"7 Watt" Ivy Bridge my arse!

[Haserath]

Speaking of removing turbo from the iGPU, I wonder how that will affect performance in an ultrabook vs. tablet.

The tablet should have a lower resolution right? (Therefore less need for iGPU turbo)

The ipad and Nexus 10 have a higher resolution than my 1080p desktop.

Anyone want to wager a guess how much power a 7 watt SDP processor uses if the OEM puts a very small cooler on it and instead uses the TjMax 105 C standard?

7.5 watts?

8 watts?

More than that?

Maximum power use is TDP under the worst case.

I assume
SDP= Max @80C
cTDP= Thermal throttling@105C(ie base clock)
TDP= Maximum power use

 

[NTMBK]

Yet a conclusion is still drawn?

The conclusion is that it is an undefined metric we don't know the meaning of, that Intel insists on comparing with the TDP metric for its other processors in order to make it look better than it really is. So, yes.

 

[Charles Kozierok]

The conclusion is that it is an undefined metric we don't know the meaning of, that Intel insists on comparing with the TDP metric for its other processors in order to make it look better than it really is. So, yes.

That, plus again, Intel deliberately made no mention of this new metric in their PR materials.

If the point of SDP was really a "more fair" way of measuring power, Intel would have been up-front about it. That they were not shows quite clearly the real intention of this metric.

 

[ShintaiDK]

The conclusion is that it is an undefined metric we don't know the meaning of, that Intel insists on comparing with the TDP metric for its other processors in order to make it look better than it really is. So, yes.

So you still draw a conclusion, yet you even state yourself that you know nothing about what SDP is?

Honestly? We have no idea what SDP means.

 

[ShintaiDK]

That, plus again, Intel deliberately made no mention of this new metric in their PR materials.

If the point of SDP was really a "more fair" way of measuring power, Intel would have been up-front about it. That they were not shows quite clearly the real intention of this metric.

Did you miss the footnotes on the slide?

 

[Charles Kozierok]

Did you miss the footnotes on the slide?

Once again, I am not talking about the slides. I'm talking about their press release. The releases are what are used by the media.

Go to Google News, and see what's being reported. Dozens of stories about Intel's "new 7 watt chips". None of them mentioning "SDP" except for a handful that mostly reference the Ars piece.

I repeat for your benefit the following.

--

Here is Intel's press release from yesterday.

Here's how they are presenting these new chips in terms of power:

Since mid-2011, Intel has led the industry in enabling Ultrabook devices aimed at providing new, richer mobile computing experiences in thin, elegant and increasingly convertible and detachable designs. To enable these innovative designs, Intel announced last September that it added a new line of processors to its forthcoming 4th generation Intel Core processor family targeted at about 10 watt design power, while still delivering the excellent performance people want and need.

Skaugen announced today that the company is bringing the low-power line of processors into its existing 3rd generation Intel Core processor family. Available now, these chips will operate as low as 7 watts, allowing manufacturers greater flexibility in thinner, lighter convertible designs. Currently there are more than a dozen designs in development based on this new low-power offering and they are expected to enable a full PC experience in innovative mobile form factors including tablets and Ultrabook convertibles. The Lenovo IdeaPad Yoga* 11S Ultrabook and a future Ultrabook detachable from Acer will be among the first to market this spring based on the new Intel processors and were demonstrated by Skaugen on stage.

Notice the deliberate mention of "design power" in the context of Haswell chips. And then, in the very next paragraph, they move on to discuss these new IBs by talking about them being "as low as 7 watts".

No mention of what those 7 watts are, or that they're using an entirely new marketing-speak measure. No mention of "SDP" in the paragraph. In fact, there's no mention of it anywhere in the release.

This is actually even more deliberately deceptive than I thought it was when I read the Ars piece, before I found the PR.

 

[Haserath]

"These chips will operate as low as 7W"

 

[Haserath]

And who really cares... You'll be buying a finished product with this inside it. Not the chip itself.

OEM's know what it means... That's all that really matters for this.

 

[Charles Kozierok]

And who really cares...

I do.

I have yet to hear a single valid explanation for why Intel conflated TDP and SDP numbers in their press materials, other than that they wanted people to think these chips had a 7W TDP.

There's example after example of how this stuff is being used to deceive journalists, and thus, consumers.

