out in the wild.

We don't even do any real performance analysis until we get the final silicon. Prior to that, we are totally focused on design validation.

Performance is determined based on the following:

1. The processor
2. The processor microcode
3. The BIOS
4. The Operating system
5. The drivers
6. The compiler code

So, unless you are looking at final silicon and you have all of the other pieces in place, making a statement about the performance of ES chips is completely pointless.

Which is why we just don't comment on those numbers. They probably aren't real. And if they are, they are not representative of what people will see when they get the real products in their hands.

Imagine making a cake. You start with a pile ingredients. You put it all in a mixing bowl. You stir it up. You put it in the oven. You let it cook. You take it out. You put frosting on it.

You are all looking at bowlful of batter and arguing about whether or not it is appetizing.

 

[Dent1]

No it really doesn't make sense make a badass board for a overclocker to overclock if the cpu is a pile of shit. Whats the point in overclocking a turd ? its like hotrodding a honda, when you could start with a corvette instead.

Marketing.

Motherboard manufacturers need a entry level, mainstream and enthusiast brand of line up. They couldn't care less about how the CPU performs or how well they overclock. If the consumer wants to buy a monster highend and expensive board they'll give them it because the market requests it and hence money is to be made.

 

[TheMailMan78]

You think X6 CPUs are good?

Read my sig.

There is nothing wrong with a 6-core.

 

[cadaveca]

Work fine for me on my protools rigs, works as good as my friends sandbridge, Do you belive everything you say ? my results don't add up to your results. Either I have a better optimized system, or your benchs favor intel products. All I know is I can runn 100 tracks with effects on each track just like my buddys brand new sandybridge machine and we both get the same performance.

So, to me dsp loads are kind of where the rubber meets the road. I also use a multithreaded daw and I have run rendering benchmarks. while the SB does outrun the thuban, its not as dramatic as one would expect. A good 5.1 surround mix with alot of effects and alot of instruments, tracks etc can take 30-45 minutes to render to wav. My thuban can build in 37min my buddys SB in about 35 using the same project files setting plugins etc. To me thats a pretty moot point.

It's not just about performance...to me, it's all about performance per watt.


Intel TDP is maximum load, while AMD TDP is "typical" load.

So, for example, just at stock, my 1100T pulls 150W. That's a fair bit more than the 125W my CPU is rated for.

My 2600K pulls 65W. So running 2x 2600K uses the same power from the wall as one X6. So yeah, I do not think X6 chips are that good, because at stock, Intel chips are a bit faster, so really, I get more than 2x the perforamcne boost via Intel.

Of course, I'm comparing 45nm AMD vs 32nm Intel. So no, it's not fair, exactly, but it is what it is. AMD's 32nm chips are expected to be 125W CPUs(again, typical load, not maximum), while Intel's is 95W.


To be honest, I don't really like Intel. I was NOT going to buy into P67 platform, but had to buy in to be able to do my job here. And I'm glad I did.

But, at hte same time, I'm an ATI/AMD fanboy, so I am very much interested in Bulldozer...

So consider, not only raw CPU grunt..consider how much power that grunt requires, too. For me, it's not JUST performance.

There is nothing wrong with a 6-core.

See the above.

 

[TheMailMan78]

It's not just about performance...to me, it's all about performance per watt.


Intel TDP is maximum load, while AMD TPU is "typical" load.

So, for example, just at stock, my 1100T pulls 150W. That's a fair bit more than the 125W my CPU is rated for.

My 2600K pulls 65W. So running 2x 2600K uses the same power from the wall as one X6. So yeah, I do not think X6 chips are that good, because at stock, Intel chips are a bit faster, so really, I get more than 2x the perforamcne boost via Intel.

Of course, I'm comparing 45nm AMD vs 32nm Intel. So no, it's not fair, exactly, but it is what it is. AMD's 32nm chips are expected to be 125W CPUs(again, typical load, not maximum), while Intel's is 95W.


To be honest, I don't really like Intel. I was NOT going to buy into P67 platform, but had to buy in to be able to do my job here. And I'm glad I did.

