Ivy Bridge models and clocks leaked

[BrightCandle]

For the average PC World buyer the onboard GPU is a lovely idea, because they can't cut the graphics back any further than this minimum. But a discrete GPU with its own power budget is going to be faster for the foreseeable future. Anyone serious about PC gaming is going to buy a card and not rely on the bare minimum provided by AMD/Intel on the CPU.

Power and transistor budget is being spent not on improving the performance of the CPU but on a GPU that most of us don't use. I would have preferred to see 6 cores rather than more GPU, because at least in the rare circumstances when I need them I am actually using them, unlike the iGPU when I am never using it. SB-E and then IB-E are the only real performance options, its a real shame they are lagging so much behind the mainstream CPU releases that bring the architectural improvements and hence raw single thread performance.

What I would really like to see is a CPU designed only to get maximum performance from a single thread. One core and all the transistors thrown at that problem. It wouldn't compete on performance/watt but having it requiring watercooling isn't ridiculous anymore with the boxed sealed loops on offer.

 

[greenhawk]

It makes sense that IVB would get ~18% better clocks at the SAME TDP. But the 3770K is not the same TDP, its lower. It wouldn't suprise me that if they made a 95W IVB chip that it could hit ~18% higher clocks.

+1, intel just traded performance for power. nothing else.

 

[IntelUser2000]

I meant in general, but I guess the average Joe doesn't care about more CPU power. Maybe Intel doesn't want to cannibalize their Sandy-E line.

If the TDP's were the same as SB's then such a line of thinking might be plausible, but given they dropped the TDP's while still hitting the clockspeeds tells me these bins are being purely driven by marketing and not by capability.

I expressed my concerns on the 3820 chip being clocked so low and being released in the same time as Ivy Bridge: http://forums.anandtech.com/showpost.php?p=32410301&postcount=18


I have another theory. They want us to shift to laptops.

Top SKU on mobile:
2960XM:
Base: 2.7GHz
Sustained 4 core: 3.0GHz
Temporary 4 core: 3.5GHz
2 core: 3.6GHz
1 core: 3.7GHz

The second fastest is the 2860QM with 200MHz off everything above. And 3rd is 2760QM with 100MHz off 2860QM, with 6MB L3 and cost of $378.

If they can up the performance(using Clock+IPC) on the 2760QM/2860QM/2960XM by 20%, the stock performance of mobile chips will essentially equal desktop chips. The configurable Turbo allowing availability of a 65W part supports this theory.

HD4000 top integrated gpu

earlier info said it had more render units or whatever, but the clockspeeds look unchanged;

It does, each EUs are twice as powerful as the previous generation and there's 33% more of them. There's also twice as many texture units. Reduced clock tells why they were estimating 60% gains rather than 80-90% that would come with truly doubling it(rather than lowering clocks).

 

[greenhawk]

but I guess the average Joe doesn't care about more CPU power.

semi true. People have been saying that since 1Ghz was common. Even before that. But people keep buying and CPUs keep getting faster.

 

[LoneNinja]

They have a better IGP, something that may compete with Llano but I doubt will match Trinity. They have lower TDP, why increase clocks when they already humiliate anything their competition(AMD) is able to produce. They'll probably clock better than Sandy Bridge for the overclocker anyway.

 

[IntelUser2000]

They have a better IGP, something that may compete with Llano but I doubt will match Trinity.

I don't think it'll match either. Even the chief architecture for Ivy Bridge graphics said they'll be behind in this generation. But it'll be pretty close to Llano, if not on par.

And the prospects for Ivy Bridge on mobile is quite fantastic. I will have to seriously re-think going with Ivy Bridge on my desktop and getting a laptop instead.

What I would really like to see is a CPU designed only to get maximum performance from a single thread. One core and all the transistors thrown at that problem.

Pollak's Rule and the fact that even using dual cores have quite a few scenarios where it falls significantly behind a quad core makes all-out single core not feasible.

 

[Hevy187]

IVB is a power efficiency upgrade from SB.. That is all. What were u expecting? A 4ghz stock clock when they have 2 crowned platforms and zero competition?

There is no need to release anything that would trump on Sandy at this point.
Think about it, would you? Or would you bleed your kings of the hill before releasing your succeding king of the hill. Its a marketing strategy.
It would hurt them to release a harder performer more than anything.


