Sign in with
Sign up | Sign in
Your question

Breaking news: K10 benches leaked from Tunisia

Last response: in CPUs
Share
April 24, 2007 10:35:42 AM

http://www.fudzilla.com/index.php?option=com_content&ta...

Could be fake.

All synthetic crap ... but something none the less.

E6400 scores more than twice as much as X2 6000+ in Sandra MM Integer, hard to say how important this bench is in real world performance. E6400 also outperforms dual FX-74's in this bench so 33% more from Barcelona over Kentsfield isnt revolutionary, but 66% MM FP would be impressive if true.
April 24, 2007 10:41:19 AM

Beats it in two test? AMD's chips usually hold well in at least one form of testing. Now the other 9 test they probably are breaking even or are at a slight disadvantage.

Could really go either way yet. Then again, monolithic cores have done better then glued cores in the past, so it could really be going for AMD.

Then again, as said, synthetics and real world often are not close enough to compare.
April 24, 2007 10:43:39 AM

Was posted elsewhere, but yeah synthetic.

Also a QX6700 scores 298,xxx for integer on the toms charts. So a 3ghz would be around 330,xxx.

AMD claimed 20% more than Intel's top part in int and 50%(?) in float.
That would work out about right.

Thing is though as you said, will this equate to anything, it's benchmarks, and there are benches where netburst is still top.
Related resources
Can't find your answer ? Ask !
April 24, 2007 11:26:22 AM

Quote:
http://www.fudzilla.com/index.php?option=com_content&ta...

Could be fake.

All synthetic crap ... but something none the less.

E6400 scores more than twice as much as X2 6000+ in Sandra MM Integer, hard to say how important this bench is in real world performance
. E6400 also outperforms dual FX-74's in this bench so 33% more from Barcelona over Kentsfield isnt revolutionary, but 66% MM FP would be impressive if true.
I agree.If those numbers carry over to other benches...looks good(if this site is believable). I just find that SANDRA numbers don't always jive with reality. I feel that way with all processors in SANDRA.
April 24, 2007 11:39:14 AM

:| (why am I up early, back to sleep)
April 24, 2007 11:39:21 AM

fudzilla does actually seem to be a trustworthy site does it not? yeah, some of them are total FUD but most of them arent. i still dont trust the scores tho... 33% and 66% sound tooo exact to be ... lets say " realistic"
April 24, 2007 11:39:47 AM

Quote:
http://www.fudzilla.com/index.php?option=com_content&ta...

Could be fake.

All synthetic crap ... but something none the less.

E6400 scores more than twice as much as X2 6000+ in Sandra MM Integer, hard to say how important this bench is in real world performance
. E6400 also outperforms dual FX-74's in this bench so 33% more from Barcelona over Kentsfield isnt revolutionary, but 66% MM FP would be impressive if true.
I agree.If those numbers carry over to other benches...looks good(if this site is believable). I just find that SANDRA numbers don't always jive with reality. I feel that way with all processors in SANDRA.

I have a feeling that the K10 architecture has lower diferences between artificial and real life benchmarks as it is the case for K7 and K8.

Anyway 20% more int perf. for 2.5 GHz Barcelona over 3.0 GHz Intel even in syntetics is waaaaaaaaayyyyyyyy more than Conroe had over K8.

After the recent bleed AMD deserves to smash Intel hands down.
April 24, 2007 11:57:18 AM

Sure synthetic benchmarks don't mean much.
But usually, high clock speed inflates synthetic results compared to real world (see Netburst, for example), and memory bandwidth/latency is another important factor which is neglected by number crunching synthetic tests (except those specifically designed to test memory performance).
Barcellona has a lower clock speed than its competitor, and has an overall better memory subsystem, so i wouldn't be too surprised to see it ahead even in real world tests.
But we should wait and see.
April 24, 2007 12:10:25 PM

Quote:
Sure synthetic benchmarks don't mean much.
But usually, high clock speed inflates synthetic results compared to real world
(see Netburst, for example), and memory bandwidth/latency is another important factor which is neglected by number crunching synthetic tests (except those specifically designed to test memory performance).
Barcellona has a lower clock speed than its competitor, and has an overall better memory subsystem, so i wouldn't be too surprised to see it ahead even in real world tests.
But we should wait and see.
Yeah, it's always been that way with SANDRA. As an example, take a 2GHz CPU(either brand), overclock it to 3GHz, and the results will be 1.5 times the original(in other words... clock-speed based), not taking FSB/HT speed, mem speed/timings,etc. into account. :x
April 24, 2007 12:13:34 PM

untill this CPU is in the hands of the true testers (non THG people) like us, these are meaningless.

