Phenom II X4 @ +6 GHz by Team Finland

[Morais]

First, Congratulations for the achievement on 6,3GHz, I liked the video of you guys having fun with these new chips. I wonder why Macci and Peter didnt used AOD for windows OC...Good job guys, macci, peter, sf3d and sampsa, i wish i was there...
Oh and gonna join the ph2 as soon as i can

[cegras]

AMD's 45nm process is different from intel's. I think AMD is using SOI, which would mean it is resistant to higher voltages .. ?

[Oldguy932]

Stock voltage is also 1.35v so going to 1.5 is like nothing.

[informal]

AMD's 45nm process is different from intel's. I think AMD is using SOI, which would mean it is resistant to higher voltages .. ?

That's correct,AMD uses SOI and it will use it at 32nm node also(apart from tuning 32nm for bulk and hk/mg too).As for vcore question,the default Vcore for Shanghai and Deneb is 1.35V. 1.55V is 14% over that value and considering low idle and load temps PhII is showing in previews we seen thus far,implies it's quite manageable under air.I'd say 1.6V would be the highest value for good air or water cooling(18% percent over stock vcore).
The power draw is very good even with the "higher" than usual def. vcore,which is saying a lot about improvement AMD is seeing with their latest process node.If 32nm node nets similar gains ,they will be quite competitive process wise.

[flippin_waffles]

Who will be the first person to build a cascade that goes below -195C? Really curious if getting any colder will help

When Sampsa hinted about something 'cooler' in the future, I immediately thought of baldeagle earlier talking about AMD experimenting with liquid helium at -269c. They seem to have advanced their competitiveness by leaps and bounds. A nice bump in IPC with a substantial improvement of their process.

[Ghostbuster]

That's correct,AMD uses SOI and it will use it at 32nm node also(apart from tuning 32nm for bulk and hk/mg too).As for vcore question,the default Vcore for Shanghai and Deneb is 1.35V. 1.55V is 14% over that value and considering low idle and load temps PhII is showing in previews we seen thus far,implies it's quite manageable under air.I'd say 1.6V would be the highest value for good air or water cooling(18% percent over stock vcore).
The power draw is very good even with the "higher" than usual def. vcore,which is saying a lot about improvement AMD is seeing with their latest process node.If 32nm node nets similar gains ,they will be quite competitive process wise.

You should remember that the power draw rises exponentially with higher voltages, not linearly. At 1.55V the power draw would theoretically increased by 32%, not 14%.

Mind you that at 1.35V the Phenom II is rated at 125W and 95W TDP.

[Ozzfest05]

nice results when is the phenom 2 officially released>

[s3rv3r]

You should remember that the power draw rises exponentially with higher voltages, not linearly. At 1.55V the power draw would theoretically increased by 32%, not 14%.

Mind you that at 1.35V the Phenom II is rated at 125W and 95W TDP.

1.35 - 100%
1.55 - x

x = 155 / 1.35 = 114.81% - 100% = 14.81%

If 1.35 is stock value.

[informal]

You should remember that the power draw rises exponentially with higher voltages, not linearly. At 1.55V the power draw would theoretically increased by 32%, not 14%.

Mind you that at 1.35V the Phenom II is rated at 125W and 95W TDP.

Of course ,i never stated power draw,i said vcore-since we are talking about vcore limits of what's safe and what's not.If you want to be even more accurate wrt power draw,it would be the 3rd degree of voltage ,not square like it is in theory(without leakage currents factored in).
Again,vcore of 1.55V is 14% over the default one.And even with 1.35V ,Shanghai is extremely power efficient chip .

[Ghostbuster]

1.35 - 100%
1.55 - x

x = 155 / 1.35 = 114.81% - 100% = 14.81%

If 1.35 is stock value.

Wrong calculation...

P = V ^ 2 / R

Let's say R = constant

Power difference = (1.55 ^ 2) / (1.35 ^ 2) = 1.318

That's roughly 31.8% (or 32%) higher.

