That being said you can always check proton db to see if your games work with it
EAC for Linux works
I think there are a few other anti-cheats which can be problematic, but again I think it can be worked around
But many other AC's like EA's Ricochet, Vanguard or Faceit AC don't support proton at all.
Wouldn't doubt that Ricochet and Vanguard have no support for it at all though
And that's without even considering vr goggles...
Again, do not read what anyone says as gospel, check the games you have and want to play even if people say they're perfect. Lots of things work but a lot of high-profile stuff still doesn't, so it won't work for everybody. And if a game requires 20 minutes of fucking around to get working reliably, it's very possibly not worth it; but IME people who desperately want to use Linux will excuse any amount of issues except for literally not being able to boot.[1] For example, Dragon Ball FighterZ and Destiny 2 don't work well under Proton (for different reasons) but ultimately, I have to use Windows as a result, no matter what anyone says here. Lots of games do work on my Steam Deck though, so that's nice.
[1] You can find tons of reports on ProtonDB like the following: marked "Playable", but then it says "You must launch the game with a custom environment variable to boot it, then it will crash every 10 minutes, which sucks, but it's playable." Literally the definition of "not playable" by any metric...
- what kernel you used + which settings
- what CPU model you used? what security mitigations were enabled?
- GPU model, GPU driver version, GPU driver configuration settings
- what wine version you used (e.g.: wine stable, staging, devel? proton? glorious eggroll? custom?)
- what wine settings you used: overrides, Windows version, installed libraries, etc.
- did it use the gamemode launcher or similar?
Playing with those knobs have a big impact on performance.
When running on wine, we could say that how you are running a game can have as much on performance impact as your specs.
Identifying the differences between setups is hard.
The amount of variables simply has no impact on the difficulty of running a benchmark. The amount of variables will impact the relevance or use value of the benchmark, but interpretation and application is the user's problem. The amount of variables has zero impact on the process of performing or recording the benchmark, and this matters because interpretation and application is easier with more benchmarks covering more cases, but your insistence that it's "hard" could only reduce that number, making you the only variable that is actually a problem.
I found that when comparing a normal Windows 10 installation to SteamOS, SteamOS performed better (slightly more consistent FPS and better battery).
When I switched to the LTS version of Windows 10 (which doesn't have a lot of the bloat in it and you don't run the risk of screwing things up by running a debloat tool) I got slightly better performance on Windows than I did on SteamOS.
This is also not on a higher end PC. I choose to keep my gaming PC windows right now just to not deal with lower performance, I paid a lot for this computer since I value performance on it.
I will say however that I am finding myself using Windows on my Steam Deck more often. Like right now I am replaying through the Kingdom Hearts series. It feels like jumping through hoops to get that to work on SteamOS where on Windows I just installed the Heroic Game Launcher (alternate app for epic games), installed it and it just works.