Intel C2D power consumption benchmark from Xbitlabs

[thilanliyan]

click

The Xtreme edition C2D is using less than half as much wattage as the FX-62...very impressive. Remember, these numbers are for just the CPU. Some sites(such as AT) gave total system power consumption, although even in those tests you can see the difference.

I can't wait to get an E6600 hopefully and try to take it to 4 GHz.

 

[dexvx]

Mmm... you probably need to add 10-15W for the Core2 Duo figures. This is because the K8 has an integrateed memory controller whereas the C2D's do not. Nevertheless, your electricity bill savings will be astronomical!!!

 

[xtknight]

Fixed link

That's interesting. The C2D Extreme is using less than the C2D 1.83 GHz at idle. :shocked:

 

[thilanliyan]

Originally posted by: dexvx
Mmm... you probably need to add 10-15W for the Core2 Duo figures. This is because the K8 has an integrateed memory controller whereas the C2D's do not. Nevertheless, your electricity bill savings will be astronomical!!!

Huh?? That doesn't make sense. Why would you add that. That is the design of the CPU...you can't pick and choose certain hardware implementations to suit you. Can you measure the power consumption of an A64 independent of its memory controller?? You can't AFAIK.

That's like me measuring power consumption of an Athlon X2 compared to a single core A64 and then dividing the X2 number in half since I'm comparing to a single core.

Oh and the link was fixed. Thanks to xtknight.

 

[meksta]

what he's saying that if it'll be closer if you add in the power consumption of the motherboard chipset, because AMD has the memory controller built in. And he's right because you don't just run a PC with just a CPU and not a motherboard.

Anand shows that at idle, an X2 system still uses less power.

 

[kknd1967]

THG showed otherwise though.
For total power consumption, it really depends on the system config.

Originally posted by: meksta
what he's saying that if it'll be closer if you add in the power consumption of the motherboard chipset, because AMD has the memory controller built in. And he's right because you don't just run a PC with just a CPU and not a motherboard.

Anand shows that at idle, an X2 system still uses less power.

 

[hans007]

Originally posted by: dexvx
Mmm... you probably need to add 10-15W for the Core2 Duo figures. This is because the K8 has an integrateed memory controller whereas the C2D's do not. Nevertheless, your electricity bill savings will be astronomical!!!

that is not a really fair amount to add. there are entire chipsets that only use 10 -15 watts (well except ones made by nvidia apparently).

the memory controller for one thing is only one part of a northbridge. its probably 5 watts at most.

 

[thilanliyan]

Originally posted by: meksta
what he's saying that if it'll be closer if you add in the power consumption of the motherboard chipset, because AMD has the memory controller built in. And he's right because you don't just run a PC with just a CPU and not a motherboard.

Anand shows that at idle, an X2 system still uses less power.

See the problem with that approach is that different chipsets use different amounts of power, so while you may be able to get a consistent readings for an A64 system, you'll get varying results for the equivalent Intel system on different boards.

However, i sort of agree, a system power comparison with as few variables as possible is probably the best measure.

 

[coldpower27]

With system power draw there are more variables to isolate. Do you use an ATI based or Nvidia based board?

With Intel the logical board would be i975X.

 

[Goi]

I don't think it's fair to add in a few watts more just because it doesn't have an IMC. That's like saying you need to add in a few watts more for a P2 because it doesn't have SSE2/3. Or, why not add a few watts to the K8 for only being a 3-issue pipeline instead of 4-issue like the Conroe, or for having only a quarter to half the L2 cache? All these are design choices that you'll have to live with, so here's no point in penalizing either with some arbitary metric.

 

[Bobthelost]

Originally posted by: dexvx
Mmm... you probably need to add 10-15W for the Core2 Duo figures. This is because the K8 has an integrateed memory controller whereas the C2D's do not. Nevertheless, your electricity bill savings will be astronomical!!!

More like 2.5W

http://www.pcreation.fr/?dossier=materiel&page=A64

Look at the size of the memory controller on an A64 chip, it's less than half the length of the die and around 1/7th the height. That means it takes up around 5-7% of the die. Assuming all parts of the chip draw the same power then you're looking at 2.5W or so for most single core chips, around the same for dual core i think.