But, at hte same time, I'm an ATI/AMD fanboy, so I am very much interested in Bulldozer...

So consider, not only raw CPU grunt..consider how much power that grunt requires, too. For me, it's not JUST performance.



See the above.

You're just a performance per watt homo.

 

[cadaveca]

You're just a performance per watt homo.

Well you know, other than memory performance, that's exactly what I will be looking at for Bulldozer.


Memory is #1. If that is taken care of, then I'll look at how much power that performance requires. It's doesn't have to BEAT Intel, but it better damn well be close.


If bulldozer is faster, than I'll take that into consideration too.

 

[TheMailMan78]

Well you know, other than memory performance, that's exactly what I will be looking at for Bulldozer.


Memory is #1. If that is taken care of, then I'll look at how much power that performance requires. It's doesn't have to BEAT Intel, but it better damn well be close.


If bulldozer is faster, than I'll take that into consideration too.

I agree. I'm stoked about the new IMC.

 

[cadaveca]

I agree. I'm stoked about the new IMC.

Yeah, but you are one of few people that heard me bashing P67 in TS before it came out, and saying there was no way I was going that route, so you KNOW I'm not an Intel fanboy, and I was forced to accept that P67 IS good.


As far as I am concerned any leaked benches, plus the info from JF-AMD, says that the INC and cache speed is exactly what is lacking in the ES chips. Lower ram speed and cache greatly reduces overall PCU heat, and they did very much say that current ES chips weren't binned for anything other than pure functionality, not clockspeed.

So there's basic working parts in those chips, but no more. I think maybe that TurboCore isn't working, and only one core actually goes over like 1200 Mhz

 

[Heavy_MG]

There is nothing wrong with a 6-core.

Exactly.

It's not just about performance...to me, it's all about performance per watt.


Intel TDP is maximum load, while AMD TDP is "typical" load.

So, for example, just at stock, my 1100T pulls 150W. That's a fair bit more than the 125W my CPU is rated for.

My 2600K pulls 65W. So running 2x 2600K uses the same power from the wall as one X6. So yeah, I do not think X6 chips are that good, because at stock, Intel chips are a bit faster, so really, I get more than 2x the perforamcne boost via Intel.

Of course, I'm comparing 45nm AMD vs 32nm Intel. So no, it's not fair, exactly, but it is what it is. AMD's 32nm chips are expected to be 125W CPUs(again, typical load, not maximum), while Intel's is 95W.

I fail to see how both TDP wouldn't be "Maximum load".
Do you have proof that a 1100T actually pulls 150 watts?
Tools such as CPU-Z and HW monitor aren't accurate for measuring actual TDP.
If it actually used more than the rated 125W.,many would have fried VRM's because most motherboards are only rated to 140 watts,overclocking would easily push it upwards of 190W. Unless you're talking about performance per watt,you are not getting a "2x performance boost" with a 2600K. Intel CPU have usually had better performance per watt,except for the Athlon 64 series.

 

[cadaveca]

Exactly.
I fail to see how both TDP wouldn't be "Maximum load".
Do you have proof that a 1100T actually pulls 150 watts?
Tools such as CPU-Z and HW monitor aren't accurate for measuring actual TDP.
If it actually used more than the rated 125W.,many would have fried VRM's.

Yes, I can fire up a 9-series based rig and show you numbers on a meter on the 8-pin CPU power line. I use this testing for my motherboard reviews, to check VRM efficiency from one board to the next.


I DO NOT use software to measure power consumption.

You'll also note, that if you check my ASUS M5A97 EVO review that I managed to pull near 300W through the $100 board's VRM.

If you want more info on TDP, and the differences in how AMD and Intel rate it, this topic has been discussed widely over the years, so I'll let you research that yourself.

Start @ Nigel Dessau's blog @ AMD.

It is not the first time AMD has tried to convince the world its ACP measurement (or 'fake-a-watt' as we here at the INQ fondly call it) is the way to go, but after reading Nigel's blog, we decided the discussion needed some INQput.