Here's another one:

But the new drop is happening in Ivy Bridge, with a drop down to 7 (SEVEN!) watts, coming in even lighter than the original expectation of 10w for Haswell. They're shipping today, instead of waiting for the newer Haswell chips later this year.

 

[ShintaiDK]

Notice the deliberate mention of "design power" in the context of Haswell chips. And then, in the very next paragraph, they move on to discuss these new IBs by talking about them being "as low as 7 watts".

Can you show me the press releases that specify TDP as the maximum power consumption for the chip at 105C Tjmax?

 

[Charles Kozierok]

Can you show me the press releases that specify TDP as the maximum power consumption for the chip at Tjmax?

I can't respond to this obvious attempt at deflection without violating forum rules.

I will note that for the fourth or fifth time you have refused to address my question about why Intel published TDP and SDP numbers in the press release without making clear that the Haswell and IB figures use different metrics.

 

[Haserath]

I do.

I have yet to hear a single valid explanation for why Intel conflated TDP and SDP numbers in their press materials, other than that they wanted people to think these chips had a 7W TDP.

There's example after example of how this stuff is being used to deceive journalists, and thus, consumers.

Here's another one:

That is what they wanted to do by the looks of it, because SDP isn't thinner and lighter since they need lower temperature to do that.

It's not completely straight forward, but they aren't lying. If somebody wants to know what the SDP or TDP is, they can find it.

At least they didn't change TDP to 7W and say that's what it is!

 

[Charles Kozierok]

That is what they wanted to do by the looks of it, because SDP isn't thinner and lighter since they need lower temperature to do that.

It's not completely straight forward, but they aren't lying. If somebody wants to know what the SDP or TDP is, they can find it.

At least they didn't change TDP to 7W and say that's what it is!

Well, they aren't overtly lying, but they are being deceptive.

Still, thanks for at least acknowledging that.

 

[Haserath]

SDP might mean average use as well... Don't really know..

Intel might be wanting to do this to try and show they don't always use 10-13W. It's always good to have another metric as long as it's consistent.

 

[Idontcare]

In a nutshell, I would expect a processor with a 7 watt SDP to need a larger cooler than a 7 watt TDP processor. (But this cooler should still be smaller than a 10 to 13 watt TDP cooler)

That is it in a nutshell.

AMD did this already, in their efforts to get Phenom up to clockspeeds that were competitive while not totally blowing out the TDP spec they drastically reduced TJmax from the 90C's to something silly low like 62C.

It was all a maneuver to get the chip to sit within a TDP spec, and one way to do that is to limit the max allowed operating temperature so the bulk of the power used is going to dynamic power use and not leakage power losses.

^ if you need your power usage to "get small" then you reduce the max-allowed operating temperature accordingly.

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.

"These chips will operate as low as 7W"

It is the same with both these companies. AMD marketing claims "performance up to ..." which is a BS disclaimer that basically means "practically never, but Frank in Engineering swears he saw it happen once so we can technically claim it is a possibility the consumer will too, maybe".

Intel does the same with their "as low as" claims. As soon as you see those sorts of CYA marketing phrases come out, you know you are being sold an idea (an expectation) that is really the exception and not the rule but you it is being sold to you in a way that is designed to leave you with the impression that it is the rule and not just the infrequent exception.

 

[anikhtos]

well intel is pulling an advertising trick
so long in how many threads intel was praised for that the tdp was the worse case scenario for their cpus??? which will be hard to reach by a user.

but now intel is trying to compete with arm and it is annoying that the marketing is all about numbers.

after all if intel wanted to promote numbers could promote the numbers of performance and battery life of a device.

when you want to buy a tablet or ultrabook what you look for ???
performance and battery life and the weight.

so why confusing people with this new metric??
and scenario usage of today will not be the same as tomorrow.
so your cpu of spd 7w how much will consume with heavier loads???

and what about the cooling solutions for the cpu??
will be for the spd for the tdp???
a lot is to be answered but it is not a valid move from the side of intel.

 

[Charles Kozierok]

Here's a blast from the past:

Intel sees no value adding another specification to our processors. As noted above, ACP is not useful for system or processor thermal engineers and end users can get more accurate power consumption values by simply measuring the actual power of their server while running their specific application. In addition, AMD does not specify exactly where in their silicon process distribution they measure ACP, so it would be impossible for Intel to create an identical ACP specification.

 

[NTMBK]

So you still draw a conclusion, yet you even state yourself that you know nothing about what SDP is?

