Confused about power consumption of C2Ds! Silence is Gold

[go1ge]

I have this *disturbing* urge to make an OC core2 system, thanks to TG and forum members. I even don't know whether I need it, I just have the urge. I just want to run some scientific simulators (FEM software and MATLAB), rip,edit DVDs etc in a shorter time than XP 3200+ at home.

However;
-I do not want to spend too much on mobo+CPU (~$300-350), although I can go higher for a very attractive system, I just want to keep it low.
-I do not want a very hot/loud system either
-I want to be able to keep my Antec Sonata with 380W
-I'm not into gaming, therefore I bought a fanless 7600GS, instead of a GT
-I do not want higher electricity bills, I admire treehuggers as well.
-I want my box live at least 2-3 years, with low maintenance.

I was trying to figure out impact of overclocking on power consumption, however could not find a reliable source.

I found this x-bit labs article Contemporary Dual-Core Desktop Processors Shootout (page 11)
which claims e6300/6400 dissipate more power than e6600/6700, x6800 dissipates least, when all at idle 8O . Don't they all have the same core, which would mean power dissipation would increase by freq. Extra cache on higher end models should also add some extra dissipation.

This could be explained maybe by idle frequencies being lower for high end models wrt low end ones. But I would not know why that could be. clock multipliers somehow require higher idle freq for 6300/6400?

I was kind of hoping that if I OC 6300 to 3GHz, it would dissipate similar to x6800, excluding peripherals (memory, chipsets etc.). Am I being too naive here?

Can anybody explain these to me? or point to a source? As an example, extreme PSU seems not to differentiate between C2Ds.

Are there rules of thumbs about overclocking power consumption estimates? e.g will keeping voltages same and overclocking 20% increase power consumption by 44% (freq. square dependence).

Resistance of metals (interconnects etc) are lower at lower temps, however semiconductor resistance increases by reduced temp.
Does anybody know impact of temp on power dissipation of CPU, or any source for such info?

I probably do not need to know these for OCing, however, urge to understand is as strong as urge to OC. I'll probably postpone OCing untill I have some better understanding about power, sigh.

Oh, I almost forgot, where does 4300 stand in terms of power consumption, stock a nd OCed?

To summarize:
1- How to figure out power consumption of C2D's, be it idle or 100% utilized (stock and OC'd, say 50%)

2- Do I really need a better cooling to reduce power consumption?

3-Will 4300 be better than 6300 in terms of power at the same OC freq?

4-Should I go for 6600 if thermal budget/noise is a concern?

 

[Mondoman]

...
-I do not want to spend too much on mobo+CPU (~$300-350), although I can go higher for a very attractive system, I just want to keep it low.

Sounds reasonable. An E6300 + decent MB.

-I do not want a very hot/loud system either

Not sure exactly what you mean by this. In general, bigger diameter fans move air with less noise.

-I want to be able to keep my Antec Sonata with 380W

Not a reasonable requirement, given that your 7600GS doesn't sip power, and you want to overclock the CPU. You need a PS with good +12V outputs; Fry's last week had a special on the Antec Neo HE 550w for around $50-60 AR, IIRC. Something like that would both increase efficiency over your current entry-level PS, and provide enough +12V output to give you good performance and reliability.

-I do not want higher electricity bills, I admire treehuggers as well.

Power is cheap. Running at 10w more for 100 hours (roughly 2 weeks at 8 hours/day) will only cost you around $0.08, depending on your cost of electricity.

I was trying to figure out impact of overclocking on power consumption, however could not find a reliable source.

The problem is that many different components are contributing to increased power consumption. For example, let's say you OC by boosting FSB speed. Not only is the CPU running faster, but so are MB components dealing with the FSB (such as the Northbridge), probably RAM and memory bus components, etc. Many of those components have voltages which are set independently.