I'm using a corsair m95. And yeah can confirm, mouse response in 5.0-5 is weird.
I'm having a lot of mouse issues with older games too. I don't think the polling rate is the only cause. It wildly varies depending on game and wine version.
Try playing Star Wars Galaxies or Warhammer Online and you'll see these problems and more, amplified, depending on the Wine/Proton version. And changing settings from Wine, Proton, or the game, do not help at all. Believe me I tried everything. On SWG disabling hardware cursor helps but then it completely ruins the experience.
Depending on the version you get:
So why am I typing all of this? Because I think that time has proved that it's impossible to find a solution that won't eventually break or fix all the issues (without introducing new ones). I think the options you find on winecfg and winetricks are too confusing (maybe misleading would be a better word) since they don't always work like stated. IMO there has to be multiple code paths you can choose that fundamentally change how input is handled. Right now you have to juggle between wine/proton versions and decide what problems you hate the least.
Just some food for thought, feel free to mark it as off-topic.
@username2222232 What model mouse are you having the problem with?
@username2222232 Also what specific game(s) are you seeing this issue in?
More info here, https://github.com/ValveSoftware/Proton/issues/3685
In addition - the recent stem client update worsened the mouse responsiveness for me in Quake Champions - despite using a custom dinpu8 implementation which read the mouse input directly using kernels evdev interface. Probably because of your gameoverlayrenderer implementation.
Please put more effort into input and output latency.
I've just downloaded Neverwinter and tried it with Proton 5.0-5 on Steam. I had input lag inside the game, I moved my mouse around and it seemed that it had about 100-120ms delay. Clicking the UI was also laggy, sometimes I had to click twice to confirm an action.
I have a Logitech G203 at 1000Hz polling rate, switching to 500Hz via Piper reduced the input lag, though it was still visible when moving my mouse too fast.
I play a game using OpenGL code that is outfitted with mouse input: DirectInput API, Win32 (with cursor ballistics), and Win32 (Raw Input excludes all Window Cursor accels).
Standard practice is to use the Win32(Raw) to negate all accels or simply use the Win32 input assuming accel should also be off. DirectInput works noticeably fine too.
On 125 Hz mouse things seem normal as on 4.11 branch (I've tested using a Microsoft Wheel Mouse). The moment you use 1000 Hz pro-gamer mouse you get treated to something awful. Maybe its the mesa version 20.0 but I doubt it because the 4.11 branch works fine.
1000 Hz a negative acceleration is deeply ingrained in my game. Accel is off on the OS-level. Raw input should negate any accel coming from the Wine environment. No... the slower I move my mouse hand the more accurate the sensitivity is to how it should be but if I move my hand very fast my distance traveled is 1/3rd as much.
In addition: the overall sensitivity in the game is now 1.80-1.84 times slower. This game uses same sensitivity as many other games (similar formula for 360 degrees) and it's clear what the original speed should be.
I'm not too optimistic if this will be fixed. I'm using 4.11 branch for now and praying it doesn't get messed up either. It's really not a surprise for me because historically even when Loki games was porting Linux titles over, many had some kind of acceleration glitches for 4-5 years before a fix was done. I don't understand why no one else is seeing this. My Logitech mouse is quite popular. Many people using Mesa 20.0, and OpenGL many games running on.
I'm hoping more people can chime in and give a penny.