Get production Samples out to market AMD and fast. Make it known that this is here.

you've got people going "well amd doesnt have anything right now, so i'm buying intel".. thats market share they're bleeding by the day.

AMD

GET PRODUCTION SAMPLES OUT! GET REAL BENCHMARKS. GET EXCITEMENT!
April 24, 2007 12:33:59 PM

I find it "strange" that faud is somehow exempt to the NDA tha tthe rest of the media was forced to follow. I am not a benchmark expert, but aren't the Sandra benchmarks usually labeled with a year? Like... Sandra 2006 or something? *Assuming that is the case* Couldn't this be another 6 year old benchmark being cherry picked to show the "massive" improvement?

It all seems fishy to me. I don't doubt the validity of the article, I just find the circumstances fishy.
April 24, 2007 12:35:06 PM

Hope those results are real ... :lol: 
April 24, 2007 1:07:25 PM

8O LIKE I SAID !!!
April 24, 2007 1:56:49 PM

It comes from FUDzilla. So, its FUD!
April 24, 2007 2:21:13 PM

Thats not benchmarks...
April 24, 2007 2:36:37 PM

Quote:
I find it "strange" that faud is somehow exempt to the NDA tha tthe rest of the media was forced to follow. I am not a benchmark expert, but aren't the Sandra benchmarks usually labeled with a year? Like... Sandra 2006 or something? *Assuming that is the case* Couldn't this be another 6 year old benchmark being cherry picked to show the "massive" improvement?

It all seems fishy to me. I don't doubt the validity of the article, I just find the circumstances fishy.


Photoshop???
April 24, 2007 2:39:35 PM

If the K10 is truly that impressive, I would wonder why AMD has the non disclosure agreement.
April 24, 2007 3:26:55 PM

I do not pay attention to Sandra and other synthetic benchmarks. I want real-world.
April 24, 2007 3:47:49 PM

blah blah blah

People can blow hot air all day long, but numbers in an article don't mean squat. Until a legitimately decent hardware review site gets hold of an actual K10 chip and performs tests, anything else is just plain bullsh*t.

Anybody can post 'K10 is XX% faster than C2D...blah blah blah' and especially coming from a site named fudzilla?? Shall we throw fudzilla in there with the inq?
April 24, 2007 4:00:16 PM

Quote:
blah blah blah

People can blow hot air all day long, but numbers in an article don't mean squat. Until a legitimately decent hardware review site gets hold of an actual K10 chip and performs tests, anything else is just plain bullsh*t.

Anybody can post 'K10 is XX% faster than C2D...blah blah blah' and especially coming from a site named fudzilla?? Shall we throw fudzilla in there with the inq?


As far as I'm concerned Fudzilla=Inq=Sharikou - all crap sites that spew whatever for hits...

I would pose the same question that Jeff_2087 posed - if the processor is THAT impressive then why the NDA? Why all the secrecy when their financials are already in the toilet? Doesn't seem very wise imo....
April 24, 2007 4:51:32 PM

A few days ago Fudzilla posted fake CPUZ shot of Barcelona. I think these numbers are fake. They're very vague and not exact. If they're real why not give exact numbers? 33% and 66% sounds suspicious to me. Like he made it up.
April 24, 2007 5:07:19 PM

Quote:
blah blah blah

People can blow hot air all day long, but numbers in an article don't mean squat. Until a legitimately decent hardware review site gets hold of an actual K10 chip and performs tests, anything else is just plain bullsh*t.

Anybody can post 'K10 is XX% faster than C2D...blah blah blah' and especially coming from a site named fudzilla?? Shall we throw fudzilla in there with the inq?


As far as I'm concerned Fudzilla=Inq=Sharikou - all crap sites that spew whatever for hits...

I would pose the same question that Jeff_2087 posed - if the processor is THAT impressive then why the NDA? Why all the secrecy when their financials are already in the toilet? Doesn't seem very wise imo....Also, if the processor was that strong, why would they give DailyTech a Radeon HD 2900 XT, for them to test on a QX6700?

DailyTech's R600 Benches

Quote:
The tests were conducted on an Intel D975XBX2 BadAxe2, Intel Core 2 Extreme QX6700 and 2x1GB DDR2-800 MHz. The operating system on the test system was Windows XP, with a fresh install before benchmarking each card. Testing of the AMD ATI Radeon HD 2900 XT was performed using the 8.361 Catalyst RC4 drivers, while the GeForce 8800 GTS used ForceWare 158.19 drivers.


