Why do PCs produce waste heat?

[Kerr Avon]

Considering how we're all supposed to be so aware of global warming, it's always surprised me how no one ever seems to question the fact that PCs kick out so much excess heat, especially since at the moment (when the weather is *very* warm around here) the PCs are making things almost unbearable in the office...

I mean, the excess heat from a PC (be it desktop or laptop) is waste heat. And waste heat is a sign of bad design (so I've heard), as either the components are drawing too much power, that they then don't need and so is left to radiate outwards, or the PC is producing waste heat as part of it's normal functioning, which is an inefficient side effect. So why aren't PCs produced that counter this, either by drawing only the power they need, or by avoiding the inefficient output of waste heat from their components?

I've heard that it's down to the inefficient design of the original chips, and that every revision/upgrade of the chips are not only carrying all of the inefficiencies of the older chips, but also adding their own new inefficiencies, as a total redesign of the chips, along with as close to 100% efficiency as possible, would be a prohibitively large expense for the chip manufacturers, when instead they can just build on the old chips which costs much less in research and design. I've no idea if that's true or not (I know nothing of CPU design or building, it might as well be magic to me) but I'd be interested to know why such (seemingly) inefficient components are the norm nowadays.

 

[mikeymikec]

Perfect efficiency is not possible, certainly on this planet. Resistant forces are at work in every scenario - in this context, a really simple example is imperfect conductive materials (in absence of almost-perfect conductive materials), being used in the processor result in energy being wasted. That wasted energy is usually in the form of heat, and more energy is required because of the imperfect materials.

A fan is then deployed to dissipate the heat before it affects the components. An imperfect motor assembly wastes energy (sound and heat), the airflow resistance on the blades results in most wasted energy (sound and probably some heat) and more energy is required to power the fan.

CPU efficiency has improved drastically. If it hadn't, to achieve today's desktop processor performance (based on P166MMX-era tech), liquid nitrogen would be required by now to cool the processor (which would probably have to be in a mainframe and require huge amounts of power). As the processor die size increases, as do the power requirements, because of the extra imperfect conductive materials being used that line the pathways/components to provide the capabilities we want.

A Pentium 166MMX used to require ~2.5V - ~3V to power it (as well as active cooling). Due to the improvements made to shrink the design of processors, less power is required (modern equivalent processors require less than a volt and at the very most 1.5V). Nowadays we can produce a processor that's probably 10 times faster which doesn't even require a heatsink, or perhaps a tiny one. If all the average user required in the present day was the performance equivalent of a P166MMX-era machine, it might be possible to pack that all into a purely solar-powered watch-sized device. Unfortunately, our needs have increased as well, though because our needs have increased, the tech has been invented to pursue that goal.

Technology advancements are doing more with less all of the time.

 

[Vectronic]

On a somewhat unrelated note, I've often wondered why no one makes use of the waste heat.

ie: Heatsinks that have a peltier in it that powers the fan that cools the heatsink.

 

[Jeff7]

On a somewhat unrelated note, I've often wondered why no one makes use of the waste heat.

ie: Heatsinks that have a peltier in it that powers the fan that cools the heatsink.

Cost vs benefit. Peltier junctions aren't terribly cheap, and a small fan uses very little electricity. It'll be a long time before that setup pays for itself, unless your electric rates are very high, and you're trying to run a high-speed Delta fan, in which case you're going to want a lot of waste heat.

And cost is not only monetary, but also in complexity.
You've already got 12V available, so use it to run the fans.
A Peltier's now another thing that can go wrong.



The next thing we're up against is, as mikeymikec said, inherent inefficiencies. Until we figure out a way of making superconducting semiconductors (if such a thing is even possible) that can run at room temperature, we're stuck with materials that have electrical resistance. You can reduce it quite a lot, but we're also working with hundreds of millions of individual transistors. In that sense, it's rather impressive that you can get so much done with well under a millionth of a watt per transistor.

 

[Throckmorton]

All the power used by a computer gets converted to heat

 

[aigomorla]

All the power used by a computer gets converted to heat

correction.

All electrictally conductive materials will generate heat as waste unless its a super conductor.

You pass current in anything... unless its non conductive... or super conductive, a by product is always heat as waste.

Problem: We have very little things we consider "super conductor" and all of them require massively insane negative temps.


Also if u could figure out a way to not make heat a by product, your probably also close in figuring out zero point energy as well.

 

[Tsavo]

You pass current in anything... unless its non conductive... or super conductive, a by product is always heat as waste.

