zlacker

[parent] [thread] 6 comments
1. pantal+(OP)[view] [source] 2026-02-02 22:18:45
They don't need to be space grade, consumer hardware will do just fine.

For AI a random bit flip doesn't matter much.

replies(3): >>ohyout+a2 >>q3k+x2 >>gbriel+u4
2. ohyout+a2[view] [source] 2026-02-02 22:26:35
>>pantal+(OP)
Random bit flips might even improve output.
replies(1): >>adastr+96
3. q3k+x2[view] [source] 2026-02-02 22:28:01
>>pantal+(OP)
Only if that bitflip happens somewhere in your actual data, vs. some GPU pipeline register that then locks up the entire system until a power cycle. Or causes a wrong address to be fetched. Or causes other nasty silent errors. Or...

Try doing fault injection on a chip some time. You'll see it's significantly easier to cause a crash / reset / hang than to just flip data bits.

'rad-triggered bit flips don't matter with AI' is a lie spoken by people who have obviously never done any digital design in their life.

4. gbriel+u4[view] [source] 2026-02-02 22:34:05
>>pantal+(OP)
As long as they stay below Van Allen belts and deal with weaker magnetic shielding in sun synchronous orbit (high latitudes).

I would say they probably something a little beefier than consumer hardware and just deal with lots of failures and bit flips.

But cooling is a bigger issue probably?

◧◩
5. adastr+96[view] [source] [discussion] 2026-02-02 22:39:29
>>ohyout+a2
Single upset events in a modern GPU are not bitflips. They destroy the surrounding circuitry and usually disable the whole unit.
replies(1): >>pantal+L8
◧◩◪
6. pantal+L8[view] [source] [discussion] 2026-02-02 22:49:46
>>adastr+96
If that happens you disable that CUDA core. If you GPU is too damaged, you deorbit the satellite.
replies(1): >>adastr+qa
◧◩◪◨
7. adastr+qa[view] [source] [discussion] 2026-02-02 22:55:03
>>pantal+L8
Yeas, and this will happen within weeks of launch with the orbits under consideration.
[go to top]