38

Today we had a weird discussion over lunch: What exactly causes power consumption in a computer, particularly in the CPU? (ETA: For obvious reasons I don't need an explanation why a hard drive, display or fans consume power – the effect there is pretty obvious.)

Figures you usually see indicate that only a percentage (albeit a large one) of the power consumption ends up in heat. However, what exactly does happen with the rest? A CPU isn't (anymore) a device that mechanically moves parts, emits light or uses other ways of transforming energy. Conservation of energy dictates that all energy going in has to go out somewhere and for something like a CPU I seriously can't imagine that output being anything but heat.

Us being computer science instead of electrical engineering students certainly didn't help in accurately answering the question.

3
  • Hei, my answer id deleted. It's the only correct answer. I know physics. C'mon. I am an electrical engineer here.
    – user4951
    Jul 25, 2012 at 10:41
  • 1
    "only a percentage (albeit a large one) of the power consumption ends up in heat." is a bit of an understatement. For the computer enclosure (or for the CPU for that matter) it is 100% for all practical purposes. There is a tiny 'rest' amount in the form of radiation emanating from the device. For an LCD the monitor it is only slightly less because it emits light.
    – Jan Doggen
    Feb 7, 2014 at 15:58
  • 2
    processor only question: electronics.stackexchange.com/questions/79166/… Jun 25, 2015 at 5:39

14 Answers 14

26

Electrons are being pushed around, that requires work. And the electrons experience "friction" as they move around, needing more energy.

If you want to push electrons into a PNP junction in order to turn it on, that requires energy. The electrons don't want to move, and they don't want to move closer together; you have to overcome their mutual repulsion.

Take the simplest cpu, a single, lone, transistor:

alt text

Electrons lose energy as they bump around, generating heat. And overcoming the electric fields of attraction and repulsion requires energy.

9
  • 4
    so, to make it short: mainly heat :)
    – akira
    Jun 2, 2010 at 15:41
  • 5
    This still doesn't explain where the rest of the energy goes, ie. what isn't lost to heat. The work you mention eventually turns in to heat and the work/energy you put in to overcome e-e repulsion doesn't disappear; it could be reused, like when you release a spring. Maybe it isn't reused - the spring is released into thin air?
    – trolle3000
    Jun 2, 2010 at 15:43
  • 7
    Energy = heat, light (RF radiation), noise (vibration) and the output signal that eventually becomes stored, transmitted and/or displayed. However, I think you underestimate the amount of heat given off by a PC.
    – Chris Nava
    Jun 2, 2010 at 16:26
  • @Chris Nava: and i think by a huge percentage :)
    – akira
    Jun 2, 2010 at 19:13
  • 1
    The energy isn't regained by you, or the power supply. Yes it's converted back into electrical energy, it's not in a position to be used by you.
    – Ian Boyd
    Jun 6, 2010 at 11:40
17

There's an interesting article on wikipedia about Landauer's principle wich states that (quote):

"any logically irreversible manipulation of information, such as the erasure of a bit or the merging of two computation paths, must be accompanied by a corresponding entropy increase in non-information bearing degrees of freedom of the information processing apparatus or its environment"

This means that (quote):

Specifically, each bit of lost information will lead to the release of an amount kT ln 2 of heat, where k is the Boltzmann constant and T is the absolute temperature of the circuit.

Still quoting:

For, if the number of possible logical states of a computation were to decrease as the computation proceeded forward (logical irreversibility), this would constitute a forbidden decrease of entropy, unless the number of possible physical states corresponding to each logical state were to simultaneously increase by at least a compensating amount, so that the total number of possible physical states was no smaller than originally (total entropy has not decreased).

So, as a consequence of the Second Law of Thermodynamics (and Landauer), some types of computations cannot be done without generating a minimum amount of heat, and this heat is not a consequence of internal CPU resistance.

Cheers!

2
  • I believe that every kind of computation can be transformed to reversible computation: "Although achieving this goal presents a significant challenge for the design, manufacturing, and characterization of ultra-precise new physical mechanisms for computing, there is at present no fundamental reason to think that this goal cannot eventually be accomplished, allowing us to someday build computers that generate much less than 1 bit's worth of physical entropy..." en.wikipedia.org/wiki/Reversible_computing
    – Infragile
    Nov 15, 2013 at 8:25
  • It depends exactly what you consider heat. All movement? Disordered movement only? How do you determined what's ordered vs. disordered movement? …
    – Geremia
    Jan 24, 2019 at 23:52
11

To add to the other excellent answers:

Figures you usually see indicate that only a percentage (albeit a large one) of the power consumption ends up in heat. However, what exactly does happen with the rest?

