> It is estimated that about 70 percent of the energy generated worldwide ends up as waste heat.
This has me a bit confused and I’m curious if someone has more insight. Doesn’t all energy generated end up as heat somewhere? Like every watt going into a bitcoin mining rig gets released as heat. I’m unsure how you’d determine what percentage of that is waste heat.
I thought the question was limited to a "solid state"[0] computer processor.
There is no "useful work" moving large things on a computer processor. In this context, 100% of the energy consumed by a computer processor is heat, right? At least I think that is the working assumption when we build CPU cooling. If a processor consumes 15W of electrically power, we better have some way to move this 15W of heat away from the processor, right?
I have another question, for a light bulb, is this still the case? I guess for a candle or an incandescent light bulb, the question is moot but as we have more and more efficient lights like the "cool" LED lights, if the light source consumes 15W of electricity, do we need 15W of cooling?
I don't think the question was in the context of computers or even electronics in general, as the original claim cited ("70 percent of the world's power production ends up as waste heat") was certainly not restricted to computing.
In the context of CPUs, yes, the heat produced is equal to the input power. However this is not at all true for light bulbs, for motors, for battery chargers, etc. A light bulb in particular transforms some of the electrical energy into heat, but some of it into emitted light.
Even in incandescent bulbs, some the electricity gets turned into heat, raising the temperature of the filament, but past some point, the temperature no longer increases and all the extra energy gets converted into light. That is, if you were to submerge an incandescent light bulb in a large quantity of water and leave it on for 10 minutes, you'd get cooler water than if you submerged just the filament (ignoring the possible shortcircuits): if the temperature of the filament is kept below the point where it emits light, than all the electrocity goes into heat; but if it starts emitting light, then some of the energy escapes and doesn't heat the water.
Not necessarily, it can also turn to motion or to electricity or to chemical processes or to other things. For example, if you're using the light bulbs to light plants, it gets turned to sugar, not to heat.
I googled LED light efficiency what percentage of electricity becomes heat and looks like only about 20% of the input is wasted as heat?
So a 15W LED light would only need 3W of cooling? It feels even more ridiculous when we put the numbers like this... There is no excuse for LED lights to not have adequate cooling or for them to fail because of overheating...
> This means that about 80 percent of the electrical energy is converted to light, while 20 percent is lost and converted into other forms of energy such as heat.
Light bulbs are build to break and be replaced, as far as I understand we could have eternal light bulbs but that’s not in the interest of light bulb makers.
That same logic might still apply for modern light solutions like LEDs. I have a more expensive LED bar which is build on top of aluminium to disperse the heat.
Did you have heard about the Dubai Lamp? It’s basically a very efficient longlife LED which they only sell in UAE.
The Dubai Lamp is produced by Phillips, and is about 3x more efficient than typical LED bulbs -- normally 9w for 60w equivalent vs 3w. Their rated life is 25,000 hours, which is 2.5x longer than a typical 10,000 hour led.
In the case of computers it turns to waste heat and whatever is radiated away is not usable energy and is essentially waste heat. Computers are basically devices for converting useful energy into waste radiation.
Agreed, in computers all of the input power becomes waste heat.
One small exception may be microntollers that use PWM to emit a signal through an antenna, or to drive a motor. I think in that case some of the energy becomes radio waves (which may never be absorbed entirely so they may not become heat) or motion in the motor.
Monitors have an external power source, I have never heard of a monitor being powered by signal coming from the CPU or GPU. That is, the control signal coming from the CPU/GPU doesn't become light, it merely helps decide what light a separate circuit might emit.
Eventually, yes, but I don't think that's what they meant. The universe is far from thermal equilibrium :)
If you push a stone up a hill, then some of the calories you burned went into the gravitational potential energy of the stone, and some up it was lost as heat.
If you compute the SHA hash of some string, then some of of the energy from the power supply went into switching the voltages in transistors, and some of it was lost as heat.
Hmm, these aren't quite the same though, are they?
