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

[Monkeywoman]

anyone seen these before? http://markbench.blogspot.com/2008/1...ith-deneb.html and http://markbench.blogspot.com/2008/1...ead-deneb.html. looks salty

[G0ldBr1ck]

anyone seen these before? http://markbench.blogspot.com/2008/1...ith-deneb.html and http://markbench.blogspot.com/2008/1...ead-deneb.html. looks salty

Wow I hope those are real.

[Rammsteiner]

3610Mhz is also over 3,6ghz.

as long as he dont specifies what "over 3,6Ghz" means, it could be just a few or serveral hundred Mhz more .

But now dont focus on the 'over 3.6Ghz' part, but more like on the 'in 30 minutes' part? I mean, no one has a fully tweaked and OC'd system in 30 minutes, even my parents know that while they know nothing about it in the first place.

So if you got in 30 minutes, over or not, 3.6Ghz OC Prime95 stable, you dont think there's a huge possibility, depending on current temps and Vcore, that he might as well get another 200Mhz? Or even another 400Mhz in that matter?

I dont know what kind of super skills you've, but I never OC'd a system to the max, or 24/7 durable, in 30 minutes.

It'd hurt AMD

When is the NDA lifted so we can see the real thing and not this damn speculation and useless blabla? Hey I SWEAR I have a 10GHz i7! ON AIR! 1,6v is NOTHING. But I won't let you see it beause I'm under my own NDA, instead I'll tease you a bit. Are you jealous?

They better release some real evidences of this soon.

Im sorry for you you got to make a post like that. How exactly was Nehalem not overhyped? There were many people claiming 4Ghz was easy on Bloomfield as well, it wouldnt be expensive etc. Well, 4Ghz ain't an average obviously and it is expensive, considering you get some real parts though.

I mean, NDA is there for a reason, grow up. They can either break the NDA and become in trouble right away or never get a preview like this again. At some point I'd rather stick to such rules tbh.

Trust me, nobody's "crying" over i7. No need to be sorry, i7 is already quite attractive as it is. Let's see what has it been now a week and half since reelase? We've already seen 5.3Ghz OC's, 4.3GHz OC's on air? Yep, and it will only get better as people tweak and learn more.

The same thing will go for AMD's offering when they release it. It will get better as time goes on, but as far as numbers...

Read the above. Just like a single 6.3Ghz LN2 Phenom II run ain't saying everything about Phenom II in general, so is a 4.3Ghz OC on air also no average.

Besides that, Intel managed to get the CPU's on time in shops, now the motherboard manufacturers have not. Thus far I didnt see anything for my self

[Ghostbuster]

anyone seen these before? http://markbench.blogspot.com/2008/1...ith-deneb.html and http://markbench.blogspot.com/2008/1...ead-deneb.html. looks salty

Yups its truly salty... Did you see their "latest" post from "September" (the blog was just created this month November!)? This one http://markbench.blogspot.com/2008_09_01_archive.html Does those pictures look familiar? It should.. The picture processor was clearly taken from OCP, they just crop it to hide the watermark.. The CPUZ came from that single and only post at http://www.overclock.net/4285114-post9.html (which later turned up at AcesHardware)

Original picture

[Shintai]

^^ Yup

Your nick should be dreambuster

[Tony]

In the harsh times of the credit crunch the bulk of end users will go for performance/cost over ultimate performance. I see this happening more and more....this is where AMD are aiming this new CPU.They did the same with the 4XXX GPU cards and its working just fine for them as we all know.

I have i7 965XE, i will hopefully soon have Phenem2 940, going on forecast/present pricing would i be able to afford the i7 965?...sorry lads no I would not, but I can afford the 940 and if it clocks like the reports are saying then i would probably have a lot of fun playing with it.

Just to remind you guys the Extreme performance market is tiny, some of the top end boards have total sales less than 200 pieces world wide... If you keep this in mind and look at the price of i7 boards you are going to see in the bigger picture... i7 is to expensive for the average end user to buy. Even those looking to get the lower end models and clock them up still have to pay out $$$ on the boards and probably upgrade to DDR3.

I have a feeling Intel's pricing will change massively over the next few weeks now, i can't see $1500 processors staying at that price and boards will come down massively also.

Competition is good don't you think ?

BTW no insider info here, just my personal opinion

[mongoled]

^^ Yup

Your nick should be dreambuster

^^and your nickname sjould be AMD HATER, now was my post needed

** OF COURSE NOT **

so why did u have to post such a snide comment in an AMD thread??

Just as the thread is coming back on track, why are u trying to derail it.

cant u just let things be?

If peoples hopes are smashed, does this make you feel good? Give u a *ardon? Or just inflate your ego??

Your post has been reported, quit trolling

[Shintai]

I guess you prefer the fake info in the blog. And sofar you are the only one derailing and trolling with personal issues.

[Ghostbuster]

Also look at the CPU Queens results, its way off.. Compare to..



from http://reviews.zdnet.co.uk/hardware/...57885-4,00.htm

Single core = 4977

Lets's assume perfect clock scaling.. from 2.7GHz to 3Ghz...

Hypothethically at 3GHz single core = (4977 x 3) / 2.7 = 5530

Let's assume perfect 100% performance scaling from 1 core to 4 cores (no performance penalty)...

Hypothethically at 3GHz 4 cores = 5530 x 4 = 22120

Still falls short of that score... 23466

Furthermore, there is no such thing as perfect performance scaling... If you noticed from the ZDNet's score, going from 1 core to 8 core performance didn't scale 100% per core.. more like 64.6%.