Actually, almost everything ends up in heat. By the law of Conservation of energy, all the energy (which is power multiplied by time) has to end up somewhere. Almost all processes inside a computer end up turning the energy into heat, directly or indirectly. For example, the fan will turn energy into moving air (=kinetic energy), however the moving air will be stopped by friction with the surrounding air, which will turn its kinetic energy into heat.

The same goes for things like radiation (light from the monitor, EM radiation from all electrical components) and sound (noises, sound from loudspeakers) a computer produces: They too will be absorbed and transformed into heat.

If you read of a "percentage" that ends up in heat, that may have referred to the power supply alone. The power supply should indeed turn a large percentage of its input into electrical power, not into heat (though it does produce some heat as well). This energy will then be turned into heat by the rest of the computer :-).

10

I am a CPU designer. Let me provide a simplest explanation I can think of.

"All electrical energy is converted into heat."

You may ask; if all electrical energy is converted into heat, who provides energy for computation?

"All electrical computation dissipates heat energy."

In a CPU (or any other semiconductor circuit), electrical computation needs two things:

  • A way to send information from one place to other (think wires)
  • A way to act on information (think transistors)

Wires in real world expend heat energy because they have non-zero resistance; transistors also expend heat energy because electrons (and holes) bump into each other and atoms causing heat.

You may now ask: so my electric burner expends all electrical energy as heat but it doesn't compute. Why other way is true (computation expending heat energy).

This is because electrons flow in the burner randomly with no specific path (not useful for computation), but in a CPU electrons flow in a precisely defined path (useful for computation) dictated by HW/circuit design. Either way, electrons move around, causing heat dissipation. In other words, the only difference between a burner and a CPU is that former has no specific electrical path ways for electrons to flow and latter does; just because electron path ways are different, it is not a reason for latter to expend less heat energy.

Let's continue hypothetical questioning. Can we pick something very different from CPUs and see how they contrast? Let's imagine a parked car on road. If I push the car forward, the work done by me (the energy supplied by me) gets converted into two things: a) Car's new momentum and b) Heat due to tire/road friction. Wait a minute, you say, Car's momentum. Something physical I can see which happened solely because I expended energy towards it (minus heat/friction). The heat from friction is lost (just like CPU heat) but momentum generated is still useful (say charging electrical battery in the car during regenerating breaking). CPU's usefulness is in operating on some information (a certain arrangements of bits) and generating a set of new pieces of information (input and output binary bits); information is abstract though; not physical. Car's usefulness is in physical world. Information is for CPU while physical world is for cars. Both radiate heat when they do something useful for us but cars do one more thing: they physically move us around. What does CPU do in physical world beyond generating heat? Nothing. Just another way to see how CPUs convert all electrical energy into heat and nothing else.

Wait a minute, this actually means; I can use CPUs as burners? What if my electrical burner is instead a CPU and I put a cooking pan over it to cook dinner. You bet! You get two things: Food and Information computation with same energy cost! Just very expensive burner though!

5

A lot of it also goes to moving your hard drive and fans, and lighting up your monitor.

Some of it goes to transmitting data over the network. Think about how much power a large radio station needs for this. The computer is doing the same thing with network data, even if it's on a much smaller scale over an ethernet line or wifi antenna.

Moreover, paths within the cpu and motherboard work pretty much the same way as the network transmissions. It takes energy to move electrons down those paths. An electron may not have much mass, but you're moving billions of them, and doing it billions of times per second.

0
4

There is also energy used in in turning memory bits on and off, plus the CPU memory must continue to use power to maintain the current memory even when nothing else is being processed. I was unable to find figures, but you have me interested now so if I do find something I will add it.

3

My understanding is that the vast majority of the energy use by a CPU is output as heat. To do work a physical system converts or moves energy - the CPU does work by converting electrical energy into heat, changing it's internal state a large number of times along the way (so some of the energy is effectively stored for a time that way).

Caveat: my practical electronics and physics training stopped around age 20 over a decade ago, unless you count reading New Scientist, so a passing physicist may be about to tell me I'm completely wrong!

1
  • 1
    Heat energy is almost always a waste product: if we could move electrons around inside a CPU without creating heat, we'd do it in a second. Jun 2, 2010 at 13:59
3

Computation is heat. While of course not all heat is computation. So the only logical answer to; How much is lost to heat? The answer is all of it.

