Desktop CPU Power Survey, April 2006

Want to talk about one of the articles in SPCR? Here's the forum for you.
Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Post by Devonavar » Tue Dec 26, 2006 10:01 am

Yes, there is. It has been in the works for a long time though ... we have no release date, you'll just have to read it when it comes out.

flokel
Posts: 28
Joined: Fri Nov 04, 2005 1:50 am
Location: Germany

Post by flokel » Tue Dec 26, 2006 12:11 pm

Ah, very good news :D

ddekany
Posts: 21
Joined: Wed Dec 21, 2005 12:40 am

Post by ddekany » Mon Jan 08, 2007 2:44 pm

flokel wrote:Would be interesting to see Core (2) Duo and AM2 (EE/SFF/65nm) competing against each other.
Also against the slowest Cedar Mills (i.e. ~3.2GHz 65nm Celerons). For most non-gamers the real-world performance difference between a near 3GHz Celeron and a E6300 is most certainly totally negligible, but the quietness and longer lifetime (because of cooler operation) and maybe the lower electricity bill does mater. And the price, of course.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Jan 08, 2007 3:19 pm

the quietness and longer lifetime (because of cooler operation) and maybe the lower electricity bill does mater.
3 of the 5 Cedar Mill Celerons actually have a higher TDP than the C2D (including the 3.2GHz), so there are no electricity savings. Basically the only reason to get a Celeron D nowadays is price; here in the UK it costs almost 3 times less than the cheapest C2D.

ddekany
Posts: 21
Joined: Wed Dec 21, 2005 12:40 am

Post by ddekany » Tue Jan 09, 2007 6:45 pm

jaganath wrote:3 of the 5 Cedar Mill Celerons actually have a higher TDP than the C2D (including the 3.2GHz)
My guess is also that C2D eats considerably less. However I didn't find any concrete tests where you could compare the power consumptions in real-world situations, so I just can't know if how much less. As of the TDP-s, well, there are marketing people there, whose intent is seldom to give you meaningful data.

ddekany
Posts: 21
Joined: Wed Dec 21, 2005 12:40 am

Post by ddekany » Fri Jan 12, 2007 3:09 pm

You know what, despite that NetBurst is such ineffective, I'm not sure if the C2D would consume less in most cases than a 3.06G Cedar Mill Celeron. After all, the Celeron is single core, and has 1/4 of cache, which together means much less transistors, and hence, I guess, much less total leakage current, and that is certainly an important factor for 65 nm CPU-s. Wouldn't that dramatically decrease the idle power consumption? Not to mention that looking at the system as a whole, the a Celeron uses 533Mhz FSB as opposed to 1066 and that might further reduce the heating of some mobo components. (I know, one should use AMD Sempron if he wants the least consumption and the lowest price... but that was not the question now. :) )

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Fri Jan 12, 2007 5:08 pm

However I didn't find any concrete tests where you could compare the power consumptions in real-world situations
Well this article compares an E6400 with a P4 631, the E6400 uses less. I realise the P4 has much more L2 cache, but this is just as a comparison. Also later on in the article it says:
A new power management method exists in the Core 2 Duo that allows the processor to accurately manage consumption even in load. This is called, Ultra Fine Grained Power Control. It consists of a very precise cutting out of areas that can be placed in sleep. Non solicited units remain in sleep even if the others run at full speed. This often happens, because it´s rare that all processor units are solicited at the same time. This ultra precise management makes it possible for better control of power consumption and thermal dissipation.
I don't think the Celeron has this.

Also Xbitlabs measured the power consumption of the E6300:

http://www.xbitlabs.com/articles/cpu/di ... ent_8.html
However, for the dedicated AMD fans I would like to mention a few facts that may change the attitude to the last chart. The thing is that S&M, just like many other tools creating ultimate processor workloads, use special floating-point operations. And it was fine for CPUs on K8 or NetBurst microarchitecture. However, Core 2 Duo processors do not get loaded to the full extent with these utilities. We had to go through a number of different burn-programs to realize that there are a few old ones that do the job much better than S&M, prime95 and others.

For example, when we resorted to BurnK6 tool that we used to use to heat up AMD K6 processors back in the days, we managed to get much higher power consumption numbers for Core 2 Duo E6300: 55W for the CPU and 229W for the platform. In other words, Intel Core based processors hit the maximum power consumption and heat dissipation in absolutely different type of tasks than their competitors and predecessors.
So it depends what you will use the CPU to do.

ddekany
Posts: 21
Joined: Wed Dec 21, 2005 12:40 am

Post by ddekany » Sat Jan 13, 2007 6:32 am

jaganath wrote:Well this article compares an E6400 with a P4 631, the E6400 uses less.
Yes, that's quite impressive, but it compares the CPU-s under high utilization. For most non-gamer rigs the idle consumption is much more important. And also, idle is where leaking has the most importance.

I understand that Core 2 has this and that tricks that seems to decrease idle consumption, while the Celeron doesn't even have basic things like SpeedStep. But I was not in the team who designed C2D, so I just can't estimate how much those features do mater in reality, or what inherent problems the architecture has in idle. So unfortunately the only thing I can believe to is measurement. That's what matters, that's the end result. And again, here I'm talking about the idle consumption.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Sat Jan 13, 2007 6:46 am

the Celeron doesn't even have basic things like SpeedStep.
Well it should be obvious that two processors, both with practically the same TDP (which is not made up by the marketing department, like you said, but by the engineering team) the one with Speedstep will use less, because the one without Speedstep will be operating at full voltage all the time, and heat is proportional to the square of the voltage. I don't think leakage will be very different between the two chips, they are made on the same process (65nm).

What application will this be used for? File serving?

It seems to me you have already made up your mind to pick the Celeron before you even asked the question, I hope I am wrong.

ddekany
Posts: 21
Joined: Wed Dec 21, 2005 12:40 am

Post by ddekany » Sat Jan 13, 2007 8:11 am

jaganath wrote:Well it should be obvious that two processors, both with practically the same TDP (which is not made up by the marketing department, like you said, but by the engineering team)
Not by the marketing department... my experience regarding hardware specs of other stuff (not on CPU-s TDP-s, as I can't measure that) doesn't support this. The usual method is that if the parameter is something "performance" related then manufacturers do the measurements with some tricks (with the clear intent of deceiving you), or they simply just lie. Respect to the exceptions.

Anyway, the interesting thing is measurements of idle consumptions for both CPU-s. We could discuss with whatever theoretical reasonings if which consumes the least, but in fact none of us knows that much about these processors to tell... And no, I didn't decided, I just wanted to know. In fact my current bet is that C2D would win in idle too, but I'm not very sure.

Post Reply