"7 Watt" Ivy Bridge my arse!

[IntelUser2000]

I will give you 100%(2x) more IPC for the IvyBridge Core over Bobcat. Now, E-450 is at 1.7GHz, that is 2x higher frequency. I will say they will have almost the same performance.

Now add 15+% more IPC for the Jaguar and >10% more frequency at the same power budget and it seams to me that IvyBridge at 800MHz will loose big time.

According to here:

Core i3 3217U, 1.8GHz: http://www.notebookcheck.net/Intel-Core-i3-3217U-Notebook-Processor.74465.0.html

AMD E-450: http://www.notebookcheck.net/AMD-E-Series-E-450-Notebook-Processor.60138.0.html

E-450/Core i3 3217U/Advantage in % for Core i3

3DMark06 CPU: 1029.9/2138.3/107.6%
3DMark Vantage CPU: 1863.9/5058.7/171.4%
Cinebench R10 ST 32-bit: 1012.8/2379/134.9%
Cinebench ST 64-bit: 1137.4/3090.3/171.7%
Cinebench MT 32-bit: 1960.9/4984.5/154.2%
Cinebench MT: 64-bit: 2204.1/6319.3/186.7%
Cinebench R11.5 ST 64-bit: 0.3/0.8/166.7%
Cinebench R11.5 MT 64-bit: 0.6/1.7/183.3%
X264 HD Benchmark 4.0 Pass 1: 16.2/56.1/246.3%
X264 HD Benchmark 4.0 Pass 2: 2.9/8.7/200%

Total ST or lightly threaded(Cinebench and X264 Pass 1): 179.9%
Total MT(Cinebench, X264 Pass 2): 181.05%
Total Averaged: 173.3%

Umm, I think they will turn out pretty close. 800MHz might actually take few % lead because clock scaling isn't linear.

 

[Nemesis 1]

I still don't see what any of that has to do with my pointing out that IDC picked the power numbers to derive the voltages and not the other way around.

I be honest I paid little attention to what you guys were discussing on that issue , I certainly not going to read it. I discoverd my own little treasures when I went direct die cooling . I hope Haswell uses Tim . I want to remove the lid and do the same as I did with IB. The point was ANAND did not randomly pick a number from thin air. He recalled the Haswell energy consumsion video runing @less than 8 watts C-15 trying to stay under 4 watts but he got the full chip running and it used 8 watts before throttling.Yes these A-15 look really fast for power consumed

 

[Nemesis 1]

. Well its pretty clear this year is going to be interesting . It was tuff going before now with all these new players the fanclube armies will grow . It will be confussing thats for sure. I hope not as confussing as this topic

 

[Exophase]

I be honest I paid little attention to what you guys were discussing on that issue , I certainly not going to read it. I discoverd my own little treasures when I went direct die cooling . I hope Haswell uses Tim . I want to remove the lid and do the same as I did with IB. The point was ANAND did not randomly pick a number from thin air. He recalled the Haswell energy consumsion video runing @less than 8 watts C-15 trying to stay under 4 watts but he got the full chip running and it used 8 watts before throttling.Yes these A-15 look really fast for power consumed

If you're not even reading my posts you have absolutely no business telling me I'm wrong or lacking comprehension. I never said anything about Anand picking numbers from thin air. What I said was clarifying IDC's methodology and was not in the least bit some kind of commentary on actual power consumption from either IB or Haswell.

I really wish you'd read more and type less, this isn't your personal blog to just randomly post whatever is on your mind in the middle of threads. Your next comment about shorting nV is not only completely off topic but not even remotely relevant to anything anyone has said.

 

[Nemesis 1]

I am reading them I just not absorbing the info.Things have changed alot with my memory over the last 6 months , I removed that . befor you posted or befor I read this post. Well What Article was it you read , You know the one I kept referring to .

To be perfectly frank I thought this topic was over the top. IVB at 7 watts ya right. Knowing IVB is not SOC i thought it useless. ivb is a stop gap nothing more if i was intel i would tell ms to use amd in there pro model surface. If I was intel I would get friendlyer with google let AMD and MS huddle together. If I was intel I would not even use this chip.

The smart thing for intel because Apple being a big customer I would do some show boating , Intel wants on Apple phones . Playing hardball with MS . Might be the smart thing for intel to do .

 

[NTMBK]

To be perfectly frank I thought this topic was over the top.

As in this thread was a bit OTT, or the topic of Ivy Bridge a 7W has been overhyped?