Sent from iPhone4s using Tapatalk

 

[Nemesis 1]

I'm probably missing the context of Lol_Wut_Axel's claims, but I was under the impression that Intel themselves were guiding us to understand that their 22nm process enabled ~18% higher clockspeeds at the same (or better) power-consumption as their 32nm process?

This assumes a number of things regarding the differences between IB and SB in terms of xtor counts and so on, but fundamentally speaking the clockspeed improvment potential is there according to Intel.

Looking at the charts IDC . I don't think so . Notice the charts showing the 3700s the 45 and 65 watts are exactly the same other than clocks . 20 watts for that amount of clock speed is a bit alarming. If you look at the chart your using . Its based on Voltage the higher the voltage the less advantage 22nm gains so at 1.37 . continue your chart see were they intersect. Also why they listing 1333 for memory . 1600 is suppose to be the base . So why have 1333 in the figure . I see the IGP clocks are down to 1150 from 1350 which is suppose to be the case. This will be the same as intels 32nm were a new stepping will come out.

 

[IntelUser2000]

Looking at the charts IDC . I don't think so . Notice the charts showing the 3700s the 45 and 65 watts are exactly the same other than clocks .

Compare the 3500 series with the 2500:

3570S: 3.1GHz
3570T: 2.3GHz

2500S: 2.7GHz
2500T: 2.3GHz

"But mommy I thought 22nm allowed better gains on lower voltages?"

These positionings are purely arbitrary. It's likely everything BUT technical.

Also why they listing 1333 for memory . 1600 is suppose to be the base .

DDR3-1600 and DDR3L-1333?

 

[Olikan]

IVB is a power efficiency upgrade from SB.. That is all. What were u expecting? A 4ghz stock clock when they have 2 crowned platforms and zero competition?

There is no need to release anything that would trump on Sandy at this point.
Think about it, would you?

yes, i would. If ivy bridge just uses 20 watts less than a sandy bridge, and can't overclock better....there would have no point for a upgrade.
Intel competes with it self

 

[Nemesis 1]

Compare the 3500 series with the 2500:

3570S: 3.1GHz
3570T: 2.3GHz

2500S: 2.7GHz
2500T: 2.3GHz

"But mommy I thought 22nm allowed better gains on lower voltages?"

These positionings are purely arbitrary. It's likely everything BUT technical.



DDR3-1600 and DDR3L-1333?

OK I see were the 1333 memory comes from .

But in the 2 charts above

3770s is 3.10 to 3.9ghz .

3770T is 2.50 to 3.7 ghz both 4000 IGP .


The top charts shows 3700K and 3700 and there base freq. is way higher other differances also . Intel did show in its roadmap a 95 watt part . But if this is correct . I will skip 22nm IB . The chances of this being correct are slim and none . Looks like home made charts to me . Weres the 95 watt part . Weres the 95 watt part?

 

[IntelUser2000]

I wonder if the reasoning is that "desktop users can easily overclock". And that some level of overclocking is expected.

http://vr-zone.com/articles/ivy-bri...e-base-clock-but-not-as-we-know-it/13154.html

As such we can expect a standard base clock of 100MHz that is said to have a "fine-tuning range" of +/-5MHz. It's not know what other base clocks Intel will implement, but a qualified guess would suggest that we're looking at 133 and 166MHz at least, but we'd also expect a few options in-between.

So even non-K SKUs will get some overclocking with limited bus options.

The top charts shows 3700K and 3700 and there base freq. is way higher other differances also .

Nemesis, it doesn't matter. The decisions are likely purely arbitrary. The reason I compared 3500 and 2500 is because they both have T and S SKUs, making it directly comparable, while for 3700 vs 2600 only Ivy Bridge has both options, making comparisons futile.

You can see most of the T and S SKUs have about 10% clock speed gain over Sandy Bridge, while the regular voltage ones gain nothing.

 

[Edrick]

yes, i would. If ivy bridge just uses 20 watts less than a sandy bridge, and can't overclock better....there would have no point for a upgrade.
Intel competes with it self

For people who care about power/performance it would. And also people who care about the +60% IGP performance increase.

 

[Genx87]

Looks like the raw performance will probably not be much greater than SB. But the thermals will be pretty significant. And thanks to AMDs incompetence. Intel has no incentive to bring out high clocked IB.