Computation is organized heat. In the form of data. What we consider to be waste heat is just disorganized data and not used for computation.

2

An eariler respondent indicated almost everything ends up in heat. That's almost correct. In fact, all the power input ends up as heat eventually. The fan was a good example. The fan will turn energy into moving air (=kinetic energy), however the moving air will be stopped by friction with the surrounding air, which will turn its kinetic energy into heat. The same concept applies to light from the monitor etc. If you put a computer system drawing 250 watts of power into a sealed room, the net result is the same as putting a 250 watt heater in the room.

2

I wanted to respond to this comment above " Think about a simple electric circuit: a device (any device) attached to a battery. Where does the electricity go? It doesn’t stop at the device; some of it is used to do whatever it is the device does, but the rest continues through the wire, back to the battery (hence the closed circuit)."

This comment is correct if we are talking about electrical current; it flows through the circuit (does work aka dissipates heat) and goes back to the battery (or power source). The current here is actually referring to the flow of electrons.

However, the original poster was referring to heat aka energy dissipated. Heat/energy-dissipitated doesn't go back to the battery. Energy is consumed from the battery and dissipated entirely through heat in the CPU. Electric current is a different matter.

1

Conservation of energy dictates that all energy going in has to go out somewhere and for something like a CPU I seriously can't imagine that output being anything but heat.

So this is basically correct. The building block of a CPU is the transistor and the capacitor (very roughly speaking). See the section on the properties of semiconductors for more information about how the electrical engineering side of things works.

Anything that impedes the flow of electricity creates heat. A wire or a transistor transmitting electricity generates heat because it still has a tiny bit of resistance. A resistor generates heat depending on how much electricity it resists. A superconductor has no resistance and so creates no heat.

A good example of this is a DC to DC converter. For stepping voltage down, you need either a resistor which will turn the extra voltage into heat, or a switching converter which opens and shuts very quickly to maintain a desired voltage. Obviously the second is much more efficient.

For a CPU, more power does not equal more calculations, because the calculations come from switching electricity between different circuits. If you don't believe me, look up "minecraft redstone computer" on YouTube for a simple and interesting example of how computers work. The heat generated in a CPU is simply a byproduct of electricity flowing through wires and has nothing to do with the number of calculations it can do.

Therefore, in theory, an infinitely efficient CPU could exist, which would take an infinitely small amount of power, but it would still require a little bit of power because there must be a certain amount of movement to get things done.

The main power usage in a CPU comes from the millions of transistor state switches and memory cell accesses that happen per second. While designing these more efficiently is definitely possible, a certain amount of electron pressure is required for this to happen.

Furthermore, a DC circuit is always a closed loop, much like a hydraulic system which has a hydraulic pump somewhere in the loop providing pressure, which in the case of a DC circuit is usually either a battery, a generator, or some kind of AC transformer. So in theory if we could find a room-temperature super-conductor, a lot of that pressure loss could be entirely eliminated, resulting in extremely power-efficient CPUs.

Nevertheless, any voltage loss anywhere in an electrical system always results in some kind of energy transfer, usually dissipating as radiation or vibration.

For more on energy efficient CPUs from a computing perspective, EPYC: A Study in Energy Efficient CPU Design as well as Google search "how does AMD improve CPU efficiency" should give you a bunch of interesting resources to explore this more in depth.

Footnote: A regular metal becomes progressively more resistant as it heats up, which in turn heats it up even more, until it reaches a point where the resistance equals the voltage. A room temperature superconductor would have to be based on a material which is not affected by temperature.

And here's a few videos that blew my mind.

Suddenly 100 watt power draw seems like a tiny amount.

0

Yes yes, a CPU converts a lot of the electricity it absorbs into heat. We all know that; that’s why we have such crazy cooling devices attached to the CPU now.

However you’re missing the most basic principle of electronics.

Your debate makes it sound like when electricity enters a light or motor, all of it is converted to light or kinetic energy, which is not the case. Think about a simple electric circuit: a device (any device) attached to a battery. Where does the electricity go? It doesn’t stop at the device; some of it is used to do whatever it is the device does, but the rest continues through the wire, back to the battery (hence the closed circuit).

A computer is no different. Charge carriers come in through the mains, enter the PSU, then to the CPU where they do their work, create heat in the process, then the rest comes out, back to the PSU, and back out to the mains.