You can try calculating the difference at 1.6V...

[biohead]

1.35 - 100%
1.55 - x

x = 155 / 1.35 = 114.81% - 100% = 14.81%

If 1.35 is stock value.

?

As I recall that is exponential.

Wrong calculation...

P = V ^ 2 / R

Let's say R = constant

Power difference = (1.55 ^ 1.55) / (1.35 ^ 1.35) = 1.318

That's roughly 31.8% (or 32%) higher.

You can try calculating the difference at 1.6V...

you mean 1.55² / 1.35²

[informal]

Ghostbutser,before you do more math,look one post above your last one...
People i never stated power draw,i stated safe Vcore for air. Power Draw is not equal to Vcore...

edit:and like i said previously,if you want to be moe accurate in real world powe draw terms,it's 3rd degree of vcore function,not a square of it(due to leakage issues).

[Ghostbuster]

Ghostbutser,before you do more math,look one post above your last one...

Corrected, still same results.

People i never stated power draw,i stated safe Vcore for air. Power Draw is not equal to Vcore...

Power draw corresponds to Vcore. I even decided not to factor in the frequency and possible leakage at higher voltages (using R as constant).

you mean 1.55² / 1.35²

Corrected... Its supposed to be "^2"..

[biohead]

I know, lel. And I think safe vcore for this new chip is yet to be determined. Everyone is just so fixed in this Intel mindset. I wouldn't worry running 1.65v 24/7 under single stage.

[informal]

@Ghostbuster
Power draw sure does correspond to Vcore,since Vcore is a variable in the very function of it,but we are talking about safe Vcore for air cooling man,not power draw...Is it that hard to understand?

PS s3rv3r's calculation on vcore difference is correct so i don't now what for you corrected him?He didn't even speak about power draw?

[Ghostbuster]

You did say "The power draw is very good even with the "higher" than usual def."... That's why I find it amusing.

Anyway, I do suspect the gates are quite thick at 45nm which is why Vcore is much higher also lower leakage (higher R). Maybe that way its able to tolerate higher voltages.

[Ghostbuster]

PS s3rv3r's calculation on vcore difference is correct so i don't now what for you corrected him?He didn't even speak about power draw?

He responded to my post...

[RPGWiZaRD]

Guys stop ruining the thread with your phatetic discussions about voltage, this is XS ffs, you shouldn't be worrying about this category, I would personally use as much vcore that shows a benefit in the OC, 0.1v increase for 100MHz isn't a reasonable reason to go further for example (unless pushing for records that is).

Looks like Phenom II will offer lots of fun for LN2 users. First time I've seen an AMD liking cold this much, that's incredible that you can just fill the whole pot without worrying about freezing! xD 4 cores @ 6GHz+ is very very nice.

Hyvää työtä Sampsa & SF3D ja muut!

[informal]

You did say "The power draw is very good even with the "higher" than usual def."... That's why I find it amusing.

Anyway, I do suspect the gates are quite thick at 45nm which is why Vcore is much higher also lower leakage (higher R). Maybe that way its able to tolerate higher voltages.

Yeah but it was in the context of "safe" Vcore,which we are trying to debate on.Not the power draw numbers which indeed are good.If one good air cooler can cool down Phenom I 9950 rated at 140W as absolute max TDP while being overvolted to 1.55V(from 1.3V default) and OCed to 3.2-3.3Ghz[power draw would be ~220-240W as absolute worst case scenario] ,you can be sure Shanghai/Deneb will be also cooled by the same cooler even if it was running at ~1.55V and 4Ghz OC.The point is now that stock Vcore is even higher ,power draw dropped due to physics of thansistors at smaller node.1.55V is just 14% over stock and is not that much for an air cooling device to control it.

He responded to my post...

And he was correct since he spoke about Vcore difference...