Of course these numbers are rough to say the least, but i think you're being a bit too paranoid with your power assumptions.

 

[rgreen83]

Originally posted by: Bobthelost
Assuming all parts of the chip draw the same power

That would be a very bad assumption to make.

 

[Stoneburner]

goi that's utterly wrong since the memory controller if not integrated is on the mobo taking in power. yes there needs to be an accounting for the IMC but who knows what the power draw on it is. THat's why it's probably best to look at the overall power consumption levels.

 

[Zebo]

A CPU by itself does'nt have much use other than making cool keychains. Worthless! Total system consumption is the only proper way to measure since thats they only way you can actually get utility from a processor.

Now lets talk about methodolgy. They measure the they current though the 12 V CPU PS line with a current meter and then do the appropiate math for power. Sounds reasonable right? Not really beacause not all motherboards "give" the processor all it's power strictly from the 12 so it's impossible to accuratly isolate CPU power. Depeding on the mobo to varying degrees some power comes from the 24 pin power connector supplying motherboard. I've done this testing myself with a fluke meter, same processor, and depending on the motherboard power varies greatly though 12 V line.

 

[BlingBlingArsch]

anybody tried to download the S&M tool they used for measuring power consumptioN?

 

[zsdersw]

Originally posted by: BlingBlingArsch
S&M tool

Whoa there, pardner. I don't know what kind of forum you think this is, but....

 

[thilanliyan]

Originally posted by: BlingBlingArsch
anybody tried to download the S&M tool they used for measuring power consumptioN?

AFAIK, it doesn't actually measure power consumption.

I think what they meant was that they use S&M to fully load the CPU then take power measurements with their own equipment. I know, the way they worded it makes it seem like they used just S&M to measure power.

I have the S&M program but can't find anywhere to measure watts or even amps.

 

[theteamaqua]

yea thats nice

 

[aka1nas]

Originally posted by: Bobthelost

Originally posted by: dexvx
Mmm... you probably need to add 10-15W for the Core2 Duo figures. This is because the K8 has an integrateed memory controller whereas the C2D's do not. Nevertheless, your electricity bill savings will be astronomical!!!

More like 2.5W

http://www.pcreation.fr/?dossier=materiel&page=A64

Look at the size of the memory controller on an A64 chip, it's less than half the length of the die and around 1/7th the height. That means it takes up around 5-7% of the die. Assuming all parts of the chip draw the same power then you're looking at 2.5W or so for most single core chips, around the same for dual core i think.

Of course these numbers are rough to say the least, but i think you're being a bit too paranoid with your power assumptions.

Why would you apply the power requirements from the A64 IMC to a C2D system? The A64 memory controller is definately going to use less power than one on a current northbridge as the A64 IMC is .09micron logic as opposed to being on the northbridge which is going to be at least a generation or two behind that. State-of-the-art manufacturing processes are not usually used on chipsets.

 

[dmens]

Originally posted by: aka1nas
Why would you apply the power requirements from the A64 IMC to a C2D system? The A64 memory controller is definately going to use less power than one on a current northbridge as the A64 IMC is .09micron logic as opposed to being on the northbridge which is going to be at least a generation or two behind that. State-of-the-art manufacturing processes are not usually used on chipsets.

slower clock, less power.

 

[aka1nas]

Originally posted by: dmens

Originally posted by: aka1nas
Why would you apply the power requirements from the A64 IMC to a C2D system? The A64 memory controller is definately going to use less power than one on a current northbridge as the A64 IMC is .09micron logic as opposed to being on the northbridge which is going to be at least a generation or two behind that. State-of-the-art manufacturing processes are not usually used on chipsets.

slower clock, less power.

That's reasonable.

 

[Markfw]

Well, I won't argue that the C2D uses less power (not including the northbridge as pointed out however), but I have 2 empty 939 sockets, and at only $153 for an X2 3800, with the minimal savings in power, I will stick with them for now. Plus I still don;t see any available anywhere.

The other argument over the 805 was (and is) the heat, which the X2's don;t have. I still have by 805 and 820 pwered down until the temps around here get better. My AC can;t keep the house to 75 with those 2 on. When I turn the off, its fine. You can feel the difference of the heat coming out of those boxes with an X2 3800@2500 right next to it.