CPU Power Efficiency

[Tim]

What percentage of power, required by a modern CPU, is wasted by producing
heat? (a 3GHz P4, for example) TIA

 

[petermcmillan_uk]

What percentage of power, required by a modern CPU, is wasted by producing
heat? (a 3GHz P4, for example) TIA

I believe that it's usually around 50-100W at max load. The thermal
characteristics for a specific processor can usually be found on the
manufacturers site, usually in the data sheet.

I did hear somebody say that Intel specifies their power ratings at 75%
of the actual consumption, but I can't find anything on their site
which states this. If that were the case then that would make Intel
CPU's hotter than Athlons. Has anybody heard of this?

I would expect the P4, 3.0GHz to produce around 85-95W at max load, but
this can be checked on the Intel site. They'll probably produce it
with different cores too.

 

[Matt]

What percentage of power, required by a modern CPU, is wasted by producing
heat? (a 3GHz P4, for example) TIA

The CPU's output is about two-thirds of its input.

 

[Matt]

The CPU's output is about two-thirds of its input.

Yikes. I was thinking of the PSU.

I don't know that your question makes sense.

I guess it would be easy enough to calculate watts/gigaflop, but that
isn't quite what you asked for.

 

[Tim]

The CPU's output is about two-thirds of its input.

So does the percentage of input power, lost to heat, increase as the CPU
speed increases? I assume that a Pentium 133 wastes less than a third of
its input power on heat.

 

[Tim]

Matt wrote:

I don't know that your question makes sense.

In other words, assuming that a 3GHz P4 draws 90 watts from the power
supply, what percentage of that 90 watts is actually used for processing
data and not winding up in the form of heat? The measured heat dissipated
can not be all of the wattage that the CPU draws, there must be more that
actually does the processing.

Theoretically, if a CPU were superconducting it would have 100% power
efficiency, dissipating no heat. I'm guessing that a 3GHz P4 is using as
much power creating heat as it is processing data, but I'm hoping others
might have more information about it.

 

[John Doe]

Yikes. I was thinking of the PSU.
lol

I don't know that your question makes sense.

I agree, I think it's an abstract question in this context of
personal computer hardware CPUs, given the same PC type.

I guess it would be easy enough to calculate watts/gigaflop, but
that isn't quite what you asked for.

That sounds good to me. For simple logic devices, efficiency is the
result of physical type (for example CMOS versus whatever high-speed
logic).

I think there is a known CPU power efficiency difference between
notebook and desktop computers.

The difficult answer to the original poster's question might have
something to do with power input versus all of the output pins. But
who knows. That question would be for a manufacturer or an
electronics group, in my opinion.

Have fun.

 

[Michael Hawes]

It's 100%! The only useful output is the data bits on the data bus to
other chips, which are negligible. A light bulb at least gives a small
quantity of light, but a CPU doesn't even do that.
Mike.

 

[Matt]

So does the percentage of input power, lost to heat, increase as the CPU
speed increases? I assume that a Pentium 133 wastes less than a third of
its input power on heat.

See my reply to my own post. Your question does not seem to make sense.

 

[DaveW]

Put it this way: a Prescott P4 3.0 GHz CPU puts out
~110 Watts of Heat !

 

[Tim]

See my reply to my own post. Your question does not seem to make sense.

Nor does your reply.