I was under the impression that ultimately none of that voltage change ends up as anything other than heat. (Despite voltage being "electric potential", that's only really meaningful in e.g., capacitors and batteries, not in circuits that keep switching.)
They differ in the sense that it's easy to imagine how you can reclaim the energy in the stone -- let it roll back down the hill.
It's harder to imagine how you can reclaim the energy of the computation. Overwriting a bit requires a little energy even in principle, and then you can't get it back. Maybe this is what the commenter I was replying to meant by "all energy generated ends up as heat." Even still, I'm tempted to make the distinction between the energy that is needed to perform the computation and the energy that is lost e.g. to Joule heating.
Fundamentally, yes. All motion does end up adding entropy to the environment. This is especially egregious for fossil fuels which are literally combusted/oxidized to move chunks of metal on wheels carrying different kinds of cargo. Electric vehicles change the dynamics somewhat but the motion of magnets still ends up contributing entropy to the environment at a somewhat reduced rate than fossil fuel vehicles.
Heat and entropy are different concepts. Also, motion can reduce entropy locally, even if globally it always increases. For example, when a puddle freezes, the puddle's entropy goes way down. The entire Earth's entropy can also go down, as long as the entropy of the universe increases.
Heat is energy that causes something to raise in temperature, and is measured in Joules. Entropy is a harder to pin down concept, related to the internal state of a system, and is measured in Joules/kelvin. In certain processes, the change in entropy is equal to the change in heat divided by the temperature of the system.
To answer the question as literally as possible, if you reuse heat from a process in order to raise the temperature of something that needs heating, you are reducing waste heat.
Here's an example, unrelated honestly to the article but just your question -- you have a pan of water on the stove, and you are continually pumping room temperature water into it. You are also continually pumping heated water out of it. This is analogous to a lot of chemical reactors but to keep it simple.
If instead you run the hot water being pumped out next to the water being pumped in, the heat can be moved from the hot side to the cold side. That means less energy is needed to heat up the water coming in, because it is preheated.
So, in practice, you do indeed release less heat energy because less is required to keep that kettle hot. The hot water being removed gets cooled down some too, for better and worse (often this is good, who wants hot chemicals right?).
This has nothing to do with the article though, which is talking about using a neural network trained to help speed up predictions of phonon dispersion relations, which is great but maybe a little unclear. A phonon is sort of a quanta of heat if you like, so knowing how they travel in a material certainly is important and objectively cool, but the 70% thing does feel like a bit of a non-sequitur.
To be honest I suspect the first sentence and second are not as related as they are implied to be, the first is simply true and the second is also true but is hardly the reason the overall industrial base of humanity is only 30% efficient. Some stuff just can't be 100% efficient, like a Carnot engine. I'm shocked it hits 30% considering thermal power stations are a major thing [1]. But I also doubt materials limits make a thermal power station less efficient by much, there's just a limit on a thermal engine.
[1] To be clear, there is no citation in the press release to explain the source of this claim. Maybe it is only including after electricity generation, or maybe it only applies to industrial facilities, or maybe they're doing some weighted average of the efficiency of different electronics. I am assuming the 70% includes generation losses.
No, some of our energy generated gets converted into potential energy and stays that way. For instance in cement production when limestone is calcinated to produce calcium oxide, or when iron ores are reduced to pig iron.
Im not sure if all such processes amount to 30% of world energy consumption though.
Energy conversion efficiency[0] describes how much of energy type A is converted into energy type B to perform a useful activity, with the remainder being lost as wasted heat.
For instance, internal combustion engines convert chemical energy into kinetic energy with up to 50% efficiency, with the rest being dissipated and requiring a radiator in the vehicle to remove the excess [wasted] heat. Which is why your car engine becomes very hot after driving the car for a little while.
Imagine your computer has a knob for "release heat" and that setting it to zero on warm days did not reduce the amount of computation done.
