efficiency (not efficacy) of CPUs/computers? is all the power input technically converted into heat?

It has recently been fairly hot where I live, especially in my room, where computer hardware seems to be increasing the temperature by a good 5-10*F over other rooms.

This got me thinking, and was wondering if any computer engineers/physicists may know the answer to this one. If I am running a power-hungry intel CPU, coupled with RAM and maybe SSD storage (I wish!) is all the power fed into my system being converted into heat? In other words, if my rig consumes a good 250W-400W, and there are no transducer devices drawing power (LEDs or lights, speakers/microphones, monitors, motors, heaters, phone chargers, etc,) and all the energy is used for data collecting, and calculating, then is *ALL* the energy converted to 250W-400W of thermal energy (heat)? or do the data operations themselves in fact require energy and maybe entropy or something plays a role? where it would seem that 400W of input yields an apparent 399.981W of energy output + data.

sort by: active | newest | oldest
kelseymh2 years ago

Yes. At a measurable level (are you really using six-significant-figure equipment to measure power?), all of the input energy becomes waste heat, and does so rather quickly.

Steve mentioned energy conservation and computation; that's a cool thing to read about in science magazines, but is irrelevant (femtojoules) to your question.

-max- (author)  kelseymh2 years ago
Lol no, it is just an example. That is me implying that there is microwatts of power "disappearing" or being converted into data manipulation.

This does not have to be a computer as an example, it can be any computation device, also including mechanical systems like gears, cogs, etc.
kelseymh -max-2 years ago

You should look up the relevant articles in scholar.google.com (that will get you actual peer-reviewed science, instead of wacko Internet stuff). Yes, there is a relationship between energy and computation; specifically, the relationship is in _erasing_ data, not in computation per se. However, quantitatively the amount of energy involved is immeasurably small for any practical system (like your PC).

Presumably quantum computation is the limiting case ?

iceng2 years ago

Don't include the display device because it is a power hog !

And then you are sending all those watts out your network to keep the WWW from freezing !

Not to mention the hand warmer pointer thingy.

BTW quantum computation does not all happen in the same dimension as your overheated room...

A way to find out although probably not 3 sig fig accurate is to remove all the fans and submerge your computer in oil and measure the temperature before and after.

That way you wont have to factor in the 2 to 4 watts that the fans will draw.

But my guess is that its not putting out 400 watts of heat, that would be more than 6 60w lightbulbs in a room.

-max- (author)  thematthatter2 years ago
So you are thinking of a large calorimetry experiment? That sounds feasible! Although that would also factor in impedances, galvanic (resistive) losses, etc. It would be nice if I can seperate those out too!

It is theorised that there is a consumption of energy in the process of computation, but I don't think it's been measured yet. It makes sense, if you understand Claude Shannon's communication theorems