The First Wayland Benchmarks From Fedora 20 Show Great Promise

[tessio]

What about pure Wayland instead of XWayland? Because as far as I understand, XWayland will always be slower than X.

[Ericg]

What about pure Wayland instead of XWayland? Because as far as I understand, XWayland will always be slower than X.

In theory...actually not. Practice is obviously different than theory, but in theory there's ways to actually get a performance bump out of XWayland because you cut out layers of X's workings.

[stan]

What about pure Wayland instead of XWayland? Because as far as I understand, XWayland will always be slower than X.

I had the same question. What's the projected timeline for GNOME running on pure Wayland (no X whatsoever)?

[mrugiero]

Performance is doing much worse than XMir, which isn't very promising at all. Performance hit is over 10% in Xonotic in comparison to XMir from a month ago which only had a 5% hit.

Thats the (part of) Linux community looking at things.

For Mir it is "regression"
For Wayland it is "promise"

Right. Except, you are comparing XMir + composite bypassing against XWayland without composite bypassing.
Please, now compare with XMir before composite bypassing.
Also, XMir expected you to run your whole desktop with such a performance hit, while XWayland is only to allow legacy apps to run. The desktop, and preferably everything, is expected to be ran natively. I'm really glad they changed their mind for 13.10, and I hope they do the same for 14.04 (either to go pure Mir, or to stay with X.org; I'd be extra happy if they'd go Wayland, but that would be dreaming too much).

Anyway, I did expected something titled as "promising" to show at least the same performance as pure X.org.

I had the same question. What's the projected timeline for GNOME running on pure Wayland (no X whatsoever)?

GNOME is running on pure Wayland. The games are running on XWayland, because there is no port yet.

Non-full-screen applications will probably run slower.

The nice thing of running a rootless X server is that all applications can be safely considered full-screen by X.

[halfmanhalfamazing]

Easy to switch to and use, Excellent!

For the "Wayland Tech Preview" in Fedora 20 the developers hope to make it fairly easy to try out Wayland by having it be an experimental option from the log-in screen.

I am glad this is how it will be. I'll definitely go poking around and submit some bug reports.

[mannerov]

Non-full-screen applications will probably run slower.

Remember that it is rootless: What is the concept of fullscreen for you?

A window without decorations would behave as if it was fullscreen.

Since Wayland compositing should be lighter than X, windowed games should perform great (when performance issues are fixed)

[drago01]

This has nothing to do with xwayland really. We currently do not try to let games by pass the compositor at all when running under wayland (yet). First you make thinks work and then you make it fast. Currently we are focused on getting stuff to run on wayland, it is a tech preview for a reason.

[Delgarde]

I had the same question. What's the projected timeline for GNOME running on pure Wayland (no X whatsoever)?

Depends on how you defined "running". You can already start a session and use Shell, but there's a bunch of stuff that doesn't work yet - e.g Gtk's Wayland port doesn't yet have working drag/drop, stuff like that.

[dee.]

Looking good for Wayland. I know someone who's going to wear a monkey suit...

(no, I'm not talking about Shuttleworth, you silly gooses)

Anyway, I'm probably going to have to install Fedora 20 on my test machine when it's released.

[jukkan]

Very promising first results actually. I'm sure it will get only better with time.