Intel Discusses Y Chips For Tablets

Status
Not open for further replies.

whyso

Distinguished
Jan 15, 2012
689
0
19,060
I guess what he means is the difference between running prime 95 and high level use (endcoding on all cores).
Or furmark vs. gaming power consumption for gpus.
 

tpi2007

Distinguished
Dec 11, 2006
475
0
18,810
Yeah, but if we want to know the chip's TDP, we should be told what it really is. Otherwise we will be greeted with a mesage "your tablet is overheating, it will now shut down until it cools down". Hmmm, I think I've read about a popular tablet doing this.

So, the thing is, talk about SDP all you want, but unless you are actually giving OEMs the freedom to actually make the SDP converge with the TDP by running downclocked versions of the CPUs, you should also provide TDP for those who wish to know.
 

TeraMedia

Distinguished
Jan 26, 2006
904
1
18,990
Isn't there a flaw in his thinking? Yes, the usage model in tablet mode may involve lower power consumption and heat generation, but as soon as you reconnect the keyboard and turn it back into a laptop or notebook, the system may be subjected to the same higher-level power consumption rates that other laptops face. So you have to design the thermals for the laptop (high power) mode, not the tablet (low power) mode. After all, it's not like you can pull the CPU out from behind the glass when you connect the keyboard.
 

tpi2007

Distinguished
Dec 11, 2006
475
0
18,810



Exactly, hence my point also.
 

ojas

Distinguished
Feb 25, 2011
2,924
0
20,810
Intel's provided both the TDP and SDP for their Y series chips, th y-series i5s for example had a TDP of 13w and an SDP of 7 or 8w, can't really remember. I think it was 7w.

I don't see the point of this post, less why it's "news" since the y-series slide leak was about a month ago. A few people speculated then that since both the y-series and surface pro launch in Jan, the surface would most likely have a y-series i5, probably the only way to make it fanless.
 

CrArC

Distinguished
Jun 5, 2006
219
0
18,690
"For the devices that we're targeting, as they move to more content-consumption (tablet-use) workloads, it doesn't make sense anymore to just give guidance to our OEMs on that extreme-set-point TDP definition. SDP is a workload that represents a more mainstream workload. For the Y processor line we'll talk about SDP because that's really the primary design point that makes sense for the kind of devices that use those processors."
Soo..... if I want to use my Y-series tablet to do something processor-intensive it will burst into flames. Awesome.
 

Blandge

Distinguished
Aug 25, 2011
316
0
18,810
[citation][nom]ojas[/nom]Intel's provided both the TDP and SDP for their Y series chips, th y-series i5s for example had a TDP of 13w and an SDP of 7 or 8w, can't really remember. I think it was 7w.I don't see the point of this post, less why it's "news" since the y-series slide leak was about a month ago. A few people speculated then that since both the y-series and surface pro launch in Jan, the surface would most likely have a y-series i5, probably the only way to make it fanless.[/citation]

Actually, it doesn't work that way. In order for the SDP of the part to remain at 7W, the OEM must keep the CPU below 80 degrees Tjmax (Max junction temp). Therefore, in order to rate a product at 7W TDP, the device must never go above 80C. That is directly tied to device cooling, so the device needs BETTER cooling in order to achieve 7W SDP. Passive cooling likely will not work.
 
G

Guest

Guest
True Blandge, but Intel's point is that in tablet mode, you won't be stressing it to those heat levels. The device rates at 13W TDP (not 7W TDP). My guess is its heavily clock and voltage regulated to keep temperatures down... perhaps its dyamic enough to know when it is connected to a base/plug, and when it is not.
 

ojas

Distinguished
Feb 25, 2011
2,924
0
20,810
[citation][nom]Blandge[/nom]Actually, it doesn't work that way. In order for the SDP of the part to remain at 7W, the OEM must keep the CPU below 80 degrees Tjmax (Max junction temp). Therefore, in order to rate a product at 7W TDP, the device must never go above 80C. That is directly tied to device cooling, so the device needs BETTER cooling in order to achieve 7W SDP. Passive cooling likely will not work.[/citation]
I don't think it works this way either. Temperature and power aren't co-related like this. And I presume by device you mean the actual chip, not the device as a whole.

What it means that, the cooling solution must be able to dissipate 7w of heat to keep Tj under 80*C, and 13w to keep Tj under TjMax. That is it, really.

You seem to be implying the inverse, but heat causes a change in temp and not the other way around. Allowing the temperature to exceed 80*C can't increase power consumption. Power consumed by a chip is given by:
P=C(V^2)f, where C is capacitance, V is voltage, and f is frequency of the circuit.