Guys stop ruining the thread with your phatetic discussions about voltage, this is XS ffs, you shouldn't be worrying about this category, I would personally use as much vcore that shows a benefit in the OC, 0.1v increase for 100MHz isn't a reasonable reason to go further for example (unless pushing for records that is).

Looks like Phenom II will offer lots of fun for LN2 users. First time I've seen an AMD liking cold this much, that's incredible that you can just fill the whole pot without worrying about freezing! xD

Hyvää työtä Sampsa & SF3D ja muut!

Sorry for that,but some people wanted to know what's considered "safe" 24/7 Vcore for air cooling.I gave my opinion,it may be i'm wrong .For definite answer we need thorough report from independent review (temperatures reported in OCed mode) and we need AMD's tech docs .

[Mav451]

The Vcore question had nothing to do with power consumption - just the ability to run it (regardless of power draw) 24/7. I don't care how hot it is or how inefficient it is, as long as it's a safe level of voltage. Again, to emphasize, the question has nothing to do with efficiency - just safety. Jeez some people are thick here.

[Zucker2k]

Yeah but it was in the context of "safe" Vcore,which we are trying to debate on.Not the power draw numbers which indeed are good.If one good air cooler can cool down Phenom I 9950 rated at 140W as absolute max TDP while being overvolted to 1.55V(from 1.3V default) and OCed to 3.2-3.3Ghz[power draw would be ~220-240W as absolute worst case scenario] ,you can be sure Shanghai/Deneb will be also cooled by the same cooler even if it was running at ~1.55V and 4Ghz OC.The point is now that stock Vcore is even higher ,power draw dropped due to physics of thansistors at smaller node.1.55V is just 14% over stock and is not that much for an air cooling device to control it.


And he was correct since he spoke about Vcore difference...


Sorry for that,but some people wanted to know what's considered "safe" 24/7 Vcore for air cooling.I gave my opinion,it may be i'm wrong .For definite answer we need thorough report from independent review (temperatures reported in OCed mode) and we need AMD's tech docs .

So after ghostbuster busted (no pun intended) your power draw argument, you now decide it's irrelevant to your argument? Why bring it up then? Since when did overclockability trump high stock voltage? Sorry, I thought AMD was the green company here? Your problem is that you are blind to the truth until you need it to make a point, your point, then all of a sudden it's relevant. Seriously, you need to heed my advice.

[informal]

The Vcore question had nothing to do with power consumption - just the ability to run it (regardless of power draw) 24/7. I don't care how hot it is or how inefficient it is, as long as it's a safe level of voltage. Again, to emphasize, the question has nothing to do with efficiency - just safety. Jeez some people are thick here.

Exactly,thanks for the post we're on the same page.
There is no tech dock for public use yet,but as Deneb launches we should know what AMD states as max. acceptable Vcore for 45nm gen. of chips.
It looks like Phenom IIs might need around 1.55V for 4Ghz OC.It will vary from chip to chip,or batch to batch,but knowing AMD's CTI prorgam,it will only get better with time .
@Zucker2k
I won't bother with your flame baiting anymore.You're just a troll that's all.

[Zucker2k]

Exactly,thanks for the post we're on the same page.
There is no tech dock for public use yet,but as Deneb launches we should know what AMD states as max. acceptable Vcore for 45nm gen. of chips.
It looks like Phenom IIs might need around 1.55V for 4Ghz OC.It will vary from chip to chip,or batch to batch,but knowing AMD's CTI prorgam,it will only get better with time .
@Zucker2k
I won't bother with your flame baiting anymore.You're just a troll that's all.

AMD

[cegras]

AMD

^ What does 'flame bait' mean to you?

Next time I'll keep my questions to a PM or something. Good god.

[Zucker2k]

^ What does 'flame bait' mean to you?

Next time I'll keep my questions to a PM or something. Good god.

I'm not flame-baiting him, I'm just telling the truth as I see it. The guy (Informal) thinks AMD is God or something, AMD is not sacrosanct, as far as I'm concerned.