Use enough current and sometimes you get also get fire and smoke as byproducts, too.

 

[cheez]

Anytime when there is a movement in molecules that's when the heat is generated. This is mother nature guys. No way around it. You can have less heat in heaven where all things are overruled by God.


cheez

 

[mikeymikec]

On a somewhat unrelated note, I've often wondered why no one makes use of the waste heat.

In extremely cold living conditions I imagine that it might be considered worthwhile to conduct every bit of waste heat (from fridges, computers, televisions, etc) and channel that heat in some useful capacity (the central heating perhaps). Otherwise it would be an awful lot of work for minimal benefit.

Even then it would seem more worthwhile to invest in energy efficiency and then direct the energy savings towards heating systems.

 

[Throckmorton]

correction.

All electrictally conductive materials will generate heat as waste unless its a super conductor.

You pass current in anything... unless its non conductive... or super conductive, a by product is always heat as waste.

Problem: We have very little things we consider "super conductor" and all of them require massively insane negative temps.


Also if u could figure out a way to not make heat a by product, your probably also close in figuring out zero point energy as well.

My point is that if a computer is using 50 watts, all 50 of those watts turn into heat. Nothing else can happen to it.

 

[mikeymikec]

My point is that if a computer is using 50 watts, all 50 of those watts turn into heat. Nothing else can happen to it.

If all of those 50 watts turned into heat, it wouldn't be a computer, it would be a heater.

Another way to look at it is, if all 50 watts turned into heat, there's no energy for computing.

 

[smackababy]

If all of those 50 watts turned into heat, it wouldn't be a computer, it would be a heater.

Another way to look at it is, if all 50 watts turned into heat, there's no energy for computing.

It would be the most efficient heater of all time at that.

The heat is wasted energy, and attempting to harness it will result in more wasted energy to achieve something with minimal gains. Unless someone can invent a extremely efficient way to convert thermal energy into any other form of energy, it won't be worth it on a small scale.

And even on large scale, it is only worth it because it is more expensive to build alternatives.

 

[piasabird]

You generate heat also. 98 degrees. Your refrigerator also generates heat along with your hot water heater You could always turn them off.

Heater, Toaster, Microwave, TV Monitors, VCR's, Cell Phones, Cable Box, Wireless Networking, Stove, oven all make heat.

 

[sm625]

Look at how much heat some cable boxes produce. Hell, look at old incandescent light bulbs. A dozen incandescents can easily produce 500 watts of waste heat. Few houses house PCs that produce that much waste heat. Most new retail PCs (with IGPs) produce less than 20W of waste heat at idle, and less than 50W under load. Notebooks, even less. I think lots of progress is being made.

 

[ElFenix]

A Pentium 166MMX used to require ~2.5V - ~3V to power it (as well as active cooling).

it only required a fan attached to the heatsink if the heatsink itself wasn't very big. if the heatsink were near as large as just the stock intel one nowadays, it likely wouldn't need a fan anywhere nearby.

 

[BrightCandle]

PCs are actually quite efficient electrical heaters. Somewhere in the region of 97% of the input power does indeed get converted to heat. Computers are largely very inefficient at the "work" they do, instead most of that energy is lost into heat and did not go into actually making the computer do anything useful at all. As I understand it the grand majority of this heat is generated in the loss of electrons into the substate silicon.

There are scientists today working on efficient mechanisms for taking heat and converting it directly into electricity but so far they are mostly not very efficient or particularly practical for the cooling of a computer. Neitherless if we could make such a thing we would expect the power consumption of our computers to drop dramatically as much of the wasted energy was recycled.

 

[Jeff7]

If all of those 50 watts turned into heat, it wouldn't be a computer, it would be a heater.

Another way to look at it is, if all 50 watts turned into heat, there's no energy for computing.

It gets turned into heat after the computing occurs.

Like a fluorescent light: Feed in 50 watts to the power supply.
All of it gets converted to heat...eventually.

Some of it gets lost as heat inside the power supply or ballast.
The remainder is used to strike the arc inside the tube.
The arc ionizes the gas within the tube, also heating it in the process.
It produces ultraviolet, which strikes the inside of the tube. Some of that gets converted to heat, but a lot gets absorbed and re-emitted as visible light.
That visible light strikes surfaces in the room. Some gets absorbed immediately, heating the surface it hits, and the remainder gets reflected. Your eye picks up a tiny bit of that light, which also heats it up (slightly). The rest of that reflected light again strikes other surfaces. The only part of that light that doesn't end up heating up your room is that which manages to escape through your windows, through the atmosphere, and into space.