7w is SDP, not TDP...which you've corrected later...so i assume it was a typo :)
 

twelch82

Distinguished
Dec 8, 2011
182
0
18,680
Let's say that you had a processor with 100 cores, that were gated off so they would completely shut off and use no power when not needed. The cores can also throttle their speed significantly between 10-100%.

The max TDP for that processor would be the power consumption with all 100 cores running at 100%. For cooling purposes, the processor would need to be able to support that, so when you design the cooling system, it would have to take that TDP in mind. In practice however, maybe usually only 1 or 2 cores are active, and not running at 100%, so - just for the sake of illustration - let's say that means the processor uses only 2% of TDP in the average case.

Now you have another processor with just one core, but it is very power hungry, and - again for the sake of illustration - let's say it's the old kind that couldn't throttle down the clockspeed at all. It has a lower overall TDP than the 100-core processor because it only has 1 core, but it consumes that much power all the time. Let's say its max TDP is 10% of the 100 core processor.

For cooling purposes, this 2nd processor would be easier to cool, because you have a lower limit to worry about for thermal dissipation. But in the average case, it actually consumes 5x more power, despite having a TDP that's 10x lower. Now these numbers are never likely to exist in a real-world scenario, but hopefully you can get the picture of what Intel is trying to message.
 

whyso

Distinguished
Jan 15, 2012
689
0
19,060
[citation][nom]ojas[/nom]I don't think it works this way either. Temperature and power aren't co-related like this. And I presume by device you mean the actual chip, not the device as a whole.What it means that, the cooling solution must be able to dissipate 7w of heat to keep Tj under 80*C, and 13w to keep Tj under TjMax. That is it, really. You seem to be implying the inverse, but heat causes a change in temp and not the other way around. Allowing the temperature to exceed 80*C can't increase power consumption. Power consumed by a chip is given by:p=C(V^2)f, where C is capacitance, V is voltage, and f is frequency of the circuit.7w is SDP, not TDP...which you've corrected later...so i assume it was a typo[/citation]

I think what he means is that transistor leakage increases as temperature increases. Transistor leakage means power lost and therefore a temperature increase (positive feedback). A chip running at the same speed at 20 degrees centigrade will use less power than one running at the same speed but at 90 degrees, because leakage will be much lower.
 
[citation][nom]ojas[/nom]I don't think it works this way either. Temperature and power aren't co-related like this. And I presume by device you mean the actual chip, not the device as a whole.What it means that, the cooling solution must be able to dissipate 7w of heat to keep Tj under 80*C, and 13w to keep Tj under TjMax. That is it, really. You seem to be implying the inverse, but heat causes a change in temp and not the other way around. Allowing the temperature to exceed 80*C can't increase power consumption. Power consumed by a chip is given by:p=C(V^2)f, where C is capacitance, V is voltage, and f is frequency of the circuit.7w is SDP, not TDP...which you've corrected later...so i assume it was a typo[/citation]

Doesn't capacitance increase a little at higher temperatures, thus increasing power consumption a little, because of increased leakage at higher temperatures?

Also, doesn't that formula only apply to digital circuits and not all of a CPU is digital, so it doesn't account for quite 100% of the chip's power consumption?

Unless I'm mistaken, it seems like you're trying to oversimplify this a little.
 

ojas

Distinguished
Feb 25, 2011
2,924
0
20,810
[citation][nom]whyso[/nom]I think what he means is that transistor leakage increases as temperature increases. Transistor leakage means power lost and therefore a temperature increase (positive feedback). A chip running at the same speed at 20 degrees centigrade will use less power than one running at the same speed but at 90 degrees, because leakage will be much lower.[/citation]
I'm not sure he meant that, but i see what you mean and it makes sense...though i would suppose that...when you design the circuit you make allowances for all that. I can't be sure, because i'm not familiar with actually implementing and designing CPU archs.
[citation][nom]blazorthon[/nom]Doesn't capacitance increase a little at higher temperatures, thus increasing power consumption a little, because of increased leakage at higher temperatures?Also, doesn't that formula only apply to digital circuits and not all of a CPU is digital, so it doesn't account for quite 100% of the chip's power consumption?Unless I'm mistaken, it seems like you're trying to oversimplify this a little.[/citation]
Yup, i'm oversimplifying it, because i don't know CPU architectures in detail on a structural level, and unlesso ne of us is, i don't think we can do anything but oversimplify things based on very basic assumptions.
Like, for example, i don't think leakage current would increase power draw so much that you'll notice it in a big way. May be a few mW at best.