Maybe heat is desirable in some contexts, but in most it is an undesirable waste product that takes too much space to eliminate. (Heat pipes, fans, and AC are what percent of data center costs?)
Phonons aren't particles, they are quasi-particles i.e. excitations of the crystal lattice that behave a lot like particles. Weird for an MIT article to get that so wrong, they aren't subatomic either as they are "made" of atoms (which is an over-simplification as well) and can span many unit cells in a crystal.
Popular science articles always get this wrong. I did PhD on magnetic skyrmions and they're widely called 'particles' in articles. You also always see a lot about 'magnetic monopoles' where it's an emergent phenomena in crystal structures.
Yeah, but it’s MIT’s marketing department. Science communication is kind of their thing.
What I’ve observed is that when one person tries to summarize detailed information from another person, there are thresholds where the information passes from overly simplified to wrong. The original author is outraged by even basic simplification, well short of the wrongness threshold; but the summarizer blows past the threshold breezily as they delete words to get the text to fit.
They're a thing that if present will maybe make a detector go click (or beep, or light up, or whatever that detector does to say "yes something is here").
Doesn't that make them a particle? Or is there a better idea of what particles are that I should be using?
The problem with "AI" is that it can mean anything that involves either classical machine learning, or deep learning, and more recently, is primarily associated with LLMs (as in people's initial reaction to this title, without reading the article, might be that LLMs are involved in this new method, which is not the case).
I don’t think this is _the_ problem, or even a problem. So what if we have a class of techniques that share some fundamental commonalities and label them all under the same broad category?
This is where most of the real value of AI actually lies. The scientific and engineering applications of machine learning.
Another MIT study: if we AI-augmented R&D widely in the US, our productive economic growth rate would double. The authors argue this would be permanent, forever increasing the rate of technological progress:
I bet a true accounting of all the economic impacts of semiconductor technologies will show these have far more value than Meta, Instagram, Snapchat, TikTok all combined. The stuff built on top of deep technology has higher PR value -- but the deeper tech has far more economic value. Same with AI.
The reason AI will do great things all levels of reality and different fields of studies (i.e. physics, biology, material science, climate, etc) is that these aspects operate from rules governing behaviour at the specific scale. These context specific rules derive from and summarise fundamental physics directly. This also explains why these methods will succeed in every domain. Analogy: Euler's rule is a topological invariance. So applies to both Spherical and Euclidean geometry.
The second, also called the Euler polyhedra formula, is a topological invariance (see topology) relating the number of faces, vertices, and edges of any polyhedron. It is written F + V = E + 2, where F is the number of faces, V the number of vertices, and E the number of edges.
The difference is that Euler’s rule doesn’t need to be verified after each use of it.
AI may speed up research by uncovering interesting patterns, but I don’t believe it’ll do great things all by itself.
AI is, after all, just a name for a statistical model of something. We’ve been using statistical models for a looong long time.
Any domain already relying heavily on statistical models (like protein folding, I believe) may really benefit from AI (like alpha fold), but domains which don’t won’t.
Protein folding historically was more reliant on physical simulations. Like, David Shaw (of DE Shaw fame) had/has a team optimizing the hell out of simulations. There are oodles of fields (like weather) doing similar, and they are finding they can use statistical models where they previously thought they couldn't.
Imagine a futuristic sci-fi scenario in which AI is so advanced that it can model so deep into physics (i.e. plank scale), a scale which can't be probed by experiments but the predictions match. We would have a probabilistic Oracle of sorts. It can give answers but no way of knowing why it works.
I don't think the why is that confusing. If you do a physical simulation, so many calculations cancel each other out or work in opposite directions. If there is a pattern to it (and it appears there is for many situations), you just short cut it.
Consider Roger Federer's ability to predict where a tennis ball will land and how to send electrical signals to his body to move in a way that will return that ball with high precision. It's pretty wild the number of short cuts he can make for what is a very complex calculation.
