Mainstream Arrandale platform (32nm cpu+45nm gpu) to have 35W TDP

Down from TDP of Centrino 2 platform

Nvidia to be in a lot of Calpella notebooks due to most notebooks should have switchable graphics.

 

You get a Woot! from me .

 

Every year new lappy

 

I hope in Q3 09 they will actually be for sale... but I don't really believe Fud.

I also don't agree w/ their statement that nVidia has had more success in 40nm since ATI already has rolled out 40nm parts and their discrete GPU sales are actually better now. Hopefully we'll see some good stuff at Computex 2009 in Taipei (June 2-6).

I wish they listed how much power the PM45 chipset consumed, so we could get an accurate comparison of how much the X4500 consumes (since GM45 is essentially the same as PM45 plus X4500). 12W seems pretty high as ATI's discrete HD 4330 GPU on 55nm tech consumes <7W and kicks the snot out of the X4500 by at least a factor of 3-4x.

Also, hope the clock speeds for the new CPUs aren't as low as they've been reported before on NBR.

 

^^^ Other sites, not just Fudzilla, have been saying the quad cores are coming in October (Q2-4) for months.

 

I hope its true but the final say is still Intel and just because it has been said before, doesn't mean it is true... even last year's release of Cantiga was Q3 08 but when did companies start selling them? Q4 at the earliest. Preorders were for months...

 

so...how does the arrandale platform compare with today's current montevina platform in terms of total power usage?

 

Not much difference between notebook with integrated graphic card.
My netbook has TDP of about 10-15W.

 

Ah okay. I was actually wondering which platform would be more energy efficient, since I'm thinking about picking up a 13 inch notebook w/ integrated graphics, and I'd like as long battery life as possible.

 

If you get a power efficient 13" notebook with integrated graphics, it'll be more power efficient than "main stream arrandale platform.".

 

No, Arrandale is much more power efficient since the processor is 32nm, and the graphics is shrunk from 65nm to 45nm. And Arrandale has new power saving features not available on current Core 2 processors.

A 13" with Arrandale will use less power than a current 13" with a Core 2 and integrated graphics.

Clarksfield notebooks to use more power than current laptop quads due to no compatability with integrated chipsets.

 

The PM45 chipset has 7W TDP and GM45 chipset has 12W TDP. I don't believe the 7W TDP for the 4330, unless its just the chip only. Since discrete graphics cards also have another PCB and memory, it would be more than that regardless.

Anyways, Calpella is Q3 and IGP variants are early Q1 next year. I don't think we'll see Calpella earlier than when Montevina released last year because manufacturers are still trying to clear out inventory. Montevina got delayed so its going to affect Calpella. I'd expect August-September release.

And the low clock speed Quad Core Clarkdales are only 35W, so you can't really compare them to Arrandale. It looks like Arrandale is a Core 2 Duo 25W + IGP replacement.

The CPU part will use less power but I'm not sure about the GMCH part. The graphics core on Arrandale will be significantly more powerful than the GM45. Upside is of course graphics performance will be much better.

 

http://en.wikipedia.org/wiki/Nehalem_(microarchitecture)

I've noticed the wiki for Nehalem generation CPU has been slightly changed. Apparently the flagship mobile quad will be operating only at 2 Ghz?

I wasn't expecting a ridiculously highly clocked quad, cos new platform needs fine tuning, look back to late 2007, original C2D mobile chips maxed out only at 2.33 Ghz (T7600), and we have 3.0 Ghz T9900 now. But I was hoping for something that can at least match the current gen CPU clock frequency.

Merom managed to raise both clock frequency, and efficiency quite a bit over the Core duo chips, while not increasing price much. I doubt that same can be said for Clarksfield. The new architecture comes with higher efficiency, though lower clock, and perhaps more heat? at higher cost than the first gen C2D mobile chips, is it one step forward, and too many back? Those i7 chips do run hot, I know, a 2.66 ghz desktop 920 chip still have TDP of 130W. But I just get my head around the new 2 Ghz mobile quad. Provided its probably more efficient per clock cycle, is that even going to match QX9300? Beside, put a i7 965 into Clevo 900F, then its exactly twice as powerful as the slowest mobile quad. Just as mobile C2Q started to catch up with desktop chips, the gap once again appears...

I'm really not convinenced of the next gen mobile chips, at least those first wave ones. These are all speculation though, won't know for sure until Q3.

 

The thing that could save the quads from their low clock speeds is turbo boost, which ratchets up the CPU speed by a multiplier on all active cores if it isn't running too hot. If only a single core is active I seem to recall reading the multiplier can be kicked up by as much as 8, meaning a full extra GHz in single threaded apps. We'll have to wait and see performance benchmarks before knowing if this pans out at all in the real world.

 

That could explain it. Though their last attempt with intel dynamic acceleration on santa rosa didn't make much difference. An extra Ghz sounds pretty awesome.

 

Nooo, its because not all versions are shown yet. The low clocked quad cores are the 35W parts. If the TDP is in line with current quad cores it should be 55W since it has a northbridge integrated. We'll probably see 2.53GHz parts when they release. Always reserve final judgment when the product releases.

 

But 2 ghz one is already reportedly $1054, I doubt there would be higher end one.

 

Arrandale is more power efficient, but it doesn't mean it use less power.
E.g. a 20W flourescent light bulb with 70% efficiency vs a christmas incandescent light at 0.5W at 10% efficiency. The example is a bit dramatic.. but never the less, it's a pretty accurate representation.

A current generation core2duo P series is 25W processor plus around 10W for chipset+integrated graphic card and around 10-20W for other devices. When idling, the power figure could be as low as 15W for the whole computer.

With Arrandale platform, everything be a little bit more efficient thus reducing the power to match existing "mainstream" power requirements while offering a little bit faster speed and capabilities.

 

They can just offer a higher wattage part with similar pricing. That's what they do with the LV and ULV CPUs. The clock speeds are lower but the pricing is more in line with faster SV CPUs.

On Lynnfield/Clarkdale the Turbo Boost is 1 speed grade for 4 cores, 4 for 2 cores and 7 for 1 core active. Even if we consider that the Turbo Boost is active 100% of the time, it would be a loss in clock speed with quad core operation.

See I think the plan is to bring quad cores more mainstream to laptops. In Nehalem and Westmere generation, the dual cores will be a value product. They gotta somehow fit the middle. They might actually lose IGP marketshare with Nehalem as there's no IGP option for the quad core versions, even though the 3D capability of the IGP on Arrandale is going to be much more potent.

 

Indeed. I finally figured out where it actually makes a difference - if you disable a core in the BIOS. Then the remaining core will run at an extra multiplier with no complaints. Other than that I've almost never seen RM CPU Clock report my CPU as having a core running at 2.4 GHz (versus 2.2 stock).

 

The dynamic overclocking feature in Nehalem generation will be more reliable because it has integrated PCU that manages power management more efficiently than the one in Penryn.

http://www.xbitlabs.com/articles/cpu/display/intel-core-i7_9.html#sect0

In Core i7, it just works! That's the reason they resorted to using a hardware managed system since Windows interfered too much with it.

 

How about disabling a core in Windows? Does it give you that boost as well?

 

You can just set priority to a program then it will work. The reason that the current Penryn dynamic overclocking doesn't work well is because windows bounces threads around. If you set priority to the program and its a single thread application, you'll get the boost.

 

What if you turn a core off from windows, what will happen?

 

I'm pretty sure it'll work, but why would you want to do that as you are wasting another core for multi thread apps.