Quantum computing: environmental impact

[Prometheus]

Apparently by 2030 21% of electricity demand will come from computers, a significant contribution to energy consumption.

My understanding of quantum computers is that because they run at near absolute zero, there is negligible electrical resistance, making them exceptionally energy efficient. However, to achieve these temperatures must require a significant amount of energy.

Overall would a quantum computing be more energy efficient than traditional computing?

[swansont]

How are you defining efficiency? Power or energy?

If quantum computing is developed and applied to the problems it’s designed for, it’s sort of an apples vs oranges comparison. The quantum computer can solve problems a traditional computer can’t (in a reasonable time) so it must be more efficient. 

IOW, if your traditional computer will take ~3 years to solve a problem a quantum computer can solve in a few hours, the QC can draw 1000x more power and still be more efficient, since it still takes ~1/10 the energy (1/10 of a day vs ~1000 days). 

If the TC simply can’t solve the problem, then an infinite supply of energy can be wasted.

[Prometheus]

19 hours ago, swansont said:

How are you defining efficiency? Power or energy?

Hmm... Which is most useful when talking about environmental impact? I'd have thought power as it takes into account time spent performing operations, which is what you seem to suggest.

 

20 hours ago, swansont said:

If quantum computing is developed and applied to the problems it’s designed for, it’s sort of an apples vs oranges comparison. The quantum computer can solve problems a traditional computer can’t (in a reasonable time) so it must be more efficient. 

I guess then the question is a practical one of whether QCs become common enough to have a significant environmental impact. The applications i've seen involve computational chemistry and similar modelling - not exactly for the masses. But could we have apps accessing cloud quantum computing for some of its processing? Could they, for instance, be used for encryption on a messaging app, requiring banks of QCs?

 

I just came across this paper which was cited on several sites to suggest QCs will reduce power consumption. However, the paper is above me so i'm not sure that's what it actually says.

[swansont]

30 minutes ago, Prometheus said:

Hmm... Which is most useful when talking about environmental impact? I'd have thought power as it takes into account time spent performing operations, which is what you seem to suggest.

No, power does not take that into account. It's a rate. Total energy takes both power and time into account.

Efficiency is usually a ratio of useful outcome compared to the total effort, e.g. useful energy/total energy, but that assumes the same job is being done, and that's not true here.

I might have a laser that's 50% efficient, in that half of the power used is emitted as light, but if the power output is 1 mW and I need 100 mW (and you can't combine systems), then the efficiency number is meaningless.

 

30 minutes ago, Prometheus said:

I guess then the question is a practical one of whether QCs become common enough to have a significant environmental impact.

As of now that looks doubtful in the foreseeable future. The main objection is that they aren't scalable. You can't put together an arbitrarily large number of qbits.