On second thought, I missed your point, I think. In my example we could reason through why the tennis ball ended where it did. In yours, we couldn't. My bad.
I agree with your point of calculations cancelling out. That's why I am not a fan of Butterfly effect. Just as Lewis coined unbirthday, we should coin Unbutterfly effect cancelling each other out.
61 comments
[ 3.2 ms ] story [ 107 ms ] thread> It is estimated that about 70 percent of the energy generated worldwide ends up as waste heat.
This has me a bit confused and I’m curious if someone has more insight. Doesn’t all energy generated end up as heat somewhere? Like every watt going into a bitcoin mining rig gets released as heat. I’m unsure how you’d determine what percentage of that is waste heat.
There is no "useful work" moving large things on a computer processor. In this context, 100% of the energy consumed by a computer processor is heat, right? At least I think that is the working assumption when we build CPU cooling. If a processor consumes 15W of electrically power, we better have some way to move this 15W of heat away from the processor, right?
I have another question, for a light bulb, is this still the case? I guess for a candle or an incandescent light bulb, the question is moot but as we have more and more efficient lights like the "cool" LED lights, if the light source consumes 15W of electricity, do we need 15W of cooling?
[0] not a technical term
In the context of CPUs, yes, the heat produced is equal to the input power. However this is not at all true for light bulbs, for motors, for battery chargers, etc. A light bulb in particular transforms some of the electrical energy into heat, but some of it into emitted light.
Even in incandescent bulbs, some the electricity gets turned into heat, raising the temperature of the filament, but past some point, the temperature no longer increases and all the extra energy gets converted into light. That is, if you were to submerge an incandescent light bulb in a large quantity of water and leave it on for 10 minutes, you'd get cooler water than if you submerged just the filament (ignoring the possible shortcircuits): if the temperature of the filament is kept below the point where it emits light, than all the electrocity goes into heat; but if it starts emitting light, then some of the energy escapes and doesn't heat the water.
So a 15W LED light would only need 3W of cooling? It feels even more ridiculous when we put the numbers like this... There is no excuse for LED lights to not have adequate cooling or for them to fail because of overheating...
> This means that about 80 percent of the electrical energy is converted to light, while 20 percent is lost and converted into other forms of energy such as heat.
That same logic might still apply for modern light solutions like LEDs. I have a more expensive LED bar which is build on top of aluminium to disperse the heat.
Did you have heard about the Dubai Lamp? It’s basically a very efficient longlife LED which they only sell in UAE.
They are damn impressive, but not "forever".
https://www.mea.lighting.philips.com/consumer/dubai-lamp
One small exception may be microntollers that use PWM to emit a signal through an antenna, or to drive a motor. I think in that case some of the energy becomes radio waves (which may never be absorbed entirely so they may not become heat) or motion in the motor.
If you push a stone up a hill, then some of the calories you burned went into the gravitational potential energy of the stone, and some up it was lost as heat.
If you compute the SHA hash of some string, then some of of the energy from the power supply went into switching the voltages in transistors, and some of it was lost as heat.
I was under the impression that ultimately none of that voltage change ends up as anything other than heat. (Despite voltage being "electric potential", that's only really meaningful in e.g., capacitors and batteries, not in circuits that keep switching.)
It's harder to imagine how you can reclaim the energy of the computation. Overwriting a bit requires a little energy even in principle, and then you can't get it back. Maybe this is what the commenter I was replying to meant by "all energy generated ends up as heat." Even still, I'm tempted to make the distinction between the energy that is needed to perform the computation and the energy that is lost e.g. to Joule heating.
A lot of it comes down to whether a process is reversible or not. I've taken multiple courses on thermodynamics and still only understand it a little.
Here's an example, unrelated honestly to the article but just your question -- you have a pan of water on the stove, and you are continually pumping room temperature water into it. You are also continually pumping heated water out of it. This is analogous to a lot of chemical reactors but to keep it simple.
