Which Rendering Backend - OpenGL 2.0 or 3.1?

[bendy]

I notice there is an option under Settings/Display and Monitor/Compositor to change the rendering backend for the desktop effects. It's defaulted to OpenGL 2.0 on my machine, but is there any benefit in using OpenGL 3.1?

I'm using the proprietary nvidia driver, if that makes a difference.

[GreyGeek]

What ever works best. GL 3.1 causes Stellarium and other graphics to flash and blink. GL 2.0 or XRender both work well on my box. I chose to go with XRender. IF I could run GL 3.1 I would.

[Rog131]

from the horse's mouth...

The Compositing Modes of KDE Plasma Workspaces Explained: https://blog.martin-graesslin.com/bl...ces-explained/

Which opengl option?: https://www.reddit.com/r/kde/comment...opengl_option/



Information about the currently running instance, which options are used,
what OpenGL driver and which effects are running.

qdbus org.kde.KWin /KWin supportInformation

KDE System Settings > Display and Monitor > Compositor > Rendering backend

At here...






As default, Rendering backend: OpenGL 2.0:

grep OpenGL ~/.xsession-errors

OpenGL vendor string: X.Org
OpenGL renderer string: Gallium 0.4 on AMD BARTS (DRM 2.43.0, LLVM 3.7.1)
OpenGL version string: 3.0 Mesa 11.2.1
OpenGL shading language version string: 1.30
OpenGL version: 3.0

If I turn the 'Rendering backend: OpenGL 3.1':

grep OpenGL ~/.xsession-errors

OpenGL vendor string: X.Org
OpenGL renderer string: Gallium 0.4 on AMD BARTS (DRM 2.43.0, LLVM 3.7.1)
OpenGL version string: 3.3 (Core Profile) Mesa 11.2.1
OpenGL shading language version string: 3.30
OpenGL version: 3.3

[bendy]

I changed to OpenGL 3.1 and it works fine, although I haven't noticed any improvements either. Checking per Rog131's command lines confirms 3.1

[dequire]

I wonder, if one turned on every KWin desktop effect, if there would be any difference at all? I'm guessing not, since I would assume that the composter and KWin effects, etc. would all be coded to use OpenGL 2.0 standards. It wouldn't make sense to require 3.0+ and so therefore not likely to gain anything from it. Unless the code would happen to execute more efficiently using the 3.0+ standard. Even if it did, it would likely not be visibly noticeable by a user.