 

[Piroko]

If that were true then why would Intel bother incorporating a reduction in max allowed operating temperature into the SDP spec in the first place?

Obviously the engineers who made the spec are expecting temperature to play a considerable role in the processor's power footprint, otherwise they would have left the power spec in place with the existing TJmax spec.

Oh I agree that temperature has a measurable impact on power consumption.
But -I'll take the i5 3439y as an example- to my understanding there is only one bin available which can be configured into two modes via Bios:
- TDP Nominal where it has to be able to run stable @ 2,3 Ghz and 105°C and
- TDP Down where it still has to run stable @ 2,3 Ghz and 105°C even if it might not reach these clocks due to cooling restraints.
And since I'm not aware that Ivy Bridge can adjust its voltage on its own if the temperature allows it I concluded that the temperature alone could not explain a >50% increase in Performance per Watt (let alone a 10% increase actually).

Again, if this were relevant then Intel would not bother to peg their TDP and SDP spec values to 105C and 80C respectively. Clearly they are expecting Tjunction to reach (and exceed) 80C such that they felt it necessary to split the spec to define power usage below 80C versus above 80C.

Yea, that probably wasn't a good example as appropriate cooling should be expected. But I don't see any technical specification which guarantees me a bin which will not exceed 80°C and 7W thermal dissipation need. The presentation sheet itself clearly states that TDP is defined at specific parameters and SDP is characterized at a given temperature and unknown load. That's a huge difference because you can't build a system around the SDP.

 

[Exophase]

- TDP Nominal where it has to be able to run stable @ 2,3 Ghz and 105°C and

Has Intel ever made a statement about a minimal set of conditions where a processor must be able to reach full turbo? Otherwise we don't really know the binning, and it may not even be the case that with only one core powered that it can always hit 2.3GHz. Or it may only be capable of sustaining it for a limited amount of time.

Otherwise all that's really known is the uarch could hit that speed (that's a given) under some voltage that the system must be capable of providing.

- TDP Down where it still has to run stable @ 2,3 Ghz and 105°C even if it might not reach these clocks due to cooling restraints.

That's the description I'd give for TDP nominal, where only the base clock is guaranteed in all realistic scenarios (strange power virus might still cause it to throttle)

I can't find anything about that confirms that cTDPdown is still attainable at the same base clock with just lower turbo headroom.

Not saying either one is the case; can anyone give me a good official description that clarifies this? Is it in the datasheets?

 

[Piroko]

The base clock is guaranteed in all realistic scenarios, true, but the data sheet clearly states that this processor can run at up to 2,3 Ghz stable and that it may get as hot as 105°C while trying to do so. I would think that Voltage must be high enough to guarantee stable operation at these extremes, no matter if they occur independently or together.

But that isn't what I was trying to point out, the stock Voltage at a given clock rate should be the same no matter if the Processor is at a chilly 45°C or close to 100°C. Take IDCs P(total)=[] and only change the temperature: The effect isn't nearly as dramatic as it would be if you additionally adjust for Voltage.

 

[mikk]

At 7 watts, Intel told us, the Core i7-3689Y runs at only 800MHzby default. Here's a crucial note, though: Intel tells us that the processor can still turbo up to that same 2.6GHz when the cooling system has available headroom.

http://www.theverge.com/2013/1/9/3856050/intel-candid-explains-misleading-7w-ivy-bridge-marketing


Base clock is just 800 Mhz in 7W mode or how does it work? Is there a different source? I would prefer to read something from Intel itself. Intel ark only shows up 1.5 Ghz base clock.

 

[IntelUser2000]

I can't find anything about that confirms that cTDPdown is still attainable at the same base clock with just lower turbo headroom.

Not saying either one is the case; can anyone give me a good official description that clarifies this? Is it in the datasheets?

Originally, Intel said that cTDPdown's Base frequency was equal to its LFM frequency, which for Ivy Bridge its 800MHz. But since then things changed. They've said that the 22nm process is better than expected, plus there's also that almost no one is using cTDP option, whether up or down(Sony Vaio Duo 11 is the only known one, and the i7 config uses the cTDPup).

Now even the 13W specced Ivy Bridge has a Base clock of 1.5GHz. That's pretty good considering they said 17W parts needed to run at 800MHz Base to get to a 13/14W cTDPdown level.

But for the cTDPup, the Base frequency goes up by 500MHz over Nominal. So one must believe there's a sacrifice in the cTDPdown for Base clocks as well. We'll see soon enough.