[Shintai]

^^ The X7460 would be 24 cores and not 12. Else yes

Also fits quite ok with the 2/3 scaling.

[justapost]

A 9950BE should get round 18700 pts at 3GHz. That would mean ~25% IPC increasement, sounds abit unrealistic. I got 16180 pts at 2.6GHz and 19900 at 3.2GHz that's an 23% performance increase with an 30% frequency increase.

[Chad Boga]

I have a feeling Intel's pricing will change massively over the next few weeks now, i can't see $1500 processors staying at that price and boards will come down massively also.

Competition is good don't you think ?

Hell Yeah!!!!!!

[Ghostbuster]

I'm looking at the Shanghai scores..

CPU Queens is very small, about 1MB and should fit in the cache easily. Thus, I think NUMA should not affect it much.

[STaRGaZeR]

You feel the need, you and others to come and save us from "swallowing the FUD"?
It seems you really underestimate XS members's capacity to separate facts and speculations.

Nobody is trying to save you, I'm trying to save myself from clicking a thread and find just

So far nobody has talked about perfomance. Who cares if it can clock to 4GHz if perfomance is equivalent to a 400MHz less C2Q. I guess we won't have any proven info about that until we reach the launch day (that doesn't include Mark's blog and similar, just in case).

[Calmatory]

In two words: get real.

...Somehow I remember that from the ones who claimed that RV770 has 480 SP. AMD wants to do the same again. "OMG 5 GHz on AMD!" two months later "OMG +6,5 GHz ON AMD!! ". They quite failed. 6 GHz is a very nice result anyway.

[Macadamia]

^^ The X7460 would be 24 cores and not 12. Else yes

Also fits quite ok with the 2/3 scaling.

No, it's 12 (twelve) cores.

an Intel four-processor board with the 7300 'Caneland' chipset and two Xeon X7460 'Dunnington' processors, each with six 2.66GHz cores. The test system had a single SATA disk.

So what's with the crazy performance delta?

[AbelJemka]

Nobody is trying to save you, I'm trying to save myself from clicking a thread and find just

So if u wanna save yourself just don't click and don't post
If u just post to say "wait launch day, X or Y spreadind FUD", that is surely increase your post count and...that'all

[Calmatory]

but why it took so long no one knows.

Their 45nm process is so much better than their 65nm process, and the huge core was a tad too big slice of the pizza for the 65nm, even Intel said that they wouldn't have done it on 65nm. Now the 45 nm seems to work fine.

So far nobody has talked about perfomance. Who cares if it can clock to 4GHz if perfomance is equivalent to a 400MHz less C2Q. I guess we won't have any proven info about that until we reach the launch day (that doesn't include Mark's blog and similar, just in case).

Take Agena results. Then multiply them by 1.05-1.1. There is your theoretical Deneb result with the 5-10 % IPC improvement over Agena. 3 GHz Agena scores 100 points in test X. How does 4 GHz Phenom II perform? 100 * 1.333 * (usually take something between 1.05 and 1.1). E.g. 100 * 1.3333 * 1.07 * 0.96(scaling factor) = 136.95. Thats 36.95 % faster than 3 GHz Agena. What about 2.6 GHz Agena? 100*0.86666*1.04(scaling factor, inefficiency from going from 2.6 to 3.) = 90.

However, that assumes that NB clock scales with the core clock which it does not AFAIK. But that should be roughly estimate.

You knew this already, I know it.

[Shintai]

No, it's 12 (twelve) cores.



So what's with the crazy performance delta?

I really think ZDnet simply was mistaken. Else it would be because of the 16MB cache? But I bet ZDnet is just mistaken and mixed things up from the test to article part.

[Macadamia]

I really think ZDnet simply was mistaken. Else it would be because of the 16MB cache? But I bet ZDnet is just mistaken and mixed things up from the test to article part.

I don't think the power consumption is indicative of FOUR X-rated Dunningtons.

[Chad Boga]

Their 45nm process is so much better than their 65nm process,

Not that it is going to make any great difference to the finished product either way, but I wonder how much of Deneb's improvements are due to the process and how much are due to fixing issues which plagued Barcelona?

Barcelona was quite a rushed design afterall.

[Zucker2k]

No, it's 12 (twelve) cores.



So what's with the crazy performance delta?

Originally Posted by ZDNet
an Intel four-processor board with the 7300 'Caneland' chipset and two Xeon X7460 'Dunnington' processors, each with six 2.66GHz cores. The test system had a single SATA disk.

If it's a four-processor board then 4x X7460s = 24 cores?

[Macadamia]

If it's a four-processor board then 4x X7460s = 24 cores?

I'm certain you can put 2 cores in a 4-way board. Isn't that hard after all.

[Shintai]

I don't think the power consumption is indicative of FOUR X-rated Dunningtons.

Then its all about the cache so to say

That also explains the "unrealistic" 25% increase from Phenom to Phenom 2 in this test. Not even to talk about the 100% change to the 7000 series Xeons.

[GoriLLakoS]

The first 2 AMD Phenom II that will be released on 9th of january will be :

1. AMD Phenom II X4 940 Black Edition, Part Number : HDZ940XCGIBOX with 125W TDP, on Socket AM2+. The default speed will be 3000MHz with 8MB Cache.
2. MD Phenom II X4 920, Part Number : HDX920XCGIBOX with 125W TDP, on Socket AM2+. The default speed will be 2800MHz with 8MB Cache.

Source here and transalted here !