4 ms·

I find myself dubious of the claim that after an hour of isolated number crunching, my GPU has done anything aside from generate heat (that wikipedia page doesn

I find myself dubious of the claim that after an hour of isolated number crunching, my GPU has done anything aside from generate heat (that wikipedia page doesn't appear to mention anything about computing information). If only 95% of it became waste heat, where is the 5%? Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste".

NeutronStar 5y ago
If you run in a circle, have you produced no work?
zamadatix 5y ago
> that wikipedia page doesn't appear to mention anything about computing information From the opening of the Wikipedia page: 'Landauer's principle is a physical principle pertaining to the lower theoretical limit of energy consumption of computation. It holds that "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".' Computation is in the first sentence, it's mentioned by name 8 times in the opening section - information 6 times. I'm not really sure how you came to the conclusion it doesn't mention anything about computing information. Even skimming it's hard to find a spot not mentioning it directly, by name. > Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste". What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem. Also note there is a difference between "energy that eventually ends up as heat" and "waste heat". All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. The latter is known as waste heat. Waste heat is not just "heat I ended up with at the end".
- patrick451 5y ago
  > Waste heat in a computer is largely resistance, power conversion inefficiencies, and cooling. Heat generated by the actual computation is known as work https://en.wikipedia.org/wiki/Work_(thermodynamics) https://en.wikipedia.org/wiki/Work_(thermodynamics) According to ctrl-f, the word computation never appears on this page. The word information never appears in this context. Same with https://en.wikipedia.org/wiki/Waste_heat https://en.wikipedia.org/wiki/Waste_heat > What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem. None if this explains why the portion of heat in a clock cycle is different if we consider the clock cycle to be doing something clever. If some of the energy is not waste, it must have taken another form. What exactly is it? When we switch a transistor non-cleverly, presumably the energy never takes this form, and thus the actual switching of the transistor took more energy . Which makes no sense. > All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. I'm not convinced this is true. It seems equivalent to the claim that all macroscopic objects moving through the universe will eventually stop, which I would ague is an assertion beyond science. > Waste heat is not just "heat I ended up with at the end". Fine, but we have to be able to point to where the other portion of energy ended up in the interim. Suppose I push black and white boulders up a hill. This leaves them with potential energy, which is somewhat less than the energy it took to get them there. We can agree that at some point in the distant future, the hill will decay, and the boulders end up back on the ground. Thus all the work eventually ended up as heat. I can choose to arrange them randomly, or I can arrange them to create binary numbers. It seems as though you are claiming that if I create binary numbers with these boulders, the energy it took to do that is more than if I arrange them randomly. I want to understand how exactly it took more joules to do that, and where this extra energy is stored.

NeutronStar 5y ago

If you run in a circle, have you produced no work?

zamadatix 5y ago

> that wikipedia page doesn't appear to mention anything about computing information From the opening of the Wikipedia page: 'Landauer's principle is a physical principle pertaining to the lower theoretical limit of energy consumption of computation. It holds that "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".' Computation is in the first sentence, it's mentioned by name 8 times in the opening section - information 6 times. I'm not really sure how you came to the conclusion it doesn't mention anything about computing information. Even skimming it's hard to find a spot not mentioning it directly, by name. > Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste". What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem. Also note there is a difference between "energy that eventually ends up as heat" and "waste heat". All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. The latter is known as waste heat. Waste heat is not just "heat I ended up with at the end".

patrick451 5y ago
> Waste heat in a computer is largely resistance, power conversion inefficiencies, and cooling. Heat generated by the actual computation is known as work https://en.wikipedia.org/wiki/Work_(thermodynamics) https://en.wikipedia.org/wiki/Work_(thermodynamics) According to ctrl-f, the word computation never appears on this page. The word information never appears in this context. Same with https://en.wikipedia.org/wiki/Waste_heat https://en.wikipedia.org/wiki/Waste_heat > What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem. None if this explains why the portion of heat in a clock cycle is different if we consider the clock cycle to be doing something clever. If some of the energy is not waste, it must have taken another form. What exactly is it? When we switch a transistor non-cleverly, presumably the energy never takes this form, and thus the actual switching of the transistor took more energy . Which makes no sense. > All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. I'm not convinced this is true. It seems equivalent to the claim that all macroscopic objects moving through the universe will eventually stop, which I would ague is an assertion beyond science. > Waste heat is not just "heat I ended up with at the end". Fine, but we have to be able to point to where the other portion of energy ended up in the interim. Suppose I push black and white boulders up a hill. This leaves them with potential energy, which is somewhat less than the energy it took to get them there. We can agree that at some point in the distant future, the hill will decay, and the boulders end up back on the ground. Thus all the work eventually ended up as heat. I can choose to arrange them randomly, or I can arrange them to create binary numbers. It seems as though you are claiming that if I create binary numbers with these boulders, the energy it took to do that is more than if I arrange them randomly. I want to understand how exactly it took more joules to do that, and where this extra energy is stored.