Ivy Bridge review thread

[Dahakon]

Originally Posted by vain

I never mentioned watts, please don't misquote me. While you may have mentioned watts, I've been talking about power consumption, of which perf/w would have an effect on the power consumption, and in the end, the temperature of the chip, holding all things equal. Here is the link again, a fourth time...

One may be talking about race to zero, which is part of powerconsumption, but that's pretty much dodging the case, as not everything is a race to zero. As example I would bring forth the usercase of gaming. How fast would IB race to zero while the rest is still working? Other than closing down the game because it looks/play horrible that is. The powerconsumption will stay high during the gaming, as the chip has to work hard. There is no escape there.

Of course, it also still does not deal with the high temperatures during idle, but let me not go into that right now.

[vain]

Originally Posted by Dahakon

One may be talking about race to zero, which is part of powerconsumption, but that's pretty much dodging the case, as not everything is a race to zero. As example I would bring forth the usercase of gaming. How fast would IB race to zero while the rest is still working? Other than closing down the game because it looks/play horrible that is. The powerconsumption will stay high during the gaming, as the chip has to work hard. There is no escape there.

You're absolutely right, but you have to agree that W/perf is most always viewed as a race to zero, whether you increase the denominator or reduce the numerator.

[accord1999]

Originally Posted by DCO

And then you came quoting people that were talking about laptops, now it seems that a hot chip is better for laptops, the spin is getting ridiculous.

But then you're also extrapolating that a laptop IVB is "hotter" only based on temperature sensor readings from the desktop IVB. If we do this, we also get that:

a) for the same performance level, notebook IVB requires less power
b) for a CPU intensive task, IVB with an identical power limit will complete it faster than Sandy Bridge, resulting in overall lower power use

Neither of which results in a hotter laptop.

The ridiculous spin is the use of a temperature reading, which may not even be very accurate, to determine how hot a laptop will get and the completely incorrect use of Physics to argue this point

[DCO]

Originally Posted by accord1999

But then you're also extrapolating that a laptop IVB is "hotter" only based on temperature sensor readings from the desktop IVB. If we do this, we also get that:

a) for the same performance level, notebook IVB requires less power
b) for a CPU intensive task, IVB with an identical power limit will complete it faster than Sandy Bridge, resulting in overall lower power use

Neither of which results in a hotter laptop.

The ridiculous spin is the use of a temperature reading, which may not even be very accurate, to determine how hot a laptop will get and the completely incorrect use of Physics to argue this point

Reviews tell that IB is hotter than SB with the same TDP ratings. I'm extrapolating the data we have, I thought we were using that data and not our imaginations.

IB has more perf/w, but it's going into notebooks with the same power requirements, not the same performance requirements. And the data we have is that it's going to be hotter, you only have to read reviews. It can end faster their task and be hotter, is not something exclusive.

Always more temperature indicates that something is hotter, always has been a bad thing, all people see that reviews show that IB is hotter, and in this forum some individuals twist this to "is going to be cooler", "you don't have real data but mine taken from my *** is valid".

[DCO]

Originally Posted by vain

I commented on fusion's post, who commented on exo's post in regards to your post, before which exo commented on your post regarding 0 watts, which came about because you responded to MY post. Follow along the post trail by clicking the arrow to see where the quotes go. While way way back it may have been me responding to you about your discussion of laptops and heat, since then you've quoted my post more than once.

I don't have laptop data either, which is why I made that comment you see up above in what you quoted.

Practically all my posts have the word notebook or laptop, I answered to a post that talked about notebooks, you were talking about notebooks, Fusion comes and say that I was talking about notebooks to someone and then you write that you didn't know that I had one notebook with IB.
At least be coherent.

It seems that it is difficult to understand. I never said their watts (TDP) are different, I said their power consumption is different. Let's take your hypothetical example...
If you extract a large file on an old old laptop that consumes X amount of power during full load, and on a newer one that does so at the same X amount of power during full load but takes half as long, then will their power consumption be identical in the end?

