Honestly, there isn’t really a “max” that fits every card. It depends a lot on the GPU model, cooling, and your power setup. Most miners just push clocks slowly until it starts crashing or temps get too high, then back it off a bit. As long as you keep temps under control and don’t go crazy with voltage, it’s usually safe. The real danger is heat and running fans 24/7, that’s what kills parts over time.
Ahh that depends..
A high energy asic.
The s21xp burns about 3500 watts it is energy dense in terms of its actual size.
Even with great cooling it can over heat really fast if cooling has an issue.
The 4 fans pull air hard and they are 140mm 6000 rpm plus fans.
If they suck a plastic wrapper or a piece of paper to the rear fan the air flow drops 30 to 60 percent in an instant.
But all the meters for the until read fans are working and the until will slowly heat up.
Then the unit may not shut down fast enough to cool off and you toasted the gear.
Been there done that.
If you were clocked with after market at 90% and set shut down at 78c . The gear will likely not die.
If you are clocked at 100% and 85c or stock maybe it gets damaged maybe not.
But if you clock to 110% and set temp to 88c. Huge chance you toast the gear.
This is also true of really high power gpus 3090s 3090tis etc.
But lite duty gear say antminer L3+ an 800 watt miner it is way harder to burn it out from over clock. Or the 3060ti gpu