One important thing to note though: Intel has said that the top attainable frequency for all cTDP modes are same. The difference is the Turbo duration.

 

[Nemesis 1]

As in this thread was a bit OTT, or the topic of Ivy Bridge a 7W has been overhyped?

No I think the actual draw will be fine . and it is a big accomplishment by intel . wonderful work people. Its going add another long run production line for a cheaper product in the same market kind kill any weeds. But I would have gave all that up. Because it takes away from what an asskicking product the Haswell 2core soc soix GT3 is going to be. Massive volumns. I think this and a refind process on 3D gates is the greatest accomplishment to this date. But thats just me.

 

[Exophase]

Originally, Intel said that cTDPdown's Base frequency was equal to its LFM frequency, which for Ivy Bridge its 800MHz. But since then things changed. They've said that the 22nm process is better than expected, plus there's also that almost no one is using cTDP option, whether up or down(Sony Vaio Duo 11 is the only known one, and the i7 config uses the cTDPup).

Now even the 13W specced Ivy Bridge has a Base clock of 1.5GHz. That's pretty good considering they said 17W parts needed to run at 800MHz Base to get to a 13/14W cTDPdown level.

But for the cTDPup, the Base frequency goes up by 500MHz over Nominal. So one must believe there's a sacrifice in the cTDPdown for Base clocks as well. We'll see soon enough.

One important thing to note though: Intel has said that the top attainable frequency for all cTDP modes are same. The difference is the Turbo duration.

Okay, so what do you think the 10W cTDPdown means for the Y series? Save for the Pentium that has that as its base TDP and runs at 1.1GHz base clock w/o turbo. So it should be capable of handling that much (unless HT makes any real difference).

 

[Wark0]

From what Intel told us :

- 7 watts is the power consumption mesured on the "Y" Ivy Bridge CPU under light load wich are typical uses on a tablet (casual game, video playback, create light content). Intel did not want to be more specific...
- 7 watts is a power constraint, the CPU will not go above 7 watts except for a short period if the CPU is configured as such. What does it mean ? The CPU will throttle down to a lower frequency to maintain 7W under heavier load than the load used to evaluate the SDP. That means under the base frequency, as low as 800 Mhz if needed...

 

[Abwx]

Technicaly , they integrated the power comsumption curve ,
that is , they took the average (arithmetic or exponential )
sliding mean over a period that is computed to take account
of the cooling and casing thermal inertias.

As such , it can be made to run at higher power drain than
if only average power is taken into account.

For instance , it can run at 13W for 2 minutes and then
being iddled at 3W for 4 minutes , the resultant SDP will be 7W.....

Depending of the integration period you can set the SDP at any
convenient value , let s say 5W if you re assumming that the CPU
run at 3W during 80% of the time , you ll end with a 13W CPU
specified for 5W SDP.....

Quote myself cause i was fully right...

Hardware.fr made a good article with some infos they
managed to exctract from Intel.

SDP is the TDP for a moderate usage scenario.

http://www.hardware.fr/focus/78/ivy-bridge-7-watts-quelques-details-sdp.html

 

[Wark0]

(Hello abwx, i am Marc from Hardware.fr )

 

[Abwx]

(Hello abwx, i am Marc from Hardware.fr )

Salut Marc , le monde est petit.

Keep on the good work , anyway.

 

[Idontcare]

From what Intel told us :

- 7 watts is the power consumption mesured on the "Y" Ivy Bridge CPU under light load wich are typical uses on a tablet (casual game, video playback, create light content). Intel did not want to be more specific...
- 7 watts is a power constraint, the CPU will not go above 7 watts except for a short period if the CPU is configured as such. What does it mean ? The CPU will throttle down to a lower frequency to maintain 7W@80°C Tjmax under heavier load than the load used to evaluate the SDP. That means under the base frequency, as low as 800 Mhz if needed...

Not to nitpick for the sake of nitpicking, but we've been trying to keep this clear in the this thread to avoid confusion down the road, you wrote "7W@80°C Tjmax" when really mean "7W@80°C Tj", correct?

The TJmax for the processor is still 105°C, but if you operate the processor at or below a junction temperature of 80°C then the 7W SDP value is valid. But nothing in the processor's throttling circuitry actually restricts the junction temperature to remain at or below 80°C, does it? (I don't believe it does but I do not have confirmation) Whereas we know the chip will throttle as needed to restrict the junction temperature to remain at or below 105°C because that is the preset TJmax value.

