refresh-rate ignored with experimental-backends

[Ristovski]

When enabling --experimental-backends the refresh-rate config option seems to be ignored under the glx backend.

Compton v6, master

[yshui]

This is intended. Also sw-opti is on deprecation track as well.

[Ristovski]

Does this mean glx will not support anything but vsync in the future?

[yshui]

@Ristovski Yes. But there is nothing else to support in regard to this. refresh-rate and sw-opti were hacks for cases when vsync doesn't work, which has become less relevant.

[morgoth6]

To be honest this bothers me a little too, but I assumed this can be change with new backend rework a forget about it. This was tested on Precision 5510 (optimus intel&nvidia) in modesetting mode using i3 desktop.

When I run glxgears without compton it shows me 60fps as expected with ~8% CPU utilization:

303 frames in 5.0 seconds = 60.508 FPS
300 frames in 5.0 seconds = 59.933 FPS`

But with compton (both regular and experimental GLX backend's with vblank&damage enabled) it takes ~20% CPU and shows:

5829 frames in 5.0 seconds = 1165.633 FPS
5936 frames in 5.0 seconds = 1187.190 FPS

And fun part I give a try to experimental xrender backend and it takes ~12% CPU and shows:

604 frames in 5.0 seconds = 120.776 FPS
601 frames in 5.0 seconds = 120.055 FPS

Using old compton I was able to get normal 60fps using swc and glx-swap-method=3 (funny this is only swap-method which works here in original compton version)

For reference:

name of display: :0.0
display: :0  screen: 0
direct rendering: Yes
Memory info (GL_NVX_gpu_memory_info):
    Dedicated video memory: 2048 MB
    Total available memory: 2048 MB
    Currently available dedicated video memory: 1903 MB
OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: Quadro M1000M/PCIe/SSE2
OpenGL core profile version string: 4.6.0 NVIDIA 430.09
OpenGL core profile shading language version string: 4.60 NVIDIA
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile

OpenGL version string: 4.6.0 NVIDIA 430.09
OpenGL shading language version string: 4.60 NVIDIA
OpenGL context flags: (none)
OpenGL profile mask: (none)

OpenGL ES profile version string: OpenGL ES 3.2 NVIDIA 430.09
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20`

**Version:** v6

### Extensions:

* Shape: Yes
* XRandR: Yes
* Present: Present

### Misc:

* Use Overlay: Yes
* Config file used: /home/morgoth/.config/compton.conf

### Drivers (inaccurate):

modesetting

I wonder as I can see "modesetting" in driver is automatic nvidia quirks introduced recently are applied ?

[Ristovski]

@yshui My point is, the old glx backend allows for setting a higher refresh-rate which get's rid of "input lag" while allowing to cap it to something sane like 120fps.

The only other way I can think of doing the same thing under the new backend is to run compton with vblank_mode=0 and then limit the fps with libstrangle or some other mechanism. Quite inefficient.

[Ropid]

@morgoth6:

This isn't the case here for me. The glxgears program always runs at 60fps, no matter if compton is running or not, and no matter what backend I try.

Perhaps you are doing something to the vblank_mode mesa setting in a script you use to start Compton?

[yshui]

@Ristovski That is not the proper way to solve the input lag.

[Ristovski]

@yshui It might not be, but it was a nice "workaround" that was simple enough. Do you know of a different way to solve it?

[yshui]

@Ristovski It's anything but simple.

I am working on the input lag, see #147

[Cat-Lady]

I would like to point out that ability to limit framerate to arbitrary number is direly missed feature, for other use cases - like, streaming the desktop remotely via encoded video stream (steam link-alike), where strangling the FPS have gigantic effect on network bandwidth required.

Sure - ideally, streaming software would allow to set the framerate itself - but in cases where it does not, this is a huge and disappointing regression.

[Shringe]

Vsync works very poorly on multi-monitor setups with X11, is there any chance this feature is ever coming back?