PCPER.com: 5 GHZ+ Phenom II Overclock on Dry Ice, 6Ghz on LN2

[Eson]

The 3DMark06 CPU score tells us that the Phenom II aint 66% faster for sure. So tell me...what makes the Phenom II system 66% faster? Its funny enough what to expect from IGP vs IGP...

Since when did Lost Planet use the same 3D engine as 3DMark06? Never.
200% faster in Super-PI doesn't mean its 200% faster in any other program.

[Sparky]

This:


Neither of us are wrong until we know setups.

Right, but what good does arguing about it do? Nothing. Just makes for more random clutter and more fuel to the fire so to speak.

[informal]

Also we don't know if they used "Cave" in Lost Planet which is CPU bound for sure.

http://www.xtremesystems.org/forums/...4&postcount=64

You have probably often read or heard that Snow is GPU bound and cave is CPU bound.

[Hornet331]

Anyone knows Hybrid-Crossfire got expanded to higher graphics cards on the 780GX.

Maybe that would be a possibility for the fps difference, paired with a cartain resolution and game settings it might bring that kind of boost?

[informal]

Anyone knows Hybrid-Crossfire got expanded to higher graphics cards on the 780GX.

Maybe that would be a possibility for the fps difference, paired with a cartain resolution and game settings it might bring that kind of boost?

Nope,AFAIK that is a no go with HD4xxx(was a no go before too).

[Marvin_The_Martian]

Arrrhhh...

So many bold words, is that a sign that what is told is the truth? Or just someone shouting to claim his opinion is superiour?

[Zucker2k]

Of course it ain't 66% faster since CPU score tests CPU.. Doh. Is this so hard to understand??Other tests are actual game tests,but we don't know the settings/res used.As i said i don't know what GPU(s) they used on intel system,maybe it was a specific game setting in which the intel system falls short(look the graphs Eson provided in which Core i7 shows similar numbers).

Informal, I think you fail to see where Ghostbuster and Shintai are coming from. Since the 3dmark '06 CPU test is a different test altogether, it's purpose (by AMD) was to show the CPU power of P2 940. Agree?

Now we also agree that we don't know the testbed for the in-game/platform benchmarks, agree?

So Looking at how close (About 8% difference) the CPU processing power is between P2 and Q9400, how do you account for as much as 66% disparity between testbeds using the two processors, when the CPU test showed at best a 10% difference. Assuming 100% scaling, it should not exceed 10% in-game disparity. Ghostbuster and Shintai are arguing that the massive disparity surely rests with GPU/IGP disparity.

What is your argument. Are you saying dragon accounts for 50% more performance in in-game situations? Or are you insisting P2 is more than 50% more efficient in certain games? In short, what in your opinion is responsible for the huge disparity?

[AbelJemka]

Informal, I think you fail to see where Ghostbuster and Shintai are coming from. Since the 3dmark '06 CPU test is a different test altogether, it's purpose (by AMD) was to show the CPU power of P2 940. Agree?

Now we also agree that we don't know the testbed for the in-game/platform benchmarks, agree?

So Looking at how close (About 8% difference) the CPU processing power is between P2 and Q9400, how do you account for as much as 66% disparity between testbeds using the two processors, when the CPU test showed at best a 10% difference. Assuming 100% scaling, it should not exceed 10% in-game disparity. Ghostbuster and Shintai are arguing that the massive disparity surely rests with GPU/IGP disparity.

What is your argument. Are you saying dragon accounts for 50% more performance in in-game situations? Or are you insisting P2 is more than 50% more efficient in certain games? In short, what in your opinion is responsible for the huge disparity?

http://www.guru3d.com/article/intel-...-965-review/16
Core i7920 3Dmark 06 score 4837
C2QX 9770 3Dmark 06 score 4819
difference = none!

difference = 30%!
Astonished? I'm sure you can scenario where i7 destroy QX in Lost Planet

[Shintai]

http://www.guru3d.com/article/intel-...-965-review/16
Core i7920 3Dmark 06 score 4837
C2QX 9770 3Dmark 06 score 4819
difference = none!

difference = 30%!
Astonished? I'm sure you can scenario where i7 destroy QX in Lost Planet

Turbo mode on/off. SMT on/off?





And 30% is overdoing it. Its 24%

Plus if 66%, Phenom 2 at 3Ghz should rival a i7 965. And the only reason i7 is so high is due to pure bandwidth. And Phenom 2 wont even be close there.

[Zucker2k]

http://www.guru3d.com/article/intel-...-965-review/16
Core i7920 3Dmark 06 score 4837
C2QX 9770 3Dmark 06 score 4819
difference = none!

difference = 30%!
Astonished? I'm sure you can scenario where i7 destroy QX in Lost Planet

It is also not a 4 core/ 4 thread comparison. A game that is able to make use of more than 4 threads will always be faster with Ci7, not to mention bandwidth.

[informal]

What is your argument. Are you saying dragon accounts for 50% more performance in in-game situations? Or are you insisting P2 is more than 50% more efficient in certain games? In short, what in your opinion is responsible for the huge disparity?