It seems like AMD's decision-makers are driving them into the ground, with one bad decision after the other. :x
April 24, 2007 5:14:41 PM

Considering the K10's are probably a Production Models atm and people would want a CPU they can relate to, not to mention most AMD CPUs atm would bottleneck a graphics card like that.
April 24, 2007 5:33:31 PM

Quote:
Considering the K10's are probably a Production Models atm and people would want a CPU they can relate to, not to mention most AMD CPUs atm would bottleneck a graphics card like that.
The general consensus in the Forumz has been that they haven't revealed the R600 numbers until the same time as Barcelona, so as to not have to bench it on Intel equipment.... which seems very logical. :?
April 24, 2007 5:55:42 PM

Quote:
Considering the K10's are probably a Production Models atm and people would want a CPU they can relate to, not to mention most AMD CPUs atm would bottleneck a graphics card like that.
The general consensus in the Forumz has been that they haven't revealed the R600 numbers until the same time as Barcelona, so as to not have to bench it on Intel equipment.... which seems very logical. :?

Yes very logical, now cough up some benches for Barcelona and R600 already.
April 24, 2007 6:28:18 PM

Still inconclusive, I'm going back to work now.
April 24, 2007 6:51:18 PM

This seems a tad fishy, or at least incomplete to me.

AMD is trying to raise, effectively, $1.2 billion in cash.

Morgan Stanley has demanded that they repay $600 million of their debt to them as the price for allowing them to issue $1.8 billion in convertable notes, so clearly their impression is that AMD's finances are not in great shape.

IF AMD did indeed have a Core2Duo Killer up and running and working having this fact widely known would seem to make selling these $1.8 billion worth of notes a whole lot easier.

Maybe I am missing something, but holding back information on a killer product at the same time you are trying to raise $1.8 billion strikes me as logically inconsistent, at best.
a b à CPUs
April 24, 2007 7:02:41 PM

What's all the hub-bub about? That article was a whole lotta nothing! Wasn't even worth clicking the link.

Oh well...
April 24, 2007 7:07:10 PM

Quote:
blah blah blah

People can blow hot air all day long, but numbers in an article don't mean squat. Until a legitimately decent hardware review site gets hold of an actual K10 chip and performs tests, anything else is just plain bullsh*t.

Anybody can post 'K10 is XX% faster than C2D...blah blah blah' and especially coming from a site named fudzilla?? Shall we throw fudzilla in there with the inq?


As far as I'm concerned Fudzilla=Inq=Sharikou - all crap sites that spew whatever for hits...

I would pose the same question that Jeff_2087 posed - if the processor is THAT impressive then why the NDA? Why all the secrecy when their financials are already in the toilet? Doesn't seem very wise imo....Also, if the processor was that strong, why would they give DailyTech a Radeon HD 2900 XT, for them to test on a QX6700?

DailyTech's R600 Benches

Quote:
The tests were conducted on an Intel D975XBX2 BadAxe2, Intel Core 2 Extreme QX6700 and 2x1GB DDR2-800 MHz. The operating system on the test system was Windows XP, with a fresh install before benchmarking each card. Testing of the AMD ATI Radeon HD 2900 XT was performed using the 8.361 Catalyst RC4 drivers, while the GeForce 8800 GTS used ForceWare 158.19 drivers.


It seems like AMD's decision-makers are driving them into the ground, with one bad decision after the other. :x After reading all coments on the r600 as well as the benches, it appears that DT had a one time shot at some benches, and not defending them not being set up to use a Barc, they (AMD) are soon to release to all the intro, so it hadnt been introed yet. Makes a lil sense, since what does 1 day make at this point? It is their plans to soon intro it when they choose..
April 24, 2007 7:15:03 PM

Considering the inq and fudzilla have been quite accurate with R600 info, I find it fairly easy to also consider K10 info accurate. Practically everything there is accurate in fact, although fairly vague in some cases. It's just a matter of reading between the lines to get the story right. In the end, the numbers are usually very close.
April 24, 2007 7:24:49 PM

Quote:
Considering the inq and fudzilla have been quite accurate with R600 info, I find it fairly easy to also consider K10 info accurate. Practically everything there is accurate in fact, although fairly vague in some cases. It's just a matter of reading between the lines to get the story right. In the end, the numbers are usually very close.