 

[Wark0]

My mistake, in fact it has to be just 7w, that's the power consumption under the light load that was characterized @ 80°C. As you said Tjmax is still 105°C.

Like already said on this thread SDP is like a cTDP down-down.

 

[SevenThirty]

My mistake, in fact it has to be just 7w, that's the power consumption under the light load that was characterized @ 80°C. As you said Tjmax is still 105°C.

Like already said on this thread SDP is like a cTDP down-down.

So if the processor can still operate at up to 105°C, then it's not really cTDP down-down, it's still just cTDP down isn't it? And it's still just a processor measured at 105°C to be above 7W, and probably close to... 10W. So if it doesn't throttle at 80°C, and it's still just a 10W cTDP processor, then SDP is just marketing. The fact that better cooling would be needed to reduce temperatures goes 180° to intels attempt to spin this as SDP is for ultrathin/tablet, since a smaller heatsink would reach TjMax (whatever it may be set to) earlier. So you take a 10W cTDP, already a gimped version of a gimped Y series, then you lower TjMax from 105°C to 80°C, and to top it off, throw a smaller than necessary cooler on it, because it's only "7W", so it can now reach Tjmax even sooner. Sounds like a win-win-win in my book. Can't wait to keep my hands warm in the winter.

 

[cbn]

You do know that Sandy Bridge's Turbo Boost 2.0 relies on the fact that thermal is slower than power? The difference is what gets the greater Turbo over Nehalem/Westmere. But eventually that runs out and they settle at whatever the constant TDP limit is. I guess depending on the situation it can be some Turbo active, or even just Base.

Apparently you can set the duration for the exponentially weighed average power(probably firmware for the manufacturers). It goes from 32 seconds for the ULV to 64 seconds on the dual core and quad. You can go as little as 1 second.

Today you have a base clockspeed and then turbo-bins above that provided you are fitting within a pre-defined TDP at TJmax.

Tomorrow you will have a base clockspeed and "deturbo-bins" below that for which the processor will underclock below the base clockspeed if the processor has been told to do so (the same as we tell our processor to let the clockspeed rise upwards when we enable turbo-core/boost at the platform level).

And just like with today's processors and 3rd party cooling solutions where you can spend more money, enable better cooling, and your turbo-core/boost remain enabled and active for far longer durations than if you simply relied on the stock HSF. So too for the "deturbo" settings, a superior cooling solution will enable the processor to run at less deturbo settings (closer to stock base clockspeed) than it might otherwise have to declock with other less effective coolers.

It is all two sides to the same coin.

Great posts.

It would be interesting to find out how much variation exists in turbo boost duration for the various ULV laptops?

 

[SlickR12345]

No kidding. Of course the official numbers are at least 10%-15% lower than they actually are.

 

[Idontcare]

It would be interesting to find out how much variation exists in turbo boost duration for the various ULV laptops?

Judging by the fact that the DIY market is now flooded with mobo's that are setup to configure "stock" turbo-bins to always be on (my 3770k runs at 3.9GHz all cores, at full load even in LinX, stock HSF), regardless the TDP constraints, I would be surprised if the various ULV laptops truly adhere to the specs for turbo boost duration themselves.

In other words I have no doubt you would see a rather striking distribution in results from one laptop to another, but the distribution would not be cleanly attributable to processor variation (raw capability) or implemented cooling solution (laptop model specific) because the data would be convoluted by bios writers who intentionally set the stock/auto parameters to ignore certain spec guidelines set by Intel. (ASUS is notorious for this)

 

[podspi]

At least with AT they try to point out when mobos do this.

I don't think end-users care too much (especially in a laptop) what shenanigans BIOS writers pull to come out ahead on the benchmarks. If it comes at the expense of battery life (and the reviewers make a note of that) people will care.

Unfortunately, if you walk into best buy they quote you running times of the machines at idle, screen off...

 

[NTMBK]

Anandtech's take on it, for anyone who missed it:

At the end of the day, 7W SDP Ivy Bridge (and future parts) are good for the industry. Intel should have simply done a better (more transparent) job of introducing them.

http://www.anandtech.com/show/6655/...-rating-to-get-there-yseries-skus-demystified

Not sure if I entirely agree with the last bit; "more transparent" implies that they were just a bit opaque about what SDP was, whereas they instead failed to mention SDP at all in the presentation which felt like intentionally misleading marketing speak.