Question gpu benchmarks - longevity?

[iNVERTEDpLAZMA]

I'm looking at the benchmarks results on benchmark.ul and I see a lack of longevity discussion and comparison.
Has anyone tried a gauntlet comparison?
Run it until it burns out and compare times?
I'm a bit less interested in which one hits the highest mark but instead which one I know will keep on trucking when I'm on my 19th hour straight.

 

[tennis2]

You've heard of folding and crypto mining, right?

 

[iNVERTEDpLAZMA]

You've heard of folding and crypto mining, right?

crypto mining, yeah. folding? no. i just googled it. why would they call that folding...? nevermind, that's beside the point.
anyway, yeah I hadn't thought about that. Still, i would like to see the data on which cards last longer than others.

 

[dotas1]

The data you are looking for is either way too huge, or way too small.

They need to include several of the exact same card in order to have a safe conclusion. They also need to test the different revisions of those cards, again, several.

The conditions that those tests were done, play a significant role as well. You can't compare cards being run in a hot, humid, dusty environment with poor airflow with others that were run within a server room at ideal conditions.

Another thing to consider is if the user has repasted the die and replaced the thermal pads either in the beginning or after some time.

There are several other factors like cooling (air, water, number and size of fans), power consumption, faulty components, bad soldering and so many more that can and may as well have changed since the first release of the card (talking about revisions).

It's either nightmerish huge or insignificant small.