Hacker News Reader

HNHacker News

Top New Best Ask Show Jobs

Comment by trhway | Hacker News Reader

Parent Full thread

trhway·

It is 700w of heat radiating by 1m2 at 70C ( ie one GPU can sit on such a radiator) What is not to work ok here?

View on HN

No air. No convection, only radiation.

That is what I said.

What you said was just a statement, not something with evidence or reasoning for why it should work. How large of an area is required to dissipate that much heat in a vacuum at a constant rate?

1m² of perfect black body in vacuum at 343K should emit about 700W, according to the Stephan-Bolzmann formula. Should do for a consumer GPU, or an H100.

Since the heat distribution over the radiator is not ideally uniform, I suppose that the emitted power will be less, if they want to keep the chips at 70C.

Since the temperature has the fourth power in the formula, having the GPU work at 100C would increase the radiator efficiency quite a bit.

>How large of an area is required to dissipate that much heat in a vacuum at a constant rate?

that is exactly what i said.

>something with evidence

middle-school physics, basic knowledge, at least if you decided to participate in such a discussion.

I read your statement as just stating what they had done (or planned to do), not that you had calculated that as the correct/sufficient amount. If you read it that way it sounds like 'they provided X amount, why shouldn't that work?' which reads like a statement without evidence of why it would work.

Sorry for misinterpreting your statement.

Reply on news.ycombinator.com