Ah, you are know talking about average power over an amount of time, but that is a change of subject, the third one, the initial one was if IB was going to be hotter or cooler on a notebook, then you changed to it was going to consume less power and now is less power over time.

It will consume close to "0 watts" when it's sitting on its rear waiting for your C2D to finish. Again, please click the link and look at what I've provided you. Look at the data on the bottom, it's visual and easy to understand. Here it is for the third time. I even bolded the part that's important to read.

Notice how ivy is just sitting there waiting for other processors to get done with the task? There is really not much more I can say to make this any clearer.

Depends on the workload, a game will finish when I want for example, but in general yes, but we are talking about different things.

As I said (my third time) IB is going to have for notebooks chips rated for the same power envelope as SB. If you want I can show you SB chips with the same TDP with less power consumption, for example some 2500k or 2500, on desktops, but we were referring to notebooks.
And perhaps this is news for you but on notebooks chips work near the max TDP, they have turbo (with a wider gap), the are no gaps of 30w (65w-->95w), you have 17, 25, 35, 45, 55w.

So if the power is going to be the same, it gets converted to heat, and IB has more problems to evacuate that heat it should end hotter and no cooler, and the all data we have from desktops corroborate this, that is hotter. And physics corroborate it.

Now if what you want to say is that it can be cooler more times because with more performance it will end it's task sooner, ok, but we weren't discussing this.

I never mentioned watts, please don't misquote me. While you may have mentioned watts, I've been talking about power consumption, of which perf/w would have an effect on the power consumption, and in the end, the temperature of the chip, holding all things equal. Here is the link again, a fourth time...

Power consumption is measured in watts.

I recommend understanding how the data is presented and what it is that is being presented.

Here is what I also said: Grab a trusty calculator and enter 0.8[*] 0.8[*] 0.8[*] 0.8 ...

I'm curious when one would reach zero. Think of this number as being a theoretical W/perf.

If you read carefully, I mention W/perf, not perf/W, and certainly not W. So did you get to zero yet?

Excuse If I find hard to follow all those changes of subjects, metrics than you talk about but later you didn't, etc.
But I was talking about watt-->heat-->heat conduction-->cooling, and you came with 0.8, so if I apply that metric to all new processors that's what we have, if you want to know when we reach zero resolve the equation or limit.

[fusion]

http://www.overclockers.com/ivy-bridge-temperatures

Seems IB is hotter because intel cheaped out on materials connecting the die to the heat spreader. The only good news from that is that laptop clips are generally lidless so they shouldn't have the same problem.

[xeizo]

Well, the last word seems to be the heatspreader actually doesn't worsen the temperatures that much. IHSes have already been popped off and IBs benchmarked. The best speculation imho is that Intel can't reach their target vcore of 0.8V because of process troubles, hello Nvidia(!), so they're stuck with a much too high vcore for the original design resulting in higher than expected temperatures.

I compared my 2600K to a 3770K in a similar system just running SuperPi 32M - both cpus at 5GHz - indeed the 3770K cooled by a Noctua NH-D14 is hotter(I run Antec H2O 620);

http://xeizo.com/blogg/?p=249

[vain]

Originally Posted by DCO

Long reply post.

I guess we can agree to disagree?

I'm apparently unable to show you the dots I'm trying to connect, and I apparently can't understand how you connect yours.

I could respond to your replies individually with what I consider sound physics and logic but we're really going no where here. My apologies.

Guess I'll wait for the real mobile deal in regards to explaining what I believe is right. Though I think we'll still be in disagreement even with those numbers, it may be easier for us to share our ideas.

[vain]

Promo for newegg ivy : http://promotions.newegg.com/intel/12-1596/index.html?

Guess they're up for sale here in the states, might have missed them earlier.

They're advertised as 77W on newegg, not sure what the sticker shows.

[realjjj]

On the egg at inflated prices.
Microcenter has them much cheaper http://www.microcenter.com/search/se...995+4294922744