Ian Boyd had a good start by pointing at a transistor, but did not follow it up with a tangible explanation of what exactly the electricity is used for (the “payoff” of the device, specifically as an analogy to the movement of a fan or light of an LED). You can do a little research into how a transistor works to really understand it, but suffice it to say that the electricity is used to physically alter the atomic arrangement of part of the transistor to allow or block electron flow. Granted its “action” is not nearly as clear or obvious as movement or light, but the energy is still used to do something (and as Ian mentioned, a bunch of heat is created when you push atoms around). I’ve seen some SEM photos of a CPU gate in action which really helps to visualize things; if I can find one, I’ll add it.

0

While most of these answers are great, and quite sufficient I think to answer the question in practical terms, I think I can provide a simple answer that suffices to meet the spirit of the question. Everyone agrees that the vast majority of the energy is converted either directly, or indirectly into heat. I would suggest that some portion of it is converted into information. This represents a very tiny slice of the total energy input, but there are smarter folks than me on these boards who could definitely calculate the exact amount per bit for you based on Shannon's entropy. The information itself has mass and energy went into it.

1
  • As it’s currently written, your answer is unclear. Please edit to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers in the help center.
    – Community Bot
    Nov 11, 2022 at 0:17
-1

I know this is old, but the answers here aren't the full story.

First, there is a 'static power consumption' component that is a resting state product of input voltage and total leakage current. Leakage current is, well, just what it sounds like. CPUs are these complicated surfaces with all kinds of conducting and nonconducting features and, when you apply voltage - like applying pressure to water in a complex system of pipes - some stuff is just going to leak. It is not an 'accident' necessarily when leaks happen, some of them are just entirely unavoidable even with today's technology, especially now that we are in quantum scales for fab processes like 3nm.

But more interesting is the transient power consumption

Transient power consumption PT can be calculated from PT = Cpd×VCC^2 ×fI×NSW where fI= input signal frequency, NSW = number of bits switching, and Cpd= dynamic power-dissipation capacitance. In the case of single-bit switching, NSW = 1.

The formula simplifies things a lot in quantum-scale land. But, it is the most direct and accepted answer to your question in a 'classical physics/integrated circuits 201' textbook sense, if we're talking about the CPU in particular. And the part that isn't described well here so far is the power being a function of frequency and switching. Really, just switching, because frequency lends itself to faster switching.

I'll try to make an analogy here because I like analogies I'm sorry if it sounds childish but I just like analogies.

Each transistor has a bucket and an agent (you or a friend). Moving the electron liquid within, like others have said, generates heat, period. That is just a magical property of moving our bucket of electrons through the ether no matter how small or large the distance is. You also have a special ('gate') bucket that you just look at. When the gate bucket is full (sort of like a binary 1), you see that and know you must 'do your job' with the electron soup you possess (or, may not possess).

Your job can vary, and it really only depends on what is on your left and your right. You may have an infinite electron source on your left and another transistor agent on your right, who wants you to fill his own bucket. Or, you might take the contents of another agent's bucket on your left and fill or empty someone else's special 'gate' bucket to send them a message to 'do their job' (this and more complicated interactions with the gate is the fundamental thing that makes computation on silicon possible and much more interesting than just resistors and capacitors).

So lets say you play that last role of telling another agent to either do his job or not. You have filled another agent's gate bucket. Now you get a message to tell him to stop doing his job. The only available option is to dump the bucket in a drain on the floor, where it moves through a long series of waste pipes and creates more heat.

So as long as everyone's gate bucket isn't changing, everyone is full or has already been instructed to empty their buckets or hand it to the next guy and there is no moving liquid. None going from agent to agent, coming out of the wall, or being dumped into the sewer. So, no heat. Depending on how all of our transistor agents are arranged though, just one incoming gate bucket changing can result in a new message propagating to tens, thousands or millions of other agents telling them to fill or empty their buckets or pass them on to the next guy. That's a lot of liquid moving and a lot of heat.

This is why your CPU can use very little power (relatively) even when it is at full turbo frequency, if it doesn't have a load. A full load of data is not being calculated, bits are not being flipped, so not a lot of heat is being generated. Except that which comes from leaky buckets or components that don't fit this analogy. looking back at that formula the transient power is zero if either frequency or the number of changing bits is 0. In either case, nothing is changing so no liquid is moving to generate heat. I'll note here that this heat is many many orders of magnitude larger than the theoretical heat limit of information change from information theory.

If you find the analogy worthless just look at this more academic pdf

https://www.ti.com/lit/pdf/scaa035&ved=2ahUKEwjVhaOLqKT8AhWDkmoFHYErDD0QFnoECBoQAQ&usg=AOvVaw20JRuAUK3tt9XyfLSzTwuf

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .