"7 Watt" Ivy Bridge my arse!

[Exophase]

Originally Posted by Idontcare

Not only does your static leakage go down with lower temperature, but you can also further reduce the operating voltage as temperature declines which in turn reduces power consumption (static and dynamic) all the further.

Out of curiosity, do you have any numbers on the power needed to provide the extra cooling, assuming that you're running the desktop in a normal environment and not with its case open in an under heated house in the middle of winter?

For instance, this tablet fan is 2W before any loss in the 5V regulator: http://www.laptop-cpu-fan.com/lenovo...Fe1xOgodnEUA7w

[Puppies04]

Originally Posted by CharlesKozierok

LOL. Didn't AMD try to play a similar game a few years ago?

Edit: Ah yes, here it is.

A few years ago? Have you seen how much power a stock "125W" FX 8350 pulls?

[AtenRa]

TJmax = TAmp + (Rθja x Power)

TAmp = Ambient Temperature
Rθja = junction-to-ambient thermal resistance

TAmp and Rθja should be constant in our case, we only have a change in the TJmax from 105c to 80c and Power from 10W to 7W.

Now, it seams to me that SDP(Scenario Design Power) is something like AMDs ACP(Average CPU Power). So, if the CPU will not dissipate 10W or 13W and it only dissipate 7W because it will not be stressed up by the software to reach the Pmax allowed (10W or 13W) then the TJmax drops(in that case to 80c).

So it has nothing to do with better/bigger Heat-sink or lower temperatures or no Turbo etc. It is just another TDP but, at 105c TJmax the TDP remains at 10W or 13W(depending of the CPU) and that is the design goal for the heat-sink.

[IntelUser2000]

SDP according to PCWatch:

Additional level below cTDPdown that manufacturers can set. The limits are both Tjmax and clocks. They do say peak power usage is lower.

So maybe I'm wrong about the Turbo thing but the basic premise is same. They cut peak clocks and lower temperatures to cut power.

[IntelUser2000]

Originally Posted by dastral

Whats a cTDP ? how can it be 3W lower than the TDP.... This is REALLY WEIRD....

cTDP: Configurable TDP
http://www.anandtech.com/show/4764/i...e-tdp-detailed

Manufacturers can set lower(cTDPdown) or higher(cTDPup) based on various scenarios. It's same as TDP.

It's for Convertible Ultrabooks where you can use cTDPdown for say, in a Tablet environment, Nominal for Laptop, or cTDPup for docked or with a cooling pad.

Ah yes, I watched AMD's CES presentation again and they did mention the Clover Trail tablet with a side by side demo. Wow does that thing suck! Lol. Not trying to flame, it just really does suck. On the bright side, I missed it before that there will be several Hondo based tablets coming to market in the weeks ahead.

With TWICE the battery life and being thinner and lighter, I doubt most non-Anandtech consumers will care.

[Idontcare]

Originally Posted by dastral

Which leads to my suspicions about ""At 80° and Nominal speed, this chip will consume 7W"
If that is correct, SDP is just a marketing ploy to impress the world with something we already knew...
"Keep it cooler, and it will consume less"

It is definitely a numbers game which has no lower bounds to the absurdity with which it could descend.

They could define SDP as the power used at 35°C and just require/expect OEM's to create magical water-cooling solutions that can keep the temperature at ambient + 7°C. Ridiculous to be sure but still technically feasible and possible.

Likewise they could define the expected workload as being tantamount to running notepad or task manager. I.e. the idle power usage becomes "SDP" and it is marketed based on that. It would be absurd, but technically legit.

Originally Posted by Exophase

Out of curiosity, do you have any numbers on the power needed to provide the extra cooling, assuming that you're running the desktop in a normal environment and not with its case open in an under heated house in the middle of winter?

For instance, this tablet fan is 2W before any loss in the 5V regulator: http://www.laptop-cpu-fan.com/lenovo...Fe1xOgodnEUA7w

I've unplugged the fans on my stock HSF and on my NH-D14 while the system was fully running and the power usage of those fans turned out to be tiny (~1W) but presumably we aren't talking about a 120mm fan to cool this 7W chip. Surely 7W could be effectively passively cooled with chasis-coupling or some such (turn entire tablet into a huge passive heatsink).

