Conversation
Notices
-
Embed this notice
Wayland niggers have convinced themselves the 33ms of sloppy cursor latency introduced by their lame ass implicitly synced compositor isn't real. Despite it being measurably real.
- :ihavenomouth: and pwm like this.
-
Embed this notice
@crunklord420 why do you need more than 3 frames per second? it's not 1998 anymore
-
Embed this notice
@wowaname yeah, on hyprland. I did all the nvidia shit, modesetting. I think it's just like that, by design. People seem to just ignore that it's a problem and it exists. It's all part of their "perfect frame" ethos, which they've already broken since it was a fundamentally flawed design decision.
X11 has hardware cursor, Wayland does not. Read the cope post by Hector Martin's vtuber anime-girl alter-ego.
https://web.archive.org/web/20250130091037/https://lobste.rs/s/oxtwre/hard_numbers_wayland_vs_x11_input_latency
-
Embed this notice
@crunklord420 wew it's really that bad?
-
Embed this notice
@wowaname and it's not just like sub-frame latency like you'd expect from a video game. It's like multiple frames. I run a 144hz monitor at it feels like 33ms, that's 2x60hz frames. Everything feels so fucking sluggish like it's quadruple buffered.
-
Embed this notice
@crunklord420 @wowaname didnt read the image yet, is this nvidia only?
-
Embed this notice
@mischievoustomato @wowaname no it's like that for everything from what I could tell.
-
Embed this notice
@crunklord420 @wowaname im not susceptible to it it seems, been running wayland for 4 years on gnome so far
-
Embed this notice
@mischievoustomato @wowaname have you ever tried running a Xorg WM without a compositor before?
-
Embed this notice
@crunklord420 @wowaname i have, and i dont recall feeling a difference when trying wayland on gnome
-
Embed this notice
@wowaname honestly it feels like a really badly programmed video game where the engine "eats" your inputs depending on your framerate.
Like even game engines understand this, they literally will have "hardware mouse" as an optional feature because they know embedding the cursor into the game framebuffer makes it feel soupy.
I don't know. I'm not certain, maybe it's acceleration, maybe it's the framebuffer, maybe it actually eats inputs.
-
Embed this notice
@crunklord420 ok but let me get this straight: is the delay only for how the cursor is displayed, or is there input latency to match? because in the former case, the cursor still doesn't represent its true position, and in the latter, they're just trading off visual latency for actual latency. both are bad in their own ways.
-
Embed this notice
@wowaname I messed with this a bit more. It just seems like when you don't have hardware cursor at 60hz, it feels awful. At 144hz it's tolerable, but when I go back to Xorg the difference is obvious.
I don't understand these people and tearing. I'm staring at my 60hz monitor and like yeah I guess I can see multiple cursors on the screen at the same time if I wiggle it fast. I guess it is tearing. Maybe it's my persistence of vision, I can't tell. But my brain says it's fine, it knows where the cursor is.
Tearing is digital motion blur, it's part of our vision, it's not a scourge that has to be destroyed.
-
Embed this notice
@wowaname in an alternate timeline I build a hardware mouse emulator and high speed light sensor to do full round trip latency testing through the Linux graphics stack. Then proceed to harass every fucking Red Hat soydev with numbers about how their shit is fucking slow and they have no excuse.