 

[Denithor]

IVB is a power efficiency upgrade from SB.. That is all. What were u expecting? A 4ghz stock clock when they have 2 crowned platforms and zero competition?

There is no need to release anything that would trump on Sandy at this point.
Think about it, would you? Or would you bleed your kings of the hill before releasing your succeding king of the hill. Its a marketing strategy.
It would hurt them to release a harder performer more than anything.

Exactly. IB isn't designed as higher performance although they are taking advantage of a portion of the reduced TDP to bump the GPU up somewhat.

They're smart enough to realize that if they put out a CPU that performs nearly the same to slightly faster on a smaller process node they get a better margin at the same selling price. Without real competition there's simply no drive to increase the power of these chips.

 

[TheWay64]

Anandtech previously said in an article that Ivy Bridge is going to have IPC increase of 4 to 6%, and GPU is going to be up to 60% faster.

Seems that they are focusing more on GPU performane and power consumption rather CPU performance. However it will still be better than AMD Crappydozer

 

[RyanGreener]

I think Intel is being smart about this. Why have more powerful processors when you can just make them become more energy efficient? Although it seems they are really focusing on the GPU aspect now so they're going to probably catch up to Llano or surpass them soon enough.

 

[Nemesis 1]

Nemesis, it doesn't matter. The decisions are likely purely arbitrary. The reason I compared 3500 and 2500 is because they both have T and S SKUs, making it directly comparable, while for 3700 vs 2600 only Ivy Bridge has both options, making comparisons futile.

You can see most of the T and S SKUs have about 10% clock speed gain over Sandy Bridge, while the regular voltage ones gain nothing.
__________________

As I read the 2 charts I assume that the top chart is for Desktop and the bottom one for notebooks . I am not buying into an intel 65watt notebook part when there are so many listed . Plus this is not what intel told us . 35 watts and below is what were suppose to get . If those charts are correct . These cpus are worse than SB and As far as new release BD did a better job . Weres the 95 watt part????

 

[Riek]

I think Intel is being smart about this. Why have more powerful processors when you can just make them become more energy efficient? Although it seems they are really focusing on the GPU aspect now so they're going to probably catch up to Llano or surpass them soon enough.

can you define 'soon'? Because IB i still months away and won't have faster graphics than the current llano within the same TDP. (ofcourse cpu power is vastly different between those 2, but that is nothing new either).

I'm more intriged about how well the gpu will support dx11 and openCL and what the results will be.

 

[ed29a]

As I read the 2 charts I assume that the top chart is for Desktop and the bottom one for notebooks . I am not buying into an intel 65watt notebook part when there are so many listed . Plus this is not what intel told us . 35 watts and below is what were suppose to get . If those charts are correct . These cpus are worse than SB and As far as new release BD did a better job . Weres the 95 watt part????

Bottom chart is not made up from notebook CPUs, they are the T and S CPUs desktop versions. According to this and this, IB will have a maximum of 77 watts TDP.

 

[Nemesis 1]

Not sure!!

 

[Nemesis 1]

Bottom chart is not made up from notebook CPUs, they are the T and S CPUs desktop versions. According to this and this, IB will have a maximum of 77 watts TDP.

Maybe. But I will bet anything with anyone the top K model will be 95watts. As for those 2 charts both showing desktop parts . BS.

 

[gmaster456]

I want that Dual Core i5.

 

[ed29a]

Maybe. But I will bet anything with anyone the top K model will be 95watts.

Doubt, they will probably use 95 watts as TDP for IB-E.

As for those 2 charts both showing desktop parts . BS.

BS? The SB quad core mobile CPUs are capped at 55 watts. And that is the extreme version, the others are 45 watt. You really think that the IB mobile quad core with a die shrink is going to consume 65 watts? I have a bridge for sale!

 

[LumbergTech]

Absolutely not true.

I used to be a big overclocker back in the day, but these days I prefer stability and ease of mind. Plus, I don't have the time that I did before to sit around and tinker for weeks to find that just right combination.

you're doing it wrong. Maybe you just expected unreasonable results...

when i bought my pc, i set my FSB and away i went with my 2.66 ghz chip @ 3.4 gz..

i put the proper cooling on there and havent ever had to fiddle with it in over 2 years