3 ms·

> can be circumvented entirely if you use the power in space e.g. in AI-training compute centers also in orbit. In addition to power requirements and data tran

> can be circumvented entirely if you use the power in space e.g. in AI-training compute centers also in orbit. In addition to power requirements and data transfer, I always thought that the main logistical issue with in-orbit computers was heat dissipation. If you have a need for compute in orbit (let’s assume this is true), you’d still need a way to dissipate heat. But this isn’t something that can happen easily in space. (But it makes me wonder if you could have a Peltier generator attached to a hot chip to remove heat and “recycle” some of that energy back into electricity… not that it would help too much with the other issue with compute in orbit - data transfer)

schmidtleonard 2y ago
Peltier gives you energy in exchange for making heat transfer worse. Or you can make heat transfer better by paying a bit of energy. The exchange rate is very poor in both directions.
- mbreese 2y ago
  I know they are very inefficient, but are they more inefficient at heat transfer than the alternative of radiant cooling? I have no idea. I know they are used for electricity generation on some other spacecraft, but I always thought it was because it was a method of last resort. But I had never considered if a Peltier would work for cooling.
  - schmidtleonard 2y ago
    Peltiers are never an alternative to radiant cooling. They do not get rid of heat, they pump it from one side to another (if you apply power) or resist its flow (if you extract power -- this is what RTGs do). You still have to get rid of the heat when it reaches the other side of the Peltier.
brianbreslin 2y ago
forgive my ignorance on this q. I always assumed based on movies that the temperatures in space were very cold (like near absolute zero). So wouldn't putting a hot data center in space just dissipate heat? Do shuttles and space stations not radiate heat?
- hydrolox 2y ago
  there is no air or medium for conductive transfer, it's purely radiative. The shuttle used it's very large doors as radiators, as far as I know. The point is that you need a lot more surface area , and that it needs to be unobstructed (since it's literally radiating out, no convention to transfer in cool air/water/medium)

schmidtleonard 2y ago

Peltier gives you energy in exchange for making heat transfer worse. Or you can make heat transfer better by paying a bit of energy. The exchange rate is very poor in both directions.

mbreese 2y ago
I know they are very inefficient, but are they more inefficient at heat transfer than the alternative of radiant cooling? I have no idea. I know they are used for electricity generation on some other spacecraft, but I always thought it was because it was a method of last resort. But I had never considered if a Peltier would work for cooling.
- schmidtleonard 2y ago
  Peltiers are never an alternative to radiant cooling. They do not get rid of heat, they pump it from one side to another (if you apply power) or resist its flow (if you extract power -- this is what RTGs do). You still have to get rid of the heat when it reaches the other side of the Peltier.

brianbreslin 2y ago

forgive my ignorance on this q. I always assumed based on movies that the temperatures in space were very cold (like near absolute zero). So wouldn't putting a hot data center in space just dissipate heat? Do shuttles and space stations not radiate heat?

hydrolox 2y ago
there is no air or medium for conductive transfer, it's purely radiative. The shuttle used it's very large doors as radiators, as far as I know. The point is that you need a lot more surface area , and that it needs to be unobstructed (since it's literally radiating out, no convention to transfer in cool air/water/medium)