If instead you run the hot water being pumped out next to the water being pumped in, the heat can be moved from the hot side to the cold side. That means less energy is needed to heat up the water coming in, because it is preheated.
So, in practice, you do indeed release less heat energy because less is required to keep that kettle hot. The hot water being removed gets cooled down some too, for better and worse (often this is good, who wants hot chemicals right?).
This has nothing to do with the article though, which is talking about using a neural network trained to help speed up predictions of phonon dispersion relations, which is great but maybe a little unclear. A phonon is sort of a quanta of heat if you like, so knowing how they travel in a material certainly is important and objectively cool, but the 70% thing does feel like a bit of a non-sequitur.
To be honest I suspect the first sentence and second are not as related as they are implied to be, the first is simply true and the second is also true but is hardly the reason the overall industrial base of humanity is only 30% efficient. Some stuff just can't be 100% efficient, like a Carnot engine. I'm shocked it hits 30% considering thermal power stations are a major thing [1]. But I also doubt materials limits make a thermal power station less efficient by much, there's just a limit on a thermal engine.
[1] To be clear, there is no citation in the press release to explain the source of this claim. Maybe it is only including after electricity generation, or maybe it only applies to industrial facilities, or maybe they're doing some weighted average of the efficiency of different electronics. I am assuming the 70% includes generation losses.
Im not sure if all such processes amount to 30% of world energy consumption though.
Energy conversion efficiency[0] describes how much of energy type A is converted into energy type B to perform a useful activity, with the remainder being lost as wasted heat.
For instance, internal combustion engines convert chemical energy into kinetic energy with up to 50% efficiency, with the rest being dissipated and requiring a radiator in the vehicle to remove the excess [wasted] heat. Which is why your car engine becomes very hot after driving the car for a little while.
The same principle applies to other systems.
[0] https://en.m.wikipedia.org/wiki/Energy_conversion_efficiency
Maybe heat is desirable in some contexts, but in most it is an undesirable waste product that takes too much space to eliminate. (Heat pipes, fans, and AC are what percent of data center costs?)
What I’ve observed is that when one person tries to summarize detailed information from another person, there are thresholds where the information passes from overly simplified to wrong. The original author is outraged by even basic simplification, well short of the wrongness threshold; but the summarizer blows past the threshold breezily as they delete words to get the text to fit.
Doesn't that make them a particle? Or is there a better idea of what particles are that I should be using?
Another MIT study: if we AI-augmented R&D widely in the US, our productive economic growth rate would double. The authors argue this would be permanent, forever increasing the rate of technological progress:
https://www.sciencedirect.com/science/article/pii/S004873332...
I bet a true accounting of all the economic impacts of semiconductor technologies will show these have far more value than Meta, Instagram, Snapchat, TikTok all combined. The stuff built on top of deep technology has higher PR value -- but the deeper tech has far more economic value. Same with AI.
The reason AI will do great things all levels of reality and different fields of studies (i.e. physics, biology, material science, climate, etc) is that these aspects operate from rules governing behaviour at the specific scale. These context specific rules derive from and summarise fundamental physics directly. This also explains why these methods will succeed in every domain. Analogy: Euler's rule is a topological invariance. So applies to both Spherical and Euclidean geometry.
The second, also called the Euler polyhedra formula, is a topological invariance (see topology) relating the number of faces, vertices, and edges of any polyhedron. It is written F + V = E + 2, where F is the number of faces, V the number of vertices, and E the number of edges.
AI may speed up research by uncovering interesting patterns, but I don’t believe it’ll do great things all by itself.
AI is, after all, just a name for a statistical model of something. We’ve been using statistical models for a looong long time.
Any domain already relying heavily on statistical models (like protein folding, I believe) may really benefit from AI (like alpha fold), but domains which don’t won’t.
It’s no silver bullet.
Consider Roger Federer's ability to predict where a tennis ball will land and how to send electrical signals to his body to move in a way that will return that ball with high precision. It's pretty wild the number of short cuts he can make for what is a very complex calculation.