I already said what i think.But i will say it again:we don't know the settings/res. used in those games,nor the GPUs in intel system.It may very well be a comparison between competing products(intel Q9400+GTX260) versus Dragon platform.Makes sense since the former is all made from non-AMD parts(intel direct CPU competitor,NV direct GPU competitor) ,while the later is all-AMD based system.For what we know it may even be intel/260GTXSLI versus AMD/4870CF.The settings could very well be high res. gaming with AA+AF where HD4800 excels..The Dragon is the enthusiast platform in the end.
Simply it's a lot of guesswork since we don't know the testbeds.But i can tell you AMD wouldn't compare their enthusiast Dragon system with intel IGP(value) based system.They don't belong to neither the same price range nor the performance range.It doesn't make any sense.

I hope this makes my stand clearer since there is a lot of confusion going around.BTW as a side note,what i've heard (unofficially of course) is that Deneb(aka Phenom II) will be a gamer dream system(from both perf. and perf./$),or to put it in blunt terms it will rock in games providing great/better fps than Core2Q at the same settings software and GPU wise(and we know Core i7 has little if no improvement over Yorkfield in gaming).Of course people would like to play their games at best res. and highest detail levels and still have great min fps and average fps. and with the cost of Deneb based system,you'll likely to get the best bang for your buck with it

[gallag]

I hope this makes my stand clearer since there is a lot of confusion going around.BTW as a side note,what i've heard (unofficially of course) is that Deneb(aka Phenom II) will be a gamer dream system(from both perf. and perf./$),or to put it in blunt terms it will rock in games providing great/better fps than Core2Q at the same settings software and GPU wise(and we know Core i7 has little if no improvement over Yorkfield in gaming).Of course people would like to play their games at best res. and highest detail levels and still have great min fps and average fps. and with the cost of Deneb based system,you'll likely to get the best bang for your buck with it

Dude, You wont get any commission so relax, Instead of more "deneb will beat nehalem in gaming" etc why don't we just wait and see, To be frank its time to put up or shut up.

[informal]

Dude, You wont get any commission so relax, Instead of more "deneb will beat nehalem in gaming" etc why don't we just wait and see, To be frank its time to put up or shut up.

I'm just telling you what i heard.Your "commission" comment is laughable and flaimbait in essence,but i won't bait.You can choose not to respond or put me on your ignore list if you don't want another opinion(in this case mine).And frankly,i'm tired of all of the ** you try to stir here and especially in AMD subforum.You done it oh so many times already that i lost count long time ago.

[justapost]

Found an other interesting comment on the planet3dnow forum.

Original Source

Im direkten Vergleich beziehe ich mich auf einen i7 965 und einen nicht übertakteten Phenom II. Taktfrequenz des letzteren ist leider aufgrund des NDA nicht freigegeben, aber sovie darf ich verratenl: sie lag unter der des 965.

Translated

In the direct comparison I refered to an i7 965 and an non overclocked Phenom II. I can not share the frequency of the later one due to NDA, but I can say it was below 965.

The comment comes from mj who was present in austin, it was related to the quote I posted yesterday.

Lot of discussion had happend meanwhile, in case you missed it. He said an Phenom II made a good figure compared to an i7 965. They ran 7zip wprime farcry and crysis.

Does mean an unknown non overclocked Phenom II (3GHz?) performs like an i7 965 (Turbo/HT unknown, 3.2GHz? QPI/NB speed?) running on higher clocks with benchmarks like wprime, 7zip, farcry and crysis.

[Tony]

My god i love how you guys dissect everything...its a friggin CPU not a cure for some deadly disease etc...LOL

I would love to see you guys in a bar after a few beers, i bet you would be hitting each other with pool cues !!

Can we not wait and see what performance is like? Anything you see now is being squeezed out under the NDA and its never gonna tell the whole story.

I'd lock this thread for a while and order everyone to have a cup of tea and a few rounds of toast while they calm down...LOL

[Motiv]

My god i love how you guys dissect everything...its a friggin CPU not a cure for some deadly disease etc...LOL

I would love to see you guys in a bar after a few beers, i bet you would be hitting each other with pool cues !!

Can we not wait and see what performance is like? Anything you see now is being squeezed out under the NDA and its never gonna tell the whole story.

I'd lock this thread for a while and order everyone to have a cup of tea and a few rounds of toast while they calm down...LOL

Think I'll stick to some bovril and a butter pie at this time of night. Ok maybe the pie is a bit much.

I'm honestly looking forward to this chip, my 939 is on it's last legs and the upgrade I did of a 4870 I did has produce...er...bugger all

Roll on Deneb.

BTW, when is it out December or Jan?

[Shintai]

Think I'll stick to some bovril and a butter pie at this time of night. Ok maybe the pie is a bit much.

I'm honestly looking forward to this chip, my 939 is on it's last legs and the upgrade I did of a 4870 I did has produce...er...bugger all

Roll on Deneb.

BTW, when is it out December or Jan?

January 8th.