[Idontcare]

Originally Posted by AtenRa

TJmax = TAmp + (Rθja x Power)

TAmp = Ambient Temperature
Rθja = junction-to-ambient thermal resistance

TAmp and Rθja should be constant in our case, we only have a change in the TJmax from 105c to 80c and Power from 10W to 7W.

Now, it seams to me that SDP(Scenario Design Power) is something like AMDs ACP(Average CPU Power). So, if the CPU will not dissipate 10W or 13W and it only dissipate 7W because it will not be stressed up by the software to reach the Pmax allowed (10W or 13W) then the TJmax drops(in that case to 80c).

So it has nothing to do with better/bigger Heat-sink or lower temperatures or no Turbo etc. It is just another TDP but, at 105c TJmax the TDP remains at 10W or 13W(depending of the CPU) and that is the design goal for the heat-sink.

You are absolutely correct, except you made the same minor mistake I kept making which is that you refer to the 80C value as a "TJmax" when what Intel is saying is that TJmax remains 105C, no change, but when the TJ is 80C then power usage can be expected to be 7W, etc.

They are all points on the TJ curve, there is one TJmax (105C) and then a bunch of other points for power usage at a given TJ. Intel just picked (characterized) the power profile when TJ is 80C in these marketing slides and called that SDP for now.

They could have picked TJ at 75C or 85C, the 80C value itself is purely arbitrary in terms of selection, but the 7W value is not arbitrary (it comes from characterizing power usage at a TJ of 80C).

It is not that TJmax is now 80C, but that when TJ is 80C you have the power usage they are reporting. (note the footnote in the image)

[AtenRa]

@IDC,

yeap my mistake, it is just TJ, TJmax is 105c

[Exophase]

Originally Posted by Idontcare

I've unplugged the fans on my stock HSF and on my NH-D14 while the system was fully running and the power usage of those fans turned out to be tiny (~1W) but presumably we aren't talking about a 120mm fan to cool this 7W chip. Surely 7W could be effectively passively cooled with chasis-coupling or some such (turn entire tablet into a huge passive heatsink).

My question is, will it need more enhanced cooling than it would otherwise in order to decrease power consumption of the CPU. And basically, at what point is this investment worth the added power consumption of the enhanced cooling.

Of course the total output of the tablet will be more than the CPU chip, also including the 3W PCH, a few W for RAM, and some consumption and efficiency loss in the PMIC. I don't know how much it helps having these components more spread out, but Haswell integrated VRMs counteracts this.

I agree they won't use a 120mm fan, if they do need one it'll be smaller and AFAIK less efficient per amount of heat removed (but I don't know how fast it'll need to spin).

Originally Posted by Idontcare

You are absolutely correct, except you made the same minor mistake I kept making which is that you refer to the 80C value as a "TJmax" when what Intel is saying is that TJmax remains 105C, no change, but when the TJ is 80C then power usage can be expected to be 7W, etc.

So is the 7W what you get running the same processor load (frequency) and merely the savings yielded from lower temperature including the ability to use a lower voltage? Or are they merely saying that throttling it to 80C will also limit power consumption to 7W, but at a reduced load from what you can get at 105C/13W? Because that second part seems like a trivial claim, except to note that it CAN scale to 7W outside of idle states. This seems like the burning question to me, with different people on this thread claiming one or the other (ShintaiDX at least seems to be claiming the former)

[Idontcare]

Originally Posted by Exophase

My question is, will it need more enhanced cooling than it would otherwise in order to decrease power consumption of the CPU.

Depends on what you are doing, obviously. It doesn't look like they are talking about superior binning, they are basically saying "look, if you can keep the temperature at 80C or below then the chip is gonna only use 7W or less, go above 80C and you are entering into 10W territory".

So there are two ways to keep the temperature low - (1) improve the thermal conductivity and dissipation (see delidding the IB for example) or (2) use apps that are less demanding on the processor such that even when loaded the chip is not burning through the watts (stop using IBT people, stick with multi-threaded notepad as your most strenuous app).