And how exactly do YOU know that their R600 info is "accurate", much less the K10?
April 24, 2007 7:38:35 PM

Ill answer that. Time. Over time the archs, and the feature listed there have been acurate, tho like was said, vague but true none the less
April 24, 2007 7:40:49 PM

Quote:
http://www.fudzilla.com/index.php?option=com_content&ta...

Could be fake.

All synthetic crap ... but something none the less.

E6400 scores more than twice as much as X2 6000+ in Sandra MM Integer, hard to say how important this bench is in real world performance
. E6400 also outperforms dual FX-74's in this bench so 33% more from Barcelona over Kentsfield isnt revolutionary, but 66% MM FP would be impressive if true.
I agree.If those numbers carry over to other benches...looks good(if this site is believable). I just find that SANDRA numbers don't always jive with reality. I feel that way with all processors in SANDRA.

I have a feeling that the K10 architecture has lower diferences between artificial and real life benchmarks as it is the case for K7 and K8.

Anyway 20% more int perf. for 2.5 GHz Barcelona over 3.0 GHz Intel even in syntetics is waaaaaaaaayyyyyyyy more than Conroe had over K8.

After the recent bleed AMD deserves to smash Intel hands down.



WAYYYYY more?????????? :roll:

Sorry to say, but go check Sandra SSE test and you'll see WAYYYY :wink: more then 33 to 66% lead from C2D over A64X2. Still, I agree with you that Intel might loose the performances crown to AMD once again.

The real question now is: Will this be enough for AMD to survive NEHALEM storm next year?

One last thing. WHY on Earth would Intel DESERVE to be smashed down by AMD? It's not like Intel has been personally writing in the posts here. They've even been doing a hell of a good job as of late. The good way to say it would be: "AMD now deserve to have a potent cpu to compete against Intel with all the good work they've done in the last few years". That's fair competition on both side, it's this simple. Let's put the fanboyism aside and look atr the facts like they are.

Before I go, I guess that Intel will put all their effort to bring forward NEHALEM if these numbers prove right. Good for us!
April 24, 2007 9:06:37 PM

Quote:
http://www.fudzilla.com/index.php?option=com_content&ta...

Could be fake.

All synthetic crap ... but something none the less.

E6400 scores more than twice as much as X2 6000+ in Sandra MM Integer, hard to say how important this bench is in real world performance
. E6400 also outperforms dual FX-74's in this bench so 33% more from Barcelona over Kentsfield isnt revolutionary, but 66% MM FP would be impressive if true.
I agree.If those numbers carry over to other benches...looks good(if this site is believable). I just find that SANDRA numbers don't always jive with reality. I feel that way with all processors in SANDRA.

I have a feeling that the K10 architecture has lower diferences between artificial and real life benchmarks as it is the case for K7 and K8.

Anyway 20% more int perf. for 2.5 GHz Barcelona over 3.0 GHz Intel even in syntetics is waaaaaaaaayyyyyyyy more than Conroe had over K8.

After the recent bleed AMD deserves to smash Intel hands down.



WAYYYYY more?????????? :roll:

Sorry to say, but go check Sandra SSE test and you'll see WAYYYY :wink: more then 33 to 66% lead from C2D over A64X2. Still, I agree with you that Intel might loose the performances crown to AMD once again.

The real question now is: Will this be enough for AMD to survive NEHALEM storm next year?

One last thing. WHY on Earth would Intel DESERVE to be smashed down by AMD? It's not like Intel has been personally writing in the posts here. They've even been doing a hell of a good job as of late. The good way to say it would be: "AMD now deserve to have a potent cpu to compete against Intel with all the good work they've done in the last few years". That's fair competition on both side, it's this simple. Let's put the fanboyism aside and look atr the facts like they are.

Before I go, I guess that Intel will put all their effort to bring forward NEHALEM if these numbers prove right. Good for us!

Hopefully, because they'll have Shanghai and Montreal to deal with at that time.
April 24, 2007 9:47:37 PM

Quote:
http://www.eetimes.com/news/semi/showArticle.jhtml;jsessionid=BUHGE03WWLO3MQSNDLPSKH0CJUNN2JVN?articleID=199200545


Sorry, but until this is resolved, it means little:

Quote:
New test results based on the SPECcpu2006 benchmarks show AMD's Barcelona processor will have "up to a 50 percent advantage in floating point performance and 20 percent in integer performance" over Intel's high-end quad-core chip running "at the same frequency," according to a statement from AMD.


