Intel Shows Off 8-Core 45nm Penryn Benchmarks

[xsbb]

And if they can't?

Running benchmarks to show Intel is superior is something Intel WOULD LOVE!!! At this point...Intel should just run and publish every benchmark and make AMD put up or shut up. If AMD responds and wins a few, Intel will still make plenty of $$$$, or start a price war. If AMD doesn't, free advertisement and just more publicity for the better chip (but, it might cannibalize CNR/KFD sales)

That is the last thing AMD should do!!!

In a time before K7 AMD used sandbagging strategy that made Intele beleve that K7 won't go beyond 600 MHz

In a time before K8 same aproach was used (some of us still remember some preliminary tests of K8 couple of months before finall version. It looked like K8 sudenlly overnight gone form 800 MHz to 2 GHz!).

So far sandbagging proved good for AMD. Thee's no reason to think diferent in a case of K10.

And this intel strategy is so expected. You know guys this thing that these companies are doing is written in the management school books, and it's not hot water!

Only thing different from the past times and now is the pressure form the vastly expanded public on the Net, and Intel knows that and its' using it's army of fanboys to hit up atmosphere!

It's just like in a sports one team is using its position in a game, and public energy in effort of not allowing opposing team counteroffensive.

All this marketing gimmicks is irrelevant. In Q3 when K10 is out we'll se the reall state of things!

Well said. AMD is quiet. Perhaps too quiet and I definitely agree with the "sandbagging" concept.

[[XC] riptide]

Anyone have an idea of overall Power Consumption of the entire system?

Is it really 100W for that 16gig of ram?

[dinos22]

"Any other questions?"

"Yes, FUGGER?"

(laughs heard from the back of the room)

[dinos22]

Makes me want a Penryn bad

you better sell your car

[fhpchris]

My personal rig,

Qx6700 @ 333*10
EVGA 680i'mfreekingslowperclock
OCZ VX2 @ 1066 cas 4-4-4
XP Pro 64 bit

Penryn is 1.07% faster

[red]

http://www.lostcircuits.com/cpu/intel_penryn2/2.shtml

Why is it that Yorkfield is around 25% better than Kentsfield in 9.5 and 10? No SSE4? Only thing I can derive is that Penryn scales better in 10 (though Kentsfield has bigger advantage in 10 vs dual core Penryn), so we can say it's more thread friendly or less bandwidth intensive?

[Nistomax]

I can't wait to see some benching of penryn overclocked. If it overclocks as well as Conroe then we could see some intense speeds and results. I'll probably get one on launch, they are expected in early 2008 correct?

[fhpchris]

http://www.lostcircuits.com/cpu/intel_penryn2/2.shtml

Why is it that Yorkfield is around 25% better than Kentsfield in 9.5 and 10? No SSE4? Only thing I can derive is that Penryn scales better in 10 (though Kentsfield has bigger advantage in 10 vs dual core Penryn), so we can say it's more thread friendly or less bandwidth intensive?

Penryn isnt 25% better, it is at most 5-10% better depending on ram speed used and chipset.

[red]

I meant on a raw basis, the quad Penryn@3.33 is around 25% better than the Kentsfield@2.93 in both benches which would seem strange as I would think Intel would push for SSE4 in a bench that they seem to quite like to show off. Maybe newer revisions of C10 will change that..

[fhpchris]

I meant on a raw basis, the quad Penryn@3.33 is around 25% better than the Kentsfield@2.93 in both benches which would seem strange as I would think Intel would push for SSE4 in a bench that they seem to quite like to show off. Maybe newer revisions of C10 will change that..

That just shows that 400mhz on cpu matters alot in Cinebench! I have no dobut that a Penryn isn't but ~7% faster than a Kents at the same clockspeeds.

You can see in CPUZ that 3333 for me is still stock voltage(although the limit of quad prime stability), so I would be more concerned about if Penryn can bench 3dmark @ 4100 mhz on great air or water (essentially the dual core speeds!)

[Brother Esau]

Indeed, its very hard not to become an Intel fanboy in this era, also very hard to avoid becoming one, at the risk of being accused of being a fanboy of the other side.

Great to see they have silicon for this already...good stuff Fugger

I like amd myself...but you have NO idea how many times I've looked at that E6320 and almost pulled the trigger...the processing power is just too tempting at times X(

Perkam

Hey, yes you are right I finally got a real great Intel Build that works 100% and I am very impressed and retract all of my former statements that I have made against the new Intel Platforms! This current build is awesome and I have yet tio really lean on it and crank up the juice and as of yet I get 3.6ghz on the E6600 and I can only imagine what it would be like with some DDR2 1200MHZ Team XTREME to bad its being sold but Overclocked and yes its that stable and a solid performer finally I guess 3 times the Charm hey gents?


EVGA 680i A1
Core 2 Duo E6600 @ 3.2GHZ X8 multi~ QDR 1600 ~ 1T
Zalman 9700
2gb OCZ DDR2 800 OCZ Plats Rev 2 4-4-4-12 subs 3-16-4-8 7.8us @ 2.1v
Corsair Domminator Ram Cooler
EVGA 7900 GT KO Superclocked
Zalman VF900 LED VGA COOLER
2~80gb Hitachi Sata II Raid 0
PC-Power & Cooling 610w
ThermalTake Armor Extreme Full Tower Case

[Omastar]

Sexy.

And that last picture with the AMD building in the mirror is cute. I see what they did there!

[alfaunits]

That is the last thing AMD should do!!!