Several processor architectures ago, AMD and rival Intel used the same methods for calculating Thermal Design Power with regard to microprocessors. From an engineering standpoint, the TDP represents the amount of power the cooling mechanism for the CPU must dissipate before failure.

AMD and Intel now differ with TDP calculations, and for different reasons. Intel's current architecture, for example, allows the CPU to exceed the TDP rating for a small period of time before the processor throttles its frequency clock in order to reduce the temperature at the processor level. AMD's current-generation processors do not practice this method, and thus AMD intentionally publishes conservative TDP ratings.

Here's an INtel whitepaper on how thier measurements differ from AMD's(dated April 2011):

http://www.intel.com/performance/resources/briefs/tdpvacp.pdf


AMD says:

TDP is not the maximum power of the processor.

http://support.amd.com/us/Processor_TechDocs/43374.pdf

(Page 80, Item #7)


It's just not that often that someone points this out.

 

[Samdbugman]

i hope that the new bulldozer totally spanks i-7, hell i hope it spanks i-9. that means intel has to lower prices, amd has to keep them down, and intel has to beat amd next year,and amd has to beat intel next year. it means keeping cpu's in a price range i can afford. i am not a fan boy, i have just bought my builds over the years that have favored amd over intel for bang for the buck.

 

[trickson]

Now I know this is just BS ! I ran the Fritz Chess BM and well my score is higher than the score of an 8 Core BD ! Some thing is just not right here ! It would also seem that my Q9650 is killing it in Cinebench as well ! And in Super PI

 

[Thatguy]

Well you know, other than memory performance, that's exactly what I will be looking at for Bulldozer.


Memory is #1. If that is taken care of, then I'll look at how much power that performance requires. It's doesn't have to BEAT Intel, but it better damn well be close.


If bulldozer is faster, than I'll take that into consideration too.

Maybe you give a crap about per/watt. I really don't. By design bulldozer should have pretty good perf/watt as thats one of the reasons to have more cores/vrs hyperthreading. Plus it should have better overall thermal spread as well.

But if the difference is say 10 watts and equal or better performance. I really don't give a shit. My flashlight charger pulls more power.

If you want a real perf/watt comparison add in process node sizing to the mix and you'll see AMD does pretty good here.

 

[cadaveca]

And sure, that's a legit perspective, 100%. I mean, I did already mention that each product line is on a differnt node already. It's NOT fair, but it's what's on the market.

However, for myself, I'm left really comparing these things, considering all aspects, as that's my job as a reviewer. It might not be important to you, but it may be important to some of my other readers, so it's something I HAVE to look at.

I'm trying really hard to keep my opinion out of this, and sticking to the facts. And the facts say that AMD CPUs regularily consume more power than TDP, and give less performance/watt compared to other products available for purchase today. Of course, when Bulldozer launches, this can change, but I do NOT expect AMD's TDP consideration to chage...a 125W CPU will, under some workloads, consume more power than TDP...that's just how AMD rates thier CPUs.

 

[Thatguy]

And sure, that's a legit perspective, 100%. I mean, I did already mention that each product line is on a differnt node already. It's NOT fair, but it's what's on the market.

However, for myself, I'm left really comparing these things, considering all aspects, as that's my job as a reviewer. It might not be important to you, but it may be important to some of my other readers, so it's something I HAVE to look at.

I'm trying really hard to keep my opinion out of this, and sticking to the facts. And the facts say that AMD CPUs regularily consume more power than TDP, and give less performance/watt compared to other products available for purchase today.

Flatten the sizing node and look at competing intel products of equal performance really does give a more fair comparison. So what if they pull a bit more. Who cares, its really only a factor with a overclock anyways. Also people look at TDP and power consumption the right way. Intel is using a fixed unit, amd is using a epa milage estimate based on use.

 

[Heavy_MG]

Flatten the sizing node and look at competing intel products of equal performance really does give a more fair comparison. So what if they pull a bit more. Who cares, its really only a factor with a overclock anyways. Also people look at TDP and power consumption the right way. Intel is using a fixed unit, amd is using a epa milage estimate based on use.