Other than that smattering of photons doing that, the majority of the 50W that went into that power supply has been converted to heat, and your room is now a little bit warmer as a result.



Entropy knows no mercy.

Look at how much heat some cable boxes produce. Hell, look at old incandescent light bulbs. A dozen incandescents can easily produce 500 watts of waste heat. Few houses house PCs that produce that much waste heat. Most new retail PCs (with IGPs) produce less than 20W of waste heat at idle, and less than 50W under load. Notebooks, even less. I think lots of progress is being made.

The reality of an incandescent bulb: It's an electric space heater that also happens to put out a little bit of useful light.

 

[Bryf50]

There's some wonky understanding of thermodynamics in this thread.

 

[Torn Mind]

It's not bad to care about the environment. It is bad to not have an inkling on what is causing its various ills and not understanding scientific terms.

The problem is global warming is heat dissipation of the planet is growing "worse" because of greenhouse gases are doing a fine-ass job of insulating the planet and keeping the heat it receives "in". These gases are produced by various process, of which one of them is from burning coal in the power plants.

So no, it is not the heat generated that is the issue. It is that the electricity used is being produced by fossil fuels that release greenhouse gases when burned.

Also, if you have a damn clue, "work" and "efficiency" in a physics context is NOT what the terms mean in a colloquial sense. In your case, it seems that you are confusing MECHANICAL efficiency here.

Computers do not do MECHANICAL work. What they do is essentially do math by passing electricity current through silicon.

CPUs have their distinct designs in how the silicon is arranged; that is their microarchitecture. There have been quite a few distinct microarchitectures, and they are quite different.

Brush up on physics, buddy.

 

[Red Squirrel]

Components today are way more powerful, thus use up more power.

Since electricity converts 100% to heat, I wonder if there is a way one could use this heat to create electricity, then feed it back into the system. Even if it's say, 25% efficient, wouldn't it do a difference?

 

[aigomorla]

Components today are way more powerful, thus use up more power.

Since electricity converts 100% to heat, I wonder if there is a way one could use this heat to create electricity, then feed it back into the system. Even if it's say, 25% efficient, wouldn't it do a difference?

TEC's can generate a current based on the delta from the hot side and cold side.

So if u had a heat sink on the hot side to suck up heat... and u had the cold side out in the south pole at sub degree temps, you could theoretically get a workable current.

 

[Bryf50]

If all of those 50 watts turned into heat, it wouldn't be a computer, it would be a heater.

Another way to look at it is, if all 50 watts turned into heat, there's no energy for computing.

It would be the most efficient heater of all time at that.

Basically every electrical device in your house is 100% efficient at turning electricity into heat. But the efficiency that we care about is how much work is done compared to the amount of energy inputted. Performance per Watt is the efficiency of a CPU.

Even the energy used for computing will eventually be turned into heat.

 

[CZroe]

Resistance and leakage, both of which are serious limitations at the scale transistors are made in CPUs and other chips. These push the limits of technology to operate at the frequencies they operate so they operate on the edge of what is tolerable. With each new advancement in manufacturing technique (typically a process shrink) we gain the ability to operate more efficiently at past/existing speeds OR the ability to operate faster if continuing to push for the edge of what is tolerable. The market forces (demand) decides which they want. People typically want faster CPUs.

 

[Torn Mind]

Basically every electrical device in your house is 100% efficient at turning electricity into heat. But the efficiency that we care about is how much work is done compared to the amount of energy inputted. Performance per Watt is the efficiency of a CPU.

Even the energy used for computing will eventually be turned into heat.

I think what gets people stumped is that the envisioning the "work"(in the colloquial sense of the term) a computer is not as intuitive as mechanical work. The latter has specific definition in mechanical physics (Force*distance) and is easy to imagine. Whereas, electricity passing through a circuit board and silicon chip(s) takes a bit more mental effort before someone "gets" it. Electricity is simply passing through a semiconductor a computer and practically none of that is converted into another form of energy. The "work" done is due to the circuits within the semiconductor are such that it they can perform calculations while the electricity is passing through.


A computer does virtually no mechanical work, so all of that electrical energy it used needs to be conserved somehow, and that's by converting it into heat eventually.

 

[Kerr Avon]

Thanks for all of the posts, it's very interesting to follow.