Plus these are tri-gate transistors, and none of us know much about them. I mean, assuming you've studied this stuff formally, i think you'd have studied about planar transistors. I'm just starting with the Digital Electronics course at college...so far we're still doing stuff like boolean algebra and logic gates which i've been doing since school. Microprocessors are still a year away.

Also consider the fact that...Haswell, despite having the same manufacturing process, has a higher TDP...
 

alextheblue

Distinguished
[citation][nom]twelch82[/nom]Let's say that you had a processor with 100 cores, that were gated off so they would completely shut off and use no power when not needed. The cores can also throttle their speed significantly between 10-100%. The max TDP for that processor would be the power consumption with all 100 cores running at 100%. For cooling purposes, the processor would need to be able to support that, so when you design the cooling system, it would have to take that TDP in mind. In practice however, maybe usually only 1 or 2 cores are active, and not running at 100%[/citation]That's a terrible comparison. We don't have 100 cores. We have 2-4 cores, and you know what? With an x86 Win8 tablet I can load up the cores and the GPU pretty good just by running some games. So you're saying the tablet shouldn't be able to do that without overheating? SDP just sounds like a poor marketing ploy, and sets a bad precedent.
 

twelch82

Distinguished
Dec 8, 2011
182
0
18,680
[citation][nom]alextheblue[/nom]That's a terrible comparison. We don't have 100 cores. We have 2-4 cores, and you know what? With an x86 Win8 tablet I can load up the cores and the GPU pretty good just by running some games. So you're saying the tablet shouldn't be able to do that without overheating? SDP just sounds like a poor marketing ploy, and sets a bad precedent.[/citation]

Overheating is handled by TDP. TDP needs to handle worst-case cooling requirements. SDP is to aid in estimating average power consumption.

The point is that some chips have a higher variance than others between the average amount of power they consume, and the amount of power they consume when fully loaded.

The theoretical 100-core chip was just to illustrate a more extreme version of a scenario, not to suggest that you can actually go out and buy one like that.
 

ojas

Distinguished
Feb 25, 2011
2,924
0
20,810
Intel determines a processor's TDP by running a few dozen workloads on the product and measuring thermal dissipation/power consumption. These workloads include individual applications, multitasking workloads (CPU + GPU for example) and synthetic measures that are more closely related to power viruses (e.g. specifically try to switch as many transistors in parallel as possible). The processor's thermal behavior in all of these workloads ends up determining its TDP at a given clock speed.

Scenario Design Power (SDP), on the other hand, is specific to Intel's Y-series SKUs. Here Intel takes a portion of a benchmark that stresses both the CPU and GPU (Intel wouldn't specify which one, my guess would be something 3DMark Vantage-like) and measures average power over a thermally significant period of time (like TDP, you're allowed to violate SDP so long as the average is within spec). Intel then compares its SDP rating to other, typical touch based workloads (think web browsing, email, gaming, video playback, multitasking, etc...) and makes sure that average power in those workloads is still below SDP. That's how a processor's SDP rating is born.

If you run a power virus or any of the more stressful TDP workloads on a Y-series part, it will dissipate 10W/13W. However, a well designed tablet will thermally manage the CPU down to a 7W average otherwise you'd likely end up with a device that's too hot to hold.

The best comparison I can make is to the data we saw in our last power comparison article. Samsung's Exynos 5 Dual (5250) generally saw power consumption below 4W, but during an unusually heavy workload we saw it jump up to nearly 8W. While Samsung (and the rest of the ARM partners) don't publicly specify a TDP, going by Intel's definition 4W would be the SoC's SDP while 8W would be its TDP if our benchmarks were the only ones used to determine those values.

As for whether or not 7W SDP parts will actually be any cooler running than conventional 10W/13W SKUs, they should be. They will run at lower voltages and are binned to be the lowest leakage parts at their target clock speeds. Acer has already announced a successor to its W700 tablet based on 7W SDP Ivy Bridge with a 20% thinner and 20% lighter chassis. The cooler running CPU likely has a lot to do with that.

Then there's the question of whether or not a 7W SDP (or a future 5W SDP Haswell/Broadwell) Core processor would still outperform ARM's Cortex A15. If Intel can keep clocks up, I don't see why not. Intel promised 5x the performance of Tegra 3 with a 7W SDP Ivy Bridge CPU. Cortex A15 should be good for around 50% better performance than Cortex A9 at similar frequencies, so there's still a decent gap to make up.

http://www.anandtech.com/show/6655/intel-brings-core-down-to-7w-introduces-a-new-power-rating-to-get-there-yseries-skus-demystified
 
Status
Not open for further replies.