4 ms·

https://en.wikipedia.org/wiki/External_Active_Thermal_Control_System https://en.wikipedia.org/wiki/External_Active_Thermal_Contro... All that gets you 70kW of

https://en.wikipedia.org/wiki/External_Active_Thermal_Control_System https://en.wikipedia.org/wiki/External_Active_Thermal_Contro... All that gets you 70kW of cooling. Radiating to vacuum isn't very efficient.

scottyah 6mo ago
Dang, sucks we can never improve any technology. Let's just call it quits, guys.
- ceejayoz 6mo ago
  Technology is wonderful. Physics still gets a say.
- cyclopeanutopia 6mo ago
  Maybe one day we can, but it's definitely not in a category "there is no doubt".
  - scottyah 6mo ago
    Of course not, where's the fun in that category?
lijok 6mo ago
And that’s sufficient for roughly 100 unoptimized DC grade H200s. Not efficient, and it doesn’t have to be, because the cooling system has 0 opex cost. And capex clearly can be made to wor
- ceejayoz 6mo ago
  OK, so you've got a football-sized solar and radiator array to support 100 H200s. Why are we not building it on land again in some abandoned mall's parking lot?
  - lijok 6mo ago
    Because it’s allegedly more expensive
dzhiurgis 6mo ago
Cooling space station is very different from cooling off chips. One requires extensive piping, other - a simple radiator.
- ceejayoz 6mo ago
  Both require the same thing - moving heat - and you’ll find plenty of piping in a datacenter for this reason.