AMD Bulldozer Eng. Sample leaked, benched

[cdawall]

Best i can come up with. Unless you are tlaking drivers, software age doesn't matter...it either goes faster, slower, or just the same. It's not there will magically appear new versions of EVERYTHING for Bulldozer...

yes and no if they are stuck in cool and quiet thats software and bios not talking thats how some of the chips react to incorrectly read p-states

 

[cadaveca]

incorrectly read p-states

 

[twilyth]

Here are some more benches from Chiphell. Problem is, if you translate the page you can't see the images.

They've added a few other things like Cinebench and Fritz Chess.

http://www.chiphell.com/thread-210890-1-1.html

Edit: OK, I think this page has been up since the 8th. Figured it was worth posting.

edit2: More info - compares 2500k to BD. Also links to benching vids.

edit3:

Super Pi (1m):

AMD FX-8130P: 21.456 sec

Core i5 2500K: 9.345 sec

AID64 CPU Benchmark:

AMD FX-8130P: 16724MB Read/ 10804MB Write/ 21071MB Copy

Core i5 2500K: 22359MB Read/ 20723MB Write/ 23677MB Copy

CineBench R11.5:

AMD FX-8130P: 5.66 Points

Core i5 2500K: 5.79 Points

Something is clearly wrong here.

 

[Wile E]

I was pointing something out I though you didn't realize. Not sure why that is a problem or "irrelevant", but whatever.

Where are you seeing IPC figures?

IPC is a derived value. Instructions per Clock. In other words, their performance in clock vs clock, core vs core against Intel is dismal according to those numbers. It's not big enough of an improvement.

So i was indeed right. The damn,worthless AOC is just bugged!

EDIT:
just saw it, there are vantage leaks out there too (also on this UK page).... these look better surely:
http://wccftech.com/2011/05/30/amd-bulldozer-fx8110-scores-81917-3dmark-vantage-cpu-test-details/
281 steps in CPU test 2 in Vantage is murderous in my opinion...i7 does not even 100,severely OCed

I'd bet money that's with CUDA on. If you want to use vantage to compare cpus, you should use CPU Test 1. It's uses no acceleration. My 980X did 5250 the last time I ran it stock. They skewed the test results on that page. Pretty dirty trick.

I want to point out that the BD's score was 27.5% higher than sandy's but the clock speed was only 11% higher for BD.

I do the math so you don't have to.

See above. They aren't telling the entire story in his link.


All that said, I think these are fake. At least I sure hope so, because the showing in the OP is very poor.

 

[seronx]

Here are some more benches from Chiphell. Problem is, if you translate the page you can't see the images.

They've added a few other things like Cinebench and Fritz Chess.

http://www.chiphell.com/thread-210890-1-1.html

Edit: OK, I think this page has been up since the 8th. Figured it was worth posting.

edit2: More info - compares 2500k to BD. Also links to benching vids.

edit3:

Something is clearly wrong here.

The results are fake

IPC is a derived value. Instructions per Clock. In other words, their performance in clock vs clock, core vs core against Intel is dismal according to those numbers. It's not big enough of an improvement.


I'd bet money that's with CUDA on. If you want to use vantage to compare cpus, you should use CPU Test 1. It's uses no acceleration. My 980X did 5250 the last time I ran it stock. They skewed the test results on that page. Pretty dirty trick.



See above. They aren't telling the entire story in his link.


All that said, I think these are fake. At least I sure hope so, because the showing in the OP is very poor.

You are correct these are fake(not because they are poor showing mainly because it isn't showing the right codewords)

Unknown CPUs in cinebench show up as

AuthenticAMD
GenuineIntel

 

[Heavy_MG]

I find it a good sign that a chip that was never meant for consumer hands, a chip that has problems booting windows, can outperform Sandy in anything.

It definitely makes waiting for real performance scores a lot easier.