How exactly does AMD compare systems at the same frequency? Do they take the supposed high end Quad core system, and underclock it? It's all a play on wording of how the test were done. Why not straight up comparasion? If Intel's top dog is running 3GHz, and AMD's is running 2.5GHz, compare those. Why lower one to the same frequency? That's ridiculous. It's like comparing which car is faster, but you have to put a speed limit on one of them first.

Until 3rd party benches arrive, it's all just talk, imo.
April 24, 2007 10:21:26 PM

Yawn,

Sandra is about as useful as your folding@home score or your MAX FPS stareing at a wall in-game. Show me MIN FPS average and typical values charted versus a Q6*** all else the same in the latest HL2. And a few runs converting a AVI to DVD (WinAVI rocks by the way) or converting VOB to XVID/DIVX. I am kinda fond of CPUMARK 2.1 as it measures int and floating operations weighted towards complexity not bufferability.

Oh and because AMD has this issue with disclosure and communication with it's customer base I just ordered my sister's computer: a Dell E520 w/ the Intel E6400, the Intel 965 chipset, and nVidia 7800GS video.

As far as i'm concerned AMD has given no reason to even consider socket AM2 and their silence tells me that a Kuma dual-core is slower than a C2D.
April 24, 2007 10:46:43 PM

Quote:
Until 3rd party benches arrive, it's all just talk, imo.


That's really what it all boils down to.
April 25, 2007 1:00:08 AM

Quote:
http://www.eetimes.com/news/semi/showArticle.jhtml;jsessionid=BUHGE03WWLO3MQSNDLPSKH0CJUNN2JVN?articleID=199200545


Sorry, but until this is resolved, it means little:

Quote:
New test results based on the SPECcpu2006 benchmarks show AMD's Barcelona processor will have "up to a 50 percent advantage in floating point performance and 20 percent in integer performance" over Intel's high-end quad-core chip running "at the same frequency," according to a statement from AMD.


How exactly does AMD compare systems at the same frequency? Do they take the supposed high end Quad core system, and underclock it? It's all a play on wording of how the test were done. Why not straight up comparasion? If Intel's top dog is running 3GHz, and AMD's is running 2.5GHz, compare those. Why lower one to the same frequency? That's ridiculous. It's like comparing which car is faster, but you have to put a speed limit on one of them first.

Until 3rd party benches arrive, it's all just talk, imo.

Ah you caught it too. :wink: Nothing like a little interproguesstulation-slight-of-hand to make the numbers appear bigger.
April 25, 2007 2:03:26 AM

What vote? I didn't vote for anything that I know of.

Vern, I know you're just adding some info, but when AMD uses statements like "at the same frequency", it just doesn't seem like they are being as forthcoming with data as they can be.

Yes, it all looks good. The numbers look good, but when they slip that frequency disclaimer, just how good are the numbers, in reality.

Like I said, it's all talk. Who uses an NDA at a product demonstration, especially when that product has been delayed, and is nearing release? It's those types of behavior that is questionable, not your data adding to the thread.

Sorry if you felt as though I jumped your post. I simply read it, and couldn't get over the whole "at the same frequency" statement.
April 25, 2007 2:22:27 AM

Well, to be honest, I haven't ever used those voting things, but maybe once, to annoy BM by giving Jack 5 stars.

As for the data, yes, it's the same old stuff, but what's even more annoying is that most of the websites that supposedly review or even analyze new products, seem to miss the entire frequency issue.

I find it strange how AMD keeps changing their guidelines on how the best CPU is determined. First it's instructions per clock, then power usage, and now it's at the same frequency. What next? How many laser scribes can fit on the heatspreader? Or the color of pins used? Just bring out your product, get some 3rd party benches, and live with the results.
April 25, 2007 2:40:16 AM

Other than to hope that AMD pulls off what they have been talking about, no.

Competition is good. I don't personally want AMD do go away, since it forces my company to push for more improvements. But part of me thinks that a lot of problems were done by the previous CEO, Barrett, and his insistence that Netburst was the future. Once we dropped Netburst, things started to roll. Hector Ruiz is doing the same thing. Claiming that AMD can squeeze more out of the existing architecture, even though it's being thoroughly trounced.

Will K10 be a great CPU? Beats the living <bleep> out of me. I haven't seen any concrete data or anything else to say yes or no.

The tech sector might want to rally behind AMD, and that's fine and dandy. But how many are really helping by pushing the same thing and hyping up K10, without really seeing it's actual power? Will all the hype and articles about how K10 is great hurt AMD, if it doesn't live up to statements written?
!