True. For the size AMD is using vs. what Intel is using,it is not that bad. You cannot ignore the difference in design,the differences mean different TDP. Both architectures are completely different anyway,it's like trying to compare a Nissan to a Toyota. AMD's TDP rating based on real use makes more sense,90% of users most likely don't go over the rated TDP.

Yeah, and you know, I have compared power consumption between 45nm products. Intel still wins there, even when overclocked. Intel has higher clocking overhead due to thier lower power consumption.


And that's why, IMHO, in the extreme scene, we see very few people are clocking AMD chips...board limitations brought on by CPU power consumption means that the gains just aren't wprth the effort.

Like don't get me wrong, I used to call myself ATI's #1 fanboy. ATI doesn't exist any more, so now my main concerns are power consumption, and best performance.

Obviously.
Plenty of people of overclock AMD chips,you just don't go and buy a $50 board and expect any overclock from it. I have a $120 4+1 phase board,with a 3.8Ghz overclock,there is no way it could handle a 300W. load.
ATI still exists,you must not have heard about the name change,ATI has still existed even after AMD bought them out,the only difference is the name on the card. Judging from you're saying that the X6's are awful,and not realizing the differences in architecture you are Intel biased and try to sway your readers in that direction as well.

 

[cadaveca]

Yeah, and you know, I have compared power consumption between 45nm products. Intel still wins there, even when overclocked. Intel has higher clocking overhead due to thier lower power consumption. My i7 870 draws 145w@ 4 GHz, my 1100T draws a bit over 300W.


And that's why, IMHO, in the extreme scene, we see very few people are clocking AMD chips...board limitations brought on by CPU power consumption means that the gains just aren't wprth the effort.

Like don't get me wrong, I used to call myself ATI's #1 fanboy. ATI doesn't exist any more, so now my main concerns are power consumption, and best performance. Differences in design doesn't matter...it's a set workload, that draws variable power based on platform, and completes in a variable time, based on performance.

You may not like that compare, but it's still definitely valid, especially when it come to the office environment.

 

[erocker]

We're talking about the "extreme scene?" TDP doesn't factor in my decision at all. I doubt if factors much into anyone's decision. SB is a more powerful (performance) chip though, that's what people look at. As far as the office environment it doesn't matter. People will buy in lots at a good price, performance plays less of a factor.

 

[trickson]

We're talking about the "extreme scene?" TDP doesn't factor in my decision at all. I doubt if factors much into anyone's decision. SB is a more powerful (performance) chip though, that's what people look at. As far as the office environment it doesn't matter. People will buy in lots at a good price, performance plays less of a factor.

I would agree . And by the looks of the leaked BM it is an under performing CPU at best . But I must say that the leaked BM are BS . Price is a big factor and then performance the power consumption is last . I do not see people or businesses buying CPU's that are all about power consumption .

 

[TheMailMan78]

We're talking about the "extreme scene?" TDP doesn't factor in my decision at all. I doubt if factors much into anyone's decision. SB is a more powerful (performance) chip though, that's what people look at. As far as the offic environment it doesn't matter. People will buy in lots at a good price, performance plays less of a factor.

This is true. Most offices would be over powered with an Athlon II. E-mail and some flash websites don't need a Phenom II. Nevermind a Sandy.

Raw power=Intel
Budget power=AMD

I mean its pretty simple.

Anyway we all sound like some old crows bickering about a CPU thats not even out yet. Lets just wait for some benches mkay?

 

[erocker]

This is true. Most offices would be over powered with an Athlon II. E-mail and some flash websites don't need a Phenom II. Nevermind a Sandy.

Raw power=Intel
Budget power=AMD

I mean its pretty simple.

A friend of mine who leads the IT department at a very large national law firm just bought 400 computers, all with AMD 1045 x6's in them.

 

[TheMailMan78]

A friend of mine who leads the IT department at a very large national law firm just bought 400 computers, all with AMD 1045 x6's in them.

lol overkill much?

 

[trickson]

lol overkill much?

LOL well it is a law firm after all .