Power Consumption of CPUs

[bosnian81]

Hi Guys,

I am wondering how much $ is it a year to run a 65W processor compared to a 125W processor.

Reason I am asking is because I was thinking of investing in E7200($130) and OCing to 3.2GHz to equal E8400 ($190) that I would no OC if I bought it. I buy computer every 2 years so if a 3.2GHz E7200 takes 100W to run, how much more $ a year is that on my bill?

And please don't estimate if you don't have a solid idea/proof.

Thanks in advance.

 

[Just_An_Engineer]

Hi Guys,

I am wondering how much $ is it a year to run a 65W processor compared to a 125W processor.

Reason I am asking is because I was thinking of investing in E7200($130) and OCing to 3.2GHz to equal E8400 ($190) that I would no OC if I bought it. I buy computer every 2 years so if a 3.2GHz E7200 takes 100W to run, how much more $ a year is that on my bill?

And please don't estimate if you don't have a solid idea/proof.

Thanks in advance.

One thing to keep in mind is that the power rating of the processor is not the same as the actual power consumption. The actual power consumption will vary with the load that is placed on the processor so you will need to factor in how the CPU will be used in order to factor in the total energy usage. If you want a rough estimate of the energy usage for the processor I would suggest finding benchmarks for the power consumed at idle and at load and then calculate a weighted average depending on what fraction of the time you computer is operating will it be running under full load. You can then multiply this average power consumption by the number of hours per year you expect to operate you computer each year, convert to kWh (divide by 1000) and then multiply by the cost per kWh found on your electric bill.

 

[einstein4pres]

JaE is right. However, to help you with the calculations:
125W-65W=60W=0.06 kW.
Assuming 24/7 operation, and 12c/kWhr

$63 = $0.12 (c/kWhr) * 0.060 (kW) * 24 (hours/day) * 365 (days/year)

Of course, if you overclock a processor, you will increase the power consumption, and the power rating is not necessarily accurate either. I'm pretty sure that it's a maximum.

12c/kWhr is what I pay for solar. Regular power would cost me 10c. Add +1c for the summer.

 

[joefriday]

Also, an e7200 at 3.2GHz probably won't consumed anywhere near 100 watts. If 65 watts was the actual power consumption at stock speeds, an overclock to 3.2GHz without an increase in vcore would likely only increase maximum power consumption by less than 10 watts.

 

[einstein4pres]

Power consumption is linear with frequency, and varies with the square of voltage.

P(F,V)=c*F*V^2, where c is a constant.

 

[gamecrazychris]

I've always had a similar question, I have a q9450 and my friend has a q6600 and both of the specs say they're 95wat cpu's but the review of the 45nm processors always say that they're much more energy efficient than the 65nm ones but how can that be if they both say they are 95 wats.

 

[4745454b]

I've always had a similar question, I have a q9450 and my friend has a q6600 and both of the specs say they're 95wat cpu's but the review of the 45nm processors always say that they're much more energy efficient than the 65nm ones but how can that be if they both say they are 95 wats.

Thats because as far as I'm aware of, thats the heat output of the CPU, not the power consumption. Meaning these CPUs put out 95W of heat energy, not consume 95W of power. Do the math. 95W / 1.3V = 73A? Few if any PSUs outputs that kind of power. Also keep in mind that AMD releases lower wattage parts by binning their CPUs. They find the CPUs that are stable with 1.3V instead of 1.4-1.45 and sell them as energy efficient CPUs. They are all made at the same time, just different bins.

I'm not sure I'd worry about this. If you are going this far, you also need to make sure you get a good motherboard, and perhaps one of those new Silver 85 Enermax PSUs.

 

[pcgamer12]

If you want to save money on electricity don't OC and buy a 80 Plus PSU.

 

[awolfe63]

Power consumption is linear with frequency, and varies with the square of voltage.

P(F,V)=c*F*V^2, where c is a constant.

Except when it doesn't

This formula is true in the absence of non-linear power management and is a good worst case estimate - but not accurate for PC CPUs.

2 main reasons:

1) When you overclock a CPU at less than 100% constant load - it finishes tasks faster and drops into low-power states faster (even is speedstep/Cool n quiet is off).

2) A substantial amount of the power on a CPU is I/O power. That is running on a different voltage that the one you set int he BIOS.

 

[Wisecracker]

Thats because as far as I'm aware of, thats the heat output of the CPU, not the power consumption. Meaning these CPUs put out 95W of heat energy, not consume 95W of power. Do the math. 95W / 1.3V = 73A? Few if any PSUs outputs that kind of power....

The MOSFETs are converting a 12-volt line to your cpu voltage (at let's say 80% efficiency). I'm lazy and just divide by 10 so I think it would be something like 7.3A instead of 73A.

 

[einstein4pres]

Meaning these CPUs put out 95W of heat energy, not consume 95W of power.

In a steady state system, if you are producing 95W of heat, then you must be consuming at least 95W of power. I don't think CPUs don't do work, so the steady state assumption should hold up.

1) When you overclock a CPU at less than 100% constant load - it finishes tasks faster and drops into low-power states faster (even is speedstep/Cool n quiet is off).

2) A substantial amount of the power on a CPU is I/O power. That is running on a different voltage that the one you set int he BIOS.

1) Which means the voltage and/or frequency is changing.
2) Good point. There is probably an additional term to the equation that I'm missing.

 

[bf2gameplaya]

Hi Guys,

I am wondering how much $ is it a year to run a 65W processor compared to a 125W processor.

Reason I am asking is because I was thinking of investing in E7200($130) and OCing to 3.2GHz to equal E8400 ($190) that I would no OC if I bought it. I buy computer every 2 years so if a 3.2GHz E7200 takes 100W to run, how much more $ a year is that on my bill?

And please don't estimate if you don't have a solid idea/proof.

Thanks in advance.

The PC Power Calculator:

http://extreme.outervision.com/psucalculator.jsp

I also suggest the amazing Kill-A-Watt meter:

http://www.p3international.com/products/special/P4400/P4400-CE.html

 

[joefriday]

Power consumption is linear with frequency, and varies with the square of voltage.

P(F,V)=c*F*V^2, where c is a constant.

I'm familiar with that formula, but observed power consumption does not correlate with the projected outcomes with that formula. In other words, that formula is a piece of crap, and in no way corresponds with real life power consumption of overclocked CPUs as evidenced by the actual measured consumption by such devices as DC clamp ammeters and kill-a-watt style watt meters. The only thing I can agree with on that formula is that an increase in vcore does make for an impressive increase in power consumption. However, a 50% overclock to on CPU using stock vcore does not equate to a 50% increase in power consumption, not even close.

 

[jsc]

The MOSFETs are converting a 12-volt line to your cpu voltage (at let's say 80% efficiency). I'm lazy and just divide by 10 so I think it would be something like 7.3A instead of 73A.

I would estimate the efficiency to be higher - like around 90%. Switching regulators with high quality inductors are surprisingly efficient.

My Q6600 at stock speed pulls about 8 amps through the EPS connector. OC'd to 3.6 GHz, it pulls 9.5 amps.