Stop with the exclamation marks (stop yelling, it just makes you seem rude).

and it's not hot water!

English term is "spinning wheel", not hot water like in EE.

Only thing different from the past times and now is the pressure form the vastly expanded public on the Net, and Intel knows that and its' using it's army of fanboys to hit up atmosphere!

Speaking of which, Nedjo is a local AMD employee, so pass the salt, please.

[dinos22]

Speaking of which, Nedjo is a local AMD employee, so pass the salt, please.

that explains a thing or two

[Fr3ak]

Whether Penryn is that much faster than C2Q isnt that important for me. But I like the way Intel presents new benches of upcomming CPUs a lot more than AMDs "no benches until NDA is over, which is the day of release". Makes me get the impression that AMD knows how slow their stuff is compared to Intel's. Of course it can the the whole opposite of that, but AMD's "we dont say a word"- attitude might not be the best marketing strategie.

The Penryn benches look promising. Now lets hope it doesnt coldbug that early..

[NapalmV5]

great! but @ 1/8-cores, 8-cores or 80-cores we're still limited by storage speed

caravan/horse vs Formula 1 thats where storage vs all other pc components performance is/was/hopefully not forever

[alfaunits]

Storage is not improving the transfer rates because most of the time it is less important; the seek times are, as soon as we hit mid-budget market (read: even Raptor vs. any other SATA).
If you want speedy storage there are options that surpass any speeds the RAM can handle (24GB/s). You guessed it - the cost is in the millions range.

[nemrod]

I have just undestand that V8 was harpertown. I was fool by the Q6800 and I have missed that was FB-Dimm... I was thinking it was something like 4x4 with ddr2. So they should have done a comparison with dual-xeon board?
So what is really new in this V8? This is only a dual-cpu configuration?
...with penryn Xeon (harpertown)

[Shintai]

I have just undestand that V8 was harpertown. I was fool by the Q6800 and I have missed that was FB-Dimm... I was thinking it was something like 4x4 with ddr2. So they should have done a comparison with dual-xeon board?
So what is really new in this V8? This is only a dual-cpu configuration?
...with penryn Xeon (harpertown)

V8 is currently nothing but a sticker on current workstation setups. Instant product so to say, no development etc needed. But its a glimmer product. Nothing useful, unless you already shopped a workstation and got the sticker as an addon.

[XS Janus]

V8 is currently nothing but a sticker on current workstation setups. Instant product so to say, no development etc needed. But its a glimmer product. Nothing useful, unless you already shopped a workstation and got the sticker as an addon.

Thats right. New mobo just for elite-desktop market and presumably based around the new chipset is what they are marketing with theese numbers!
Hopefully the boards will not be a more expensive than current server boards and that we will see other mobo makers producing it not just Asus.

[Movieman]

V8 is currently nothing but a sticker on current workstation setups. Instant product so to say, no development etc needed. But its a glimmer product. Nothing useful, unless you already shopped a workstation and got the sticker as an addon.

I have to laugh when I read comments like this. Nothing usefull you say?
Computational power beyond any quad single socket gaming or workstation that exists.
I'm telling you straight, you guys have to work with one of these to appreciate just how much power they really have.
Just a little comparison for you to think on:
Over at HWBOT, the highest score I saw on Sisoftsandra math test app 2 months ago was just over 81,000.That was on a Kentsfield on LN2 at just under 5000mhz.
I decided to run it just for laughs to see what would happen.
I mean a dual clover at everyday 24/7 full load speeds. No OS tweaks, nothing special, stock win2K3 server ent 64 bit.
The result: 119,000+

Thats right. New mobo just for elite-desktop market and presumably based around the new chipset is what they are marketing with theese numbers!
Hopefully the boards will not be a more expensive than current server boards and that we will see other mobo makers producing it not just Asus.

Asus really isn't into this as yet.
The best boards seem to be supermicro and tyan so far.
I really wish they were cheaper to build. I'd love to see all you guys with one.
Just incredible machines to work with.

[Iconyu]

Only read the first page and the numbers look good, but one thing bothers me. AMD haven't leaked any benches that show that the K10 is clock for clock faster than the K8. And this is something they should have been able to do, even if only half the revisions in architecture worked.

So unless K10 is nothing but 2 K8's bolted together, what have we seen really?

[Movieman]

Only read the first page and the numbers look good, but one thing bothers me. AMD haven't leaked any benches that show that the K10 is clock for clock faster than the K8. And this is something they should have been able to do, even if only half the revisions in architecture worked.

So unless K10 is nothing but 2 K8's bolted together, what have we seen really?

I think the answer is we really have to wait.
My "guess" is that AMD ran into a problem that they are working on and will fix given some time. That is the only logical conclusion I can come to as if things were all rosy they'd have brought out a dual socket system at at least 2500mhz at Computex to rival what Intel is showing.
We're they the big guy and not the little guy I could see them playing head games but not in their position.
They simply can't afford financially to play games here so I assume there is some sort of problem they are working on.

[XS Janus]

Damn I'm so tempted to build a 2s comp, always have been... But there was almost no excuse for building one untill now.
They are SO irrationaly "needed"
With all the encoding, crunching and bloated software to come, I mean ofcourse!

Must hype myself some more!

[awdrifter]

I agree with that assumption. The AMD slides' perofrmance estimates are probably what they are aiming for, but the results of the cinebench is probably what the current status is, they probably encountered some big issue and haven't been able to fix them.