Agreed.
The ES B0/B1 stepping that chiphell tested has bad performance because the multiplier is off,so clocks are actually lower ,the turbo core is broken/disabled,and some cache was apparently disabled,there was even word that some cores were shut off to keep the TDP low. It was only designed for motherboard manufacturers to test for compatiblity.
Only so much can improved in a new revision,however hopefully IPC improves with B2.
Also IMO,it's silly to use Super Pi,which is optimized for Intel,to compare Intel to AMD.

 

[seronx]

Agreed.
The ES B0/B1 stepping that chiphell tested has bad performance because the multiplier is off,so clocks are actually lower ,the turbo core is broken/disabled,and some cache was apparently disabled,there was even word that some cores were shut off to keep the TDP low. It was only designed for motherboard manufacturers to test for compatiblity.
Only so much can improved in a new revision,however hopefully IPC improves with B2.
Also IMO,it's silly to use Super Pi,which is optimized for Intel,to compare Intel to AMD.

It's fake

Don't get your eye strings in bundle

You do realize CMT has 80% more throughput in Intel optimized workloads(That is for servers but the same applies to Desktops)((30% more throughput is the lowest amount))

SuperPi is also 1 core so it doesn't use SMT or CMT


If it was real

The most realistic result you will see

is

FX-4XXX - X < 9.345 seconds

i7 2500K - 9.345 seconds

Super Pi uses the x87 ISA(Which is AMD Optimized)
(Reason why AMD has gotten lackluster superpi scores before is because they shot themselves in the foot I don't know how but they purposely reduced x87 performance by half which has been fixed in the Bulldozer Architecture)

 

[hellrazor]

Anybody else notice it's running @ .93 volts?

 

[inferKNOX]

Those might be true for that sample, but I in no way believe that that's the same CPU AMD is going to sell us, so this thread has no point other than to show how fast a punctured bicycle can go.
I have a feeling that Zambezi has been the sort of headache for AMD that Fermi has been for nVidia, or probably worse since AMD has the task of catching up to the competition on top of all it's stresses.

 

[seronx]

Those might be true for that sample, but I in no way believe that that's the same CPU AMD is going to sell us, so this thread has no point other than to show how fast a punctured bicycle can go.

I wouldn't believe anything WCCF whatever publishes

If there is a leak the only place I'll believe it is from

Ars Technica

I have a feeling that Zambezi has been the sort of headache for AMD that Fermi has been for nVidia, or probably worse since AMD has the task of catching up to the competition on top of all it's stresses.

I doubt it....It's mainly they are regretting going Fabless, This year is probably the only year(Full out) they will be able to go 32nm then next year GloFo announces 22nm 4D Transistors lol
(GloFo expects to upgrade most of its Fabs to 20/22nm by 2012(end of 2012) and they are working with AMD to make a new fab process that should rival intels 22nm 3D transistors)

 

[pantherx12]

The super pi scores are in line with phenom 2 scores.

Silly fakers.

 

[user21]

AMD showed their world's 1st slowest 8 core AMD FX processor and still cant get a better bench LOL

 

[Damn_Smooth]

AMD showed their world's 1st slowest 8 core AMD FX processor and still cant get a better bench LOL

AMD didn't show anything. If these are real, they are from a chip that was never supposed to be released to the public and it was used to test compatibility at best.

I'm not sure if you were attempting to troll, but if you were, that was pretty weak.

 

[Heavy_MG]

AMD showed their world's 1st slowest 8 core AMD FX processor and still cant get a better bench LOL

Seriously?
Read above.
The bench test that was "leaked" is a Engineering Sample,a chip only for testing. It has bugs,and possibly certain things missing such as L2 and L3 cache,or the clock speed is capped,so the clock speeds are actually less than what is shown.
AMD doesn't want true benchmarks leaked before the release of FX,thus the possible reason behind the broken L2 & L3 cache and capped clock speeds.
However, there were some performance issues,AMD is currently working on a new revision.

