Did anyone try just air OC first, or did they go straight at the LN2?

I heard tha it was just under 4ghz with air.

Am i missing sth? almost 18S for crunching 1 M of super pi with a 5ghz cpu? i will pass that.

^ Yeah, I noticed that too. i7 I think gets like 7s or so.

Well you two are just a couple of negative nancys!

This could still be that Q6600 killer that AMD is hoping for!

Did anyone notice the vcore. 1,8 is huge!!!
But thumbs up for AMD, and those who OCd it

Yep. I think the fact that AMD fixed their OC problem is really the point here. Sure, Phenom II is still far behind the Core i7, but its a step in the proper direction. The wrong direction would be like what Intel did with the Pentium 4 back in the day.

Now if only AMD could figure out how Intel took such an architectural jump over them and rebound. Does anyone have any specs, does this really even compete with the Q6600 as was joked about?

Correct me if im brong but it seems PhenomII @4.9ghz is two second slower in SuperPi...? I got 16sec on 1mb file I thought at those speeds PhenomII will be much faster lol

[IMG]http://i33.tinypic.com/2lxumg2.jpg[/IMG]

I don't think AMD is out to displace the i7 right now. They are aiming at the Core 2 series. If they can manage to place themselves between the Core 2s and i7s then they may be in good shape. We'll have to wait and see.

SuperPi is not core optimized, and has favored Intel architecture since the Athlon 64 days.

Exactly. The idea for AMD is to become the mainstream while Core i7 is still in its enthusiast stage. It'll be a huge benefit for them if they can outpace Core 2 in the months leading up to i7 becoming mainstream.

LOL my SUPERPI time on my 6000+ stock was 29 seconds. depressing

although it seems that Tom's was doing something different than AMD for overclocking since AMD managed 6+ GHz on 1.9Vcore? Although either way, I'd like a quad and 3 GHz is definitely a plus.

+1

anyone else notice that the picture at which the time was taken says the clock was only @ 4ghz?

http://www.tomshardware.com/galler [...] -png-.html

So at 4ghz it got 17.769 seconds the multiplier is only at 20x on the picture that shows the pi calculations when the 4.9ghz is at 24x

My pentium dual core @3.0 ghz runs super pi 1M in 20 seconds. I'll pass on phenom II.

Indeed.

SuperPi only uses 1 core and is optimized for Intel CPUs. It's also not the end-all, be-all of benchmarks by a long shot. In most real-life application benchmarks involving number-crunching, AMD has historically been better than Intel clock-for-clock. Mind you, I haven't seen the benches for Shanghai vs. Nehalem.

BTW, my Xeon E3110 @ 3.87 GHz does it in 16 seconds, but the current Phenom X4 9950 BE @ 3.4 GHz kicks its ass in anything multithreaded and certain applications, plus it seems quicker for basic functions in Vista.

Holy mother fucking butt monkeys! There's actually tards out there who think it's a failure because of the superpi score? That's fucking super-tard, get off the enthusiast site noobs. Superpi is strictly an intel extreme overclocking bench, has no bearing on actual performance, it just favors intel's architecture. If that's what you base your cpu purchases on then have fun with your P4. I'm sure you'll be praising it's 1M score as it chokes every card you put in your system.

I think some people are seeing the big picture. Yes it is too early to say anything about AMD's new chip. Real world numbers will mean more then overclock numbers. But price is going to be the big factor. If they are going to charge $300 bucks for a PhenomII then AMD is just going to hurt itself.

Q6600 killer? Intel is set to make the Q6600 EOL 1H 2009... i wanna hope its a helluva lot better than Q6600 a new AMD cpu competing with an EOL cpu from intel yay... NOT

Apparently, there are people who buy computers only to calculate Pi up to 1 millionth digit using a specific old program.

Didn't someone years ago get a P4 to almost 7Ghz on LN2? P4s overclocked like crazy, it didn't make them good when compared to the Athlons. Super Pi is just one benchmark, but for the most part you can probably expect the Phenom II to be a Phenom I that runs at modern clock speeds. I dont think theres going to be any 40% performance gains here

i thought these were supposed to come close to this achievement on air...

[citation][nom]Jerky_san[/nom]anyone else notice that the picture at which the time was taken says the clock was only @ 4ghz?http://www.tomshardware.com/galler [...] -png-.htmlSo at 4ghz it got 17.769 seconds the multiplier is only at 20x on the picture that shows the pi calculations when the 4.9ghz is at 24x[citation]

I noticed that. A near 25% clockspeed increase could help the superpi tome to come down a couple seconds, since people seem so concerned. Not sure how many gamers are calculated PI to the nth degree in their spare time though...

Ummmm... unless I've missed something here, Core i7 is slower than Penryn which is about the same as Conroe, so in effect, a Q6600 killer is a Core i7 killer. And Intel CPUs have always been faster at SuperPi, even when they were slower in real life against Athlon64s(although that score seems a tad odd). Since 3.5ghz is a typical high-end 24/7 stable overclock of an Intel quad, I think it's safe to say that AMD has caught up and maybe even surpassed Intel.

Im thinking we will see much better clocks than this. Again, a pitiful small mention, too bad this wont hold up

perhaps a DDR3 variant of the Phenom II will hit 6ghz, if we compare the DDR1 and DDR2 Athlon 64's the memory controller and its lower power requirement gave it a little more head room etc?

SuperPi was actually created in 1995 and doesnt use anything like multi-cores, SSE instructions etc - it shines on decent cache/memory sub system designs and solid core designs - i think id find it more valid then PCMark etc in some reguard, and YES is does seem to run better on Intel platforms lately and thats because Intel platforms HAVE been faster lately.

For reference:
P4 2M @7.4ghz got around 20 seconds
E6600 @2.4/1066 got 20 seconds
E6600 @3.2/1600 got 15 seconds

Ummm, dont be disappointed by those numbers, it may be that whoever oceed it didnt know what they were doing.

"We got to test some Phenom II X4 CPUs with liquid nitrogen (LN2) during our visit and peformed air and LN2 overclocking competition against Team AMD (macci & Pete Hardman) in front of the live audience.

We managed to run Phenom II X4 940 Black Edition CPU-Z stable at 6,28 GHz with 4 cores enabled, Crysis CPU timedemo at 6,1 GHz and 3DMark05 & 3DMark06 at 6,0 GHz. We used 1,95V vcore for CPU and cooled Dragon F1 Extreme Edition copper container to -173 degrees Celsius. AMD suggested us to run HyperTransport-link at 1 GHz (HT 1.0) and we didn’t see any coldbugs during our tests. "

It appears the HTT was set waaay to high

http://plaza.fi/muropaketti/artikk [...] la-typella

1,95v!?!

I don't think 1.8 is that bad with below ambient cooling. With intels you often see numbers starting with a 2.

I noticed the bus speed in the cpu-z picture being relatively low - wouldn't it have made a difference if it was to run faster at a lower multiplier? or is that only of importance if the memory controller isn't on the cpu?

as for the post above me - kinda impressive, but air cooled speeds is what most of us want to know about, cause that's what we can reproduce.

TH you suck! How could you manage to screw OC-ing? There is a LOT of proof over the i-net that this CPU is clockable to ~6Ghz on LN2, while you manage to get it to