Early Ubuntu 14.04 Intel XMir Benchmarks

[curaga]

what is the point of using Xmir? i don't see one at all.

Because using XWayland through Waylandmir is too much layers.

[blackout23]

what is the point of using Xmir? i don't see one at all.

You don't have a choice with some applications. If the toolkit that the application uses doesn't have a Mir/Wayland backend there is no way you could use it otherwise. Wayland and Mir aren't drop in replacements for X11 and everything just magically starts working on these protocols.
Wayland has rather good toolkit support in most modern toolkits without needing out of tree patches. GTK2 and Qt4 will probably never get a backport for Mir or Wayland. Mir has not a good toolkit support in upstream toolkits. Canonical is working on some patches for GTK, SDL etc. Wayland has support in GTK+,Qt5, SDL, EFL, Clutter upstream.

When an application uses X11 calls directly rather than through a toolkit you'd need XMir/XWayland aswell. But almost no apps use the X API directly.

[Attent?ter]

Xmir??

what is the point of using Xmir? i don't see one at all.

[Veerappan]

In this case, wouldn't a percent difference(between Xorg vs Xmir or Xwayland) be the fair comparison between them, even when running different mesa versions?

Or are you saying that changes in mesa to support XMir and XWayland would also contribute to a difference? Just curious, no rhetoric please.

I believe that there have actually been recent changes in Mesa which improve the performance of Wayland running on top of Mesa. I'd have to dig a bit for specific patches, but I believe this to generally be the case.

[dh04000]

Obviously there can't be made a direct comparison between XWayland and XMir, because of the different versions of Mesa that has been used for testing. And in the past there has been only made XWayland perfomance tests and not Wayland performance tests. It's a big difference.
Just to clarify this...

In this case, wouldn't a percent difference(between Xorg vs Xmir or Xwayland) be the fair comparison between them, even when running different mesa versions?

Or are you saying that changes in mesa to support XMir and XWayland would also contribute to a difference? Just curious, no rhetoric please.

[Bestia]

Anyone knows how the SDL Mir port is coming along? I think there is already some out of tree code. Wayland support has been merged into SDL2 2 weeks ago and should be available in the next release.

Brandon Schaefer works on SDL Mir support and he has now a three branches of code. Two for SDL 2 and one for SDL 1.2.

https://code.launchpad.net/~brandontschaefer/libsdl/sdl-mir-support

https://code.launchpad.net/~brandontschaefer/+junk/sdl-mir-support-v2

https://code.launchpad.net/~brandontschaefer/+junk/sdl1.2-mir

[profoundWHALE]

The performance is impressive considering everything.

[blackout23]

Anyone knows how the SDL Mir port is coming along? I think there is already some out of tree code. Wayland support has been merged into SDL2 2 weeks ago and should be available in the next release.

[mrugiero]

Pretty decent indeed.

I'm also glad they dropped the idea of running the desktop in XMir.

[Steffo]

Obviously there can't be made a direct comparison between XWayland and XMir, because of the different versions of Mesa that has been used for testing. And in the past there has been only made XWayland perfomance tests and not Wayland performance tests. It's a big difference.
Just to clarify this...