[Tony]

Jan 8th/9th I think....Right when CES is in full swing.

[informal]

Found an other interesting comment on the planet3dnow forum.

Original Source


Translated


The comment comes from mj who was present in austin, it was related to the quote I posted yesterday.

Lot of discussion had happend meanwhile, in case you missed it. He said an Phenom II made a good figure compared to an i7 965. They ran 7zip wprime farcry and crysis.

Does mean an unknown non overclocked Phenom II (3GHz?) performs like an i7 965 (Turbo/HT unknown, 3.2GHz?) running on higher clocks with benchmarks like wprime, 7zip, farcry and crysis.

Thanks justapost!
So the journalist says the comment was about non-clocked/stock Phenom 2 versus i965.The only model listed in online shops that could compete with that core i7 is 3Ghz a940.I guess in those 3 apps the results were close call so he made that comment.But averall,i think Phenom 2 will compete with middle core i7 model @ 2.93Ghz (i940).
I'm interested what wprime numbers does a940@3Ghz achieve.7zip is i/o intensive and faster mem. access + higher L3 clock could help a good bit there.As for fc2 and crysis tests the difference was probably small/non-existent since both are mostly gpu limited in normal gaming situations.
Thanks again for sharing the link.

In other news,Hexus is reporting another price cut round fror AMD lineup,about 10%-20% across the board:

AMD announces another round of desktop CPU price cuts
Scott Bicheno - 26 Nov 08, 4:01pm
As promised

AMD sent out a newsletter to its channel partners today confirming the price cuts it predicted when it last adjusted prices at the end of August.

The cuts will be effective as of the start of December but the new prices weren't published in the newsletter.

This is no hindrance for HEXUS.channel, however, as it gives us an excuse to pick up the phone to Raj Suman, product marketing director at giant distie Avnet, to get the word on the street.

Asked what he thought the significance of the move was, Suman said: "This strengthens AMD's position in the mid market as it enables it to offer even better bang4buck."

The prices below represent approximate street prices from Suman in the context of the volume sales with which he operates.



So we're looking at price cuts of 10-20 percent pretty much across the board from AMD. It has consistently stated it is targeting them mid market and it's certainly done that here.

With Intel's Core i7 targeting the enthusiast market AMD clearly sees this as an opportunity to make further inroads into the mainstream desktop space in advance of the launch of its next generation of processor in the new year.

Tell us what you think; do these price cuts make AMD processors more appealing than they did before? Let us know in the HEXUS.community.

[Motiv]

Thanks guys, lays that misinformation to rest.

I'm hoping the Deneb will be a marked improvement and allow easy overclocking and a price/perfomance that will keep it on par with intels setup.

I doubt they will match the top end but they certainly don't look like being low end Intel equivalents anymore on the desktop environment (low to mid+ I suspect).

[Zucker2k]

Good to see price-cuts from AMD too, let the price war begin.

[justapost]

Thanks justapost!
So the journalist says the comment was about non-clocked/stock Phenom 2 versus i965.The only model listed in online shops that could compete with that core i7 is 3Ghz a940.I guess in those 3 apps the results were close call so he made that comment.But averall,i think Phenom 2 will compete with middle core i7 model @ 2.93Ghz (i940).

Must say I assume they ran with turbo and ht off. wrime is getting an >20% boost from HT. Hardspell had showed only 5% increase at 2.3GHz for deneb vs agena. An 9950BE should need 12.5-13sec at 3GHz, that translates into 11.9-12.4sec on an deneb. An i965 with HT off (Turbo?) reached 11,47sec here. He did not say anything new afterall.

[informal]

Must say I assume they ran with turbo and ht off. wrime is getting an >20% boost from HT. Hardspell had showed only 5% increase at 2.3GHz.
An 9950BE should need 12.5-13sec at 3GHz, that translates into 11.9-12.4sec. An i965 with HT off (Turbo) rached 11,47sec here. He did not say anything new afterall.

Wait,there were core i7s there too(at AMD's event)??Or mj is speaking of his own tests?If he speaks of his own core i7 tests,i doubt they would turn off HT and turbo since both are out of the box features and legit.

[justapost]

Wait,there were core i7s there too(at AMD's event)??Or mj is speaking of his own tests?If he speaks of his own core i7 tests,i doubt they would turn off HT and turbo since both are out of the box features and legit.

It was at the amd event.

I grabbed the numbers from that ht4u i posted before.

i965 HT off: 11,47s
i965 HT on: 9,27s

I got 11.7s at 3.2GHz and 14,4s at 2.6GHz, means 12.5s at 3GHz.

Means deneb must be 8% faster to compete with i965 HT off and ~35% faster to compete with i965 HT on. But wait won't 4x1MB fit in denebs bigger L3 cache in opposite to agena?

[JohnJohn]

Guys don´t you think you are a little overhyping deneb? i mean, i would really really really LOVE to see i7 being destroyed by deneb, but there is like 1% possibilities of that happening *snif*

Why don´t we let it be and wait instead of creating higher hopes and in the end deneb will be a flop because it didn´t deliver what it promised?