If you intend to keep the processor load the same (running same demanding apps), then obviously you need better cooling to get it below 80C if it otherwise operates above 80C when running the same app. If you configure the TDP to be the SDP then you are basically dialing down the effective CPU loading (TDP limit throttling) such that the processor is guaranteed to not exceed TJ of 80C when it hits SDP...performance suffers but it doesn't cost you any more because you can use the same crappy cooling solution.

Originally Posted by Exophase

And basically, at what point is this investment worth the added power consumption of the enhanced cooling.

Of course the total output of the tablet will be more than the CPU chip, also including the 3W PCH, a few W for RAM, and some consumption and efficiency loss in the PMIC. I don't know how much it helps having these components more spread out, but Haswell integrated VRMs counteracts this.

I agree they won't use a 120mm fan, if they do need one it'll be smaller and AFAIK less efficient per amount of heat removed (but I don't know how fast it'll need to spin).

I have absolutely no clue as to the BOM expense structure these guys are looking at on their end. Its something I've never had the opportunity to be exposed to on the other side of the fence.

I know with our desktop processors there is a very small window where replacing our stock HSF's with 3rd party coolers (at our own expense) reduces power consumption enough to justify the expense.

Your 3rd party HSF needs to cost you less than $30 and it needs to lower your operating temps by roughly 20C if you have any hope of breaking even on your investment over the course of 2-3 yrs of operation.

[piesquared]

Originally Posted by IntelUser2000

...

With TWICE the battery life and being thinner and lighter, I doubt most non-Anandtech consumers will care.

TWICE the battery life in what scenario? I haven't seen the review of that Vizio tablet, where did you see it? The only thing I saw was a stuttering mess on Clover Trail, of course that kind of performance might give slightly longer battery life. I think most people are going to want a tablet with a rich 1080p display that is useful, balanced and dominates gaming as opposed to an e-reader that can play lower quality movies. They might as well go to Barnes&Noble and pick up the latest top seller.

[edit]
Anyway, last post from me on this /off topic. This topic is about intel misleading consumers, not Clover Trail's bottom of the barrel performance.

[Charles Kozierok]

Originally Posted by Exophase

So is the 7W what you get running the same processor load (frequency) and merely the savings yielded from lower temperature including the ability to use a lower voltage? Or are they merely saying that throttling it to 80C will also limit power consumption to 7W, but at a reduced load from what you can get at 105C/13W? Because that second part seems like a trivial claim, except to note that it CAN scale to 7W outside of idle states.

This kind of puzzles me as well. Which comes first, the chicken (lower power) or the egg (lower temps)?

[Idontcare]

Originally Posted by Exophase

So is the 7W what you get running the same processor load (frequency) and merely the savings yielded from lower temperature including the ability to use a lower voltage? Or are they merely saying that throttling it to 80C will also limit power consumption to 7W, but at a reduced load from what you can get at 105C/13W? Because that second part seems like a trivial claim, except to note that it CAN scale to 7W outside of idle states.

Originally Posted by CharlesKozierok

This kind of puzzles me as well. Which comes first, the chicken (lower power) or the egg (lower temps)?

It is both, of course, because you (the OEM) have the choice of which path you wish to pursue for your particular product.

Design an aggressive cooling solution such that you are able to keep the temperature below 80C when it would have otherwise rocketed past 80C with a cheaper cooling solution and you will reap the benefit of then having a 7W footprint where others have a 10W product.

Or...keep the same inexpensive cooling solution but set the cTDP to be equal to the SDP and the product will throttle itself (reducing performance but keeping power usage in check) such that it only uses 7W.

Intel is saying "we aren't going to make the choice for you, you are free to choose which path you wish to pursue in creating the products you eventually want to bring to market and sell to your customers, but we are making both options available to you, either path will net you a 7W footprint, gl gg gb & gn"

[Hans de Vries]

Originally Posted by Idontcare

It is definitely a numbers game which has no lower bounds to the absurdity with which it could descend.