Yes, I am in fact partly drawing a conclusion based on that fact. Intel compared TDP of some devices against this lower metric which has no real meaning as we don't know how it is defined, and didn't tell us that it was doing this.

 

[Haserath]

I figured people would understand that as low as means 7-13W power use, but idle would be lower than this so it's hard to define what the chip uses at whatever moment(unless these leak more than desktop chips?).

7W could be done with enough cooling though... But then you might as well have a bigger battery.

 

[NTMBK]

I figured people would understand that as low as means 7-13W power use

Their press release and press conference never mentioned that these chips actually have a 13W TDP. So no, I wouldn't expect people to come to that conclusion.

 

[sm625]

After reading this whole thread I still do not have a definitive answer on what SDP actually is. But I am leaning towards Idontcare's analysis. So we can expect the same performance from a 7W chip with lower TjMax as we get from a 10W chip at 105 TjMax? The problem is we don't know how either of these chips perform!

 

[piesquared]

Well if intel's objective was to confuse the hell out of people, they've definitely succeeded. So this STD/STP/SDP of 7W is something that is achieved only under satisfactory/ideal conditions, and if it doesn't reach the advertised STD it is entirely the OEM's problem for not running a beefy enough cooling system. [sarcasm]That should work out well with intel's cool running 22nm node[/sarcasm] I never would have thought 'beefy cooling system' and 'tablets' would have to be used in the same sentence.
I guess it wouldn't be that bad if intel would have marketed it as a minimum 10W product instead of misleading consumers.

 

[Idontcare]

After reading this whole thread I still do not have a definitive answer on what SDP actually is. But I am leaning towards Idontcare's analysis. So we can expect the same performance from a 7W chip with lower TjMax as we get from a 10W chip at 105 TjMax? The problem is we don't know how either of these chips perform!

Well if intel's objective was to confuse the hell out of people, they've definitely succeeded. So this STD/STP/SDP of 7W is something that is achieved only under satisfactory/ideal conditions, and if it doesn't reach the advertised STD it is entirely the OEM's problem for not running a beefy enough cooling system. [sarcasm]That should work out well with intel's cool running 22nm node[/sarcasm] I never would have thought 'beefy cooling system' and 'tablets' would have to be used in the same sentence.
I guess it wouldn't be that bad if intel would have marketed it as a minimum 10W product instead of misleading consumers.

In a sense it is really no different from the concept of turbo-boost/core in which both companies advertise the availability of clockspeed and performance headroom above and beyond their base clocks "provided certain thermal and power limits are maintained".

It is all "limited case" scenario usage patterns for which these extraneous features actually pan out.

Intel is just flipping that same idea around and going the other direction with it, provided you are willing/able to keep the max operating temperature below a specific value then you can count on your processor's power footprint to fit within a smaller spec value.

 

[krumme]

I think what IB really brought to the table, compared to sb, was good 18w performance.

I still feel my ib ulv is a bit of a stretch, and it didnt come cheap that samsung 9 series x3c.

How much better is a 7w ib than the new kabini? I mean if its on average 30% faster, and still relatively high idle usage, for 500% ekstra cost there is even less than a niche market for it.

I dont think using a product that is optimal at say 25-35w for a 7w target is going to work in the new market. The consumer benefit is simply to small. And i think Intel knows that pretty well, and i read this move as more of a signal of intentions and future trends than something that is actually going to sell in several millions for this spring.

 

[Nemesis 1]

And if we're comparing Intel's chips to AMD's Temash and Kabini, don't forget that they still need a PCH at 3W: http://ark.intel.com/products/64342/Intel-BD82UM77-PCH For a 16W total TDP for the solution. Whereas Temash has an integrated FCH, so the chip's TDP covers both.

Yep this is 2006 . Intel already showed its cards and your debating IVB . Haswell will be out befor Temash . They already gave must the info about the SOC SOiX haswell. I doubt any recall what SOiX is and some clearly don't know what SOC is. Or somehow AMDs Temash SoC is better than intels when its highest C state is 6 weres as intel soix can get close to zerp usage state and be always connected, Temash can do none of this . also the 2 core haswell only slightly bigger than temash . Because of use of G3. 4 core haswell is 184mm2 so 2 core be about 160mm2. 2006 everyone doubted Intels word 7 years later history repeats itself and intel leaves egg on all faces .