esync = alpha version

fsync = beta version

ntsync = final release

Ntsync got rid of performance degradation that can occur with some games under esync and fsync and that’s the why it's allowed to go in the mainline kernel, it has no downside.

fsync isn't faster than ntsync, it's merely a workaround to match Linux to Windows synchronization primitives. From ntsync's official description:

It exists because implementation in user-space, using existing tools, cannot match Windows performance while offering accurate semantics.

So without this, you either have a huge perfomance hit in case of an accurate implementation or you have good performance, but might run into edge cases where software doesn't work well or at all because it's not accurate (see github.com/ValveSoftware/Proto… for examples)

fsync issues · Issue #2922 · ValveSoftware/Proton

If you can reproduce a crash or performance regression with Proton 4.11 that consistently goes away with PROTON_NO_FSYNC=1, please post it here.

^GitHub

I don't think his statement is true though. If reddit.com/r/linux_gaming/comm… is not manipulated in any way, games with lots of these calls still get big improvements with ntsync over fsync (about 30% in this particular case, which is a massive boost). So while nobody can rule out that his statement may be true on average or in general, there are still cases where ntsync offers a tangible advantage – be it improved FPS or the fact that the game runs at all.

Edit: in the video that the thread is about, fsync didn't beat ntsync in a single one (or I missed it when jumping through it). In the best one, they were exactly tied. Sure, the difference wasn't really big, but again there are titles not working with fsync.

However, I want to stress that I'm not trying to talk about fsync. It's a good solution that significantly improved performance. But ntsync is, from everything I've seen, almost always better; how much depends on the case, and it never seems to be worse.