Frankly I wouldn't even be surprised to see a 2014 3W Haswell smartphone
demo setting a new world record in SunSpider below 200ms....

With IVR's (integrated voltage regulators) you can in principle run Haswell with
125W for a period of 200 ms from an ultra-capacitor in a phone.

External power supplies show real power dissipation, but not anymore with IVR's.

The 0.65W TDP Atom Z515 with up to 1.2 GHz and Intel Burst Performance
technology requires an external 2.8W (1.1V 2.5A) power supply for instance
which is 4.3 times the specified TDP:

http://www.cpu-world.com/CPUs/Atom/I...66UC009DV.html
http://www.intel.com/content/dam/www...-datasheet.pdf

and that's the TDP not even the SDP or whatever.


Hans.

[TuxDave]

Originally Posted by Idontcare

It is both, of course, because you (the OEM) have the choice of which path you wish to pursue for your particular product.

Design an aggressive cooling solution such that you are able to keep the temperature below 80C when it would have otherwise rocketed past 80C with a cheaper cooling solution and you will reap the benefit of then having a 7W footprint where others have a 10W product.

Or...keep the same inexpensive cooling solution but set the cTDP to be equal to the SDP and the product will throttle itself (reducing performance but keeping power usage in check) such that it only uses 7W.

Intel is saying "we aren't going to make the choice for you, you are free to choose which path you wish to pursue in creating the products you eventually want to bring to market and sell to your customers, but we are making both options available to you, either path will net you a 7W footprint, gl gg gb & gn"

Very accurate summary, especially the first point of giving the OEMs to chance to get a competitive advantage through cooling innovations. Although I guess it's debatable whether Intel openly gave the OEMs the choice or was it the OEMs demanding that Intel give them that choice.

[sm625]

This begs the question, why does surface pro have a fan? Why not just limit it to 7W so it doesnt need a fan? And why does 2012 ivy bridge still require a 3 watt PCH anyway? The whole topic is moot so long as this remains true. You've got 3 watts in the PCH and probably another 3 watts lost in the power supplies, another 2 watts in memory/NAND and 3 in the display. So even if the cpu is limited to 7 watts, you're still burning through 18 watts, or looking at 2~2.5 hours battery life under load. And then there's wifi/cellular on top of all that? Ugh. This is miserable any way you slice it.

[Charles Kozierok]

Okay, sure, they have a choice. But didn't they always?

[Olikan]

that remembers me that ivy's HD 4000 would be 3 times faster... than in a very small font saying: against HD-2000

not a lie, but it is a really cheap move

this remembers me that... haswell igp might be 5 times faster than HD-2500

[ShintaiDK]

Originally Posted by sm625

This begs the question, why does surface pro have a fan? Why not just limit it to 7W so it doesnt need a fan? And why does 2012 ivy bridge still require a 3 watt PCH anyway? The whole topic is moot so long as this remains true. You've got 3 watts in the PCH and probably another 3 watts lost in the power supplies, another 2 watts in memory/NAND and 3 in the display. So even if the cpu is limited to 7 watts, you're still burning through 18 watts, or looking at 2~2.5 hours battery life under load. And then there's wifi/cellular on top of all that? Ugh. This is miserable any way you slice it.

The Surface Pro uses a regular 17W product as far as I know.

Its the combined savings that matters. Display makers, memory makers etc also pursue the same. 21W vs 18W is still almost a 15% difference for example.

[Idontcare]

Originally Posted by CharlesKozierok

Okay, sure, they have a choice. But didn't they always?

For the "implement better cooling to get lower power usage at same performance" option, yes, that option has always been there.

But they didn't have the option to set the max TDP to any number within a range of values (that is what cTDP brings to the table) nor did they have the option of being guaranteed by Intel that if they keep temperature below a specific value (80C in this case) then they can formally spec (and advertise as such) the product as being a "7W Processor" which is a big deal for OEMs selling to resellers and retailers.

