If they worked to any acceptable level of efficacy then they could be tolerated. They're only tolerated by people who think they work as well as they claim to work (security theater) but anyone who knows about the performance impacts and/or are tech-savvy enough to understand it is a rootkit and potential exploit (that would fully pwn your device) hates them.
Some cheats are getting rather sophisticated now. There's an ever-increasing number of Pi-devices where the cheating is done externally.
They're also chosen by users when the game is filled with cheater. Counterstrike 2 is an example of this with players moving to FaceIT and ESEA (with kernel anti cheat) as the higher ranks of official competitive matchmaking are filled with cheaters.
Proven by who and what proof? Because Denuvo is the only one outspoken about how it doesn't impact performance despite all evidence to the contrary and they provide no evidence of their own beyond claiming it doesnt. Then saying they'll prove it doesn't and then backing out of proving it.
DRM and anti-cheat aren't the same though. That link is talking about denuvo DRM, not denuvo anti-cheat. Also, just because one implementation impacts performance doesn't mean they all have to.
I'll believe it when Irdeto manages to provide any evidence amounting to more than "Just believe us".
Both the anti-tamper and anti-cheat affect performance and it's incredibly noticeable to anyone who isn't building a new bleeding-edge hardware PC every year or two.
Some cheats are getting rather sophisticated now. There's an ever-increasing number of Pi-devices where the cheating is done externally.
https://www.youtube.com/watch?v=QpvwjC1_Luo
https://www.youtube.com/watch?v=revk5r5vqxA