Why these benchmarks target X11 directly instead SDL to make them window system agnostic?
Can you have SDL libraries for the X11 backend and another for wayland backend installed in the same system?
Last edited by newwen; 12-26-2013 at 08:33 PM.
Michael is right, performance are not becoming better yet. And there's no reason for that right now. The main point now is make things work at least in XWayland. Later, apps supported by Wayland will appear and we start to seek bottlenecks and enhance code.
Like the others said, take any game you could possibly run on SDL for Wayland and compare it with the same game on XWayland or X11. Certainly this is possible with some work? That's where the really interesting comparisons would lie, even if it's not purely scientific. Perhaps next year these kinds of tests will become more common.
Most of the games tested haven't even been ported to SDL2 yet, at least not in released form, which is a prereq for using the Wayland back-end.
Originally Posted by scionicspectre
Since when all the other Mir threads where related to Wayland?
Originally Posted by Hamish Wilson
I don't say that for you personally. I am just tired of those comments and thats why I did this comment. My mistake that I feed the trolls.
Lets get this thread back on track.
Why everyone expect Wayland and/or Mir to offer better berformance than Xorg? Their existence has to do with better quality, not performance. Am I wrong?
Originally Posted by rudregues
Last edited by verde; 12-27-2013 at 03:29 AM.
Also performance. Both Mir API and wayland protocol require less roundtrips than X11 for displaying something, and are both asynchronous, so we do expect better native performance (in terms of framerate but also in terms of time-to-screen / lag).
Originally Posted by verde
When considering Mir + XMir or wayland + Xwayland, you have a full X server + the Mir/wayland layer compared to a full X server, so the performance can only be equal at best.
Regarding your Mir/wayland taunts, please consider that Mir is not (should not) be criticized on a technical basis, as it's architecture is very similar to wayland, although more targeted / less generic.
What is considered "bad" is that they could have entirely (and as easily) implemented the whole "Mir thing" as a wayland shell, fitting every technical necessity that they have put forward so far for Mir. Instead, they created their own API, which although being open source, is not made to be used by any other DE apart unity, so that their work will be absolutely useless for the entire gnu/linux community except for canonical.
And this as been (justifiably) considered as a dick move by said community, which has (understandably) stirred some harsh (and sometimes not so rational) comments.
Because Canonical announced the project as being against Wayland with a barrage of FUD. There is a reason why this keeps coming up.
Originally Posted by verde