This is about legitimizing in the business sense what was available to the hobbyist in the past. Turbo-boost/core was one step in that direction, adding the performance tuning plan was another (OEMs can now sell pre-OC'ed rigs that are legitimately covered under warranty), and configurable TDP is yet another step in that same direction of legitimizing a product segment so that it can then be monetized and brought from the fringe and into the mainstream (where the volumes are to be found).

[Haserath]

Originally Posted by sm625

This begs the question, why does surface pro have a fan? Why not just limit it to 7W so it doesnt need a fan? And why does 2012 ivy bridge still require a 3 watt PCH anyway? The whole topic is moot so long as this remains true. You've got 3 watts in the PCH and probably another 3 watts lost in the power supplies, another 2 watts in memory/NAND and 3 in the display. So even if the cpu is limited to 7 watts, you're still burning through 18 watts, or looking at 2~2.5 hours battery life under load. And then there's wifi/cellular on top of all that? Ugh. This is miserable any way you slice it.

The surface pro has a fan because it's just that cool.

[Haserath]

One thing that gets me is why don't we have a tweak able laptop yet... I'd love to boost battery life and reduce heat with some voltage tuning.

[Exophase]

Originally Posted by Idontcare

It is both, of course, because you (the OEM) have the choice of which path you wish to pursue for your particular product.

Design an aggressive cooling solution such that you are able to keep the temperature below 80C when it would have otherwise rocketed past 80C with a cheaper cooling solution and you will reap the benefit of then having a 7W footprint where others have a 10W product.

So you're saying that better cooling alone will decrease power consumption for the same workload.

But then you said you don't know what workload actually measures, that it could be anything. If what you're saying is true then it's whatever the same workload 10W cTDP is. Now I have no idea what that represents, but cTDP isn't really new so maybe there's some information out there. I expect Intel should still be able to run the CPU cores at its base clock at least, under "normal" programs that fully load it (ie not doing weird power virusy things). And with the GPU only doing minimal GUI stuff.

The problem is that nowhere do I see that Intel is clearly describing this. Maybe to you it's implied by the slide, but I think there's still some room for ambiguity. 10W to 7W is a huge reduction just from 25 degrees C at TJ, which I imagine is nowhere close to saying a 25C reduction in ambient; the temperature of the cooling elements on the tablet are going to be nowhere close to 25C over the air around you.

The savings should be less dramatic if already starting at lower voltage levels, especially given that Intel's 22nm voltage/clock curve rises more steeply than on its older processors. For your graph you have things fixed at a high 1.29V which I'm sure exaggerates the results.

[Ferzerp]

Originally Posted by Exophase

So you're saying that better cooling alone will decrease power consumption for the same workload.

But then you said you don't know what workload actually measures, that it could be anything. If what you're saying is true then it's whatever the same workload 10W cTDP is. Now I have no idea what that represents, but cTDP isn't really new so maybe there's some information out there. I expect Intel should still be able to run the CPU cores at its base clock at least, under "normal" programs that fully load it (ie not doing weird power virusy things). And with the GPU only doing minimal GUI stuff.

The problem is that nowhere do I see that Intel is clearly describing this. Maybe to you it's implied by the slide, but I think there's still some room for ambiguity. 10W to 7W is a huge reduction just from 25 degrees C at TJ, which I imagine is nowhere close to saying a 25C reduction in ambient; the temperature of the cooling elements on the tablet are going to be nowhere close to 25C over the air around you.

The savings should be less dramatic if already starting at lower voltage levels, especially given that Intel's 22nm voltage/clock curve rises more steeply than on its older processors. For your graph you have things fixed at a high 1.29V which I'm sure exaggerates the results.

Where are you arriving at the assumption that it is anything less than 100% theoretical load at 80C? I didn't see this other than information introduced by posters trying to criticize the idea.

[Exophase]

Originally Posted by Ferzerp

Where are you arriving at the assumption that it is anything less than 100% theoretical load at 80C? I didn't see this other than information introduced by posters trying to criticize the idea.

I'm NOT assuming that. It's others who are assuming that it necessarily can do 100% theoretical load (relative to 10W at Tjmax) at 80C Tj consuming 7W. I'm saying it isn't clear to me that this is being guaranteed by Intel.