 

[Nesters]

Nice trolling over there, still waiting for Bulldozer, i'm gonna get one. I will support AMD as long as it's reasonable to do so.

 

[twilyth]

I'm not sure if you were attempting to troll, but if you were, that was pretty weak.

What he said. We have pretty high standards around here before we grant anyone the status of "troll". You're really going to have to put your back into it if you expect to make as a troll round these parts.

 

[Horrux]

Nice trolling over there, still waiting for Bulldozer, i'm gonna get one. I will support AMD as long as it's reasonable to do so.

Same here. I think AMD will surprise by offering a very competitive product with BD.

 

[user21]

TO ALL: JUST FOLLOW THE LINK HERE..................

http://www.youtube.com/watch?v=FQtmri400SU

and here

http://www.youtube.com/watch?v=sVlZ6_niyoo

 

[Damn_Smooth]

TO ALL: JUST FOLLOW THE LINK HERE..................

http://www.youtube.com/watch?v=FQtmri400SU

and here

http://www.youtube.com/watch?v=sVlZ6_niyoo

You do realize that they didn't run any benches at E3 right?

I don't see what you're trying to get at here.

 

[user21]

You do realize that they didn't run any benches at E3 right?

I don't see what you're trying to get at here.

and i dont see your getting the idea here! when intel made benches from their prototype they should promising results and so did amd in the past and now you all are talking about an AMD not finalized so y they did those benches anyway ? even the x6 is slower then intel quads, clearly i dont see a point mentioning a prototype and the the real thing is there! lol they didn't showed benches cause they know its slow hehe atleast from intel's extreme i7......if you dig in more you can actually consider the lga2011 platform with quad channel memory support x79 chip and running 8core processors then where would be this BULLDOZER be going ? x6 is slower then intel quad then what can be actually expected from an 8cored one

 

[Dent1]

What has AMD's current Phenom II X6's being slower than Intel's current Sandybridges got to do with AMD Bulldozer?

Secondly.

Why are you basing a argument on leaked benchmarks that could possibly be faked, wrong, inaccurate, doctored or tested without a proper methodology.

Thirdly.

Why are you basing your argument on a prototype. We don't even know if it's a prototype, it could be 100% fake.

 

[St.Alia-Of-The-Knife]

why is the product number blacked out on the pics

however, i changed the contrast and brightness on the 5th pic and noticed that there wasnt any product number so why was it blacked out???

 

[user21]

The education system in Pakistan obviously needs a reform.

What has AMD's current Phenom II X6's being slower than Intel's current Sandybridges got to do with AMD Bulldozer?

Secondly.

Why are you basing a argument on a leaked benchmarks that could possibly be faked, wrong, inaccurate, doctored or tested without a proper methodology.

Thirdly.

Why are you basing your argument on a prototype. We don't even know if it's a prototype, it could be 100% fake.

you think e3 was a joke and it could be 100% fake i think your brain needs a reform here! if the x6 is slow the x8 might not be the best thats a forecast, if your obsessed with something slower then intel live up as you like mate

http://www.fudzilla.com/processors/item/22996-amd-officially-announces-fx-brand
DOES THIS LOOKS FAKE TO YOU ?????


and this is a bonus news
http://www.fudzilla.com/processors/item/22997-amds-zambezi-up-and-running-at-e3

 

[TheMailMan78]

quotes modified post

First off fudzilla? Really?!

Second of all I have been to Pakistan pre 9/11. Not impressed.

 

[Nesters]

Ok, so what's your point?

If Bulldozer was that bad as you want it to be, there would possibly be AMD acquisition underway because shareholders wanted to get some money out of it while they can...

... and recent price cuts on PII lineup would insta-kill Bulldozer.

It's true that Bulldozer is architecture for servers but it also means it's future proof for desktops.
Bulldozer is going at least for entry high-end with S2011 considered as launched for sure.
They still have Llano and Bobcat which both seems to get positive feedback from OEM's.

AMD doesn't need top performer to make money and increase their sales volume.