Provider Object Support For RandR 1.5

[entropy]

Please allow an amateurish question:

Is this in any way connected to a noticeable performance penalty?
Shoveling the framebuffer to another GPU doesn't sound like a terrible task.
1280x1024@32bit@60Hz (raw) should be 300 MB/s if my quick calculation
doesn't fail. But maybe there's more to it.

[airlied]

How about configuring outputs on a single GPU, so it can be used on multi-seat setup? I skimmed through the patch, but didn't saw such configuration options.

Got nothing to do with randr since randr runs inside an X server, and a X server is used by one user. multi-seat needs an X server per-user, so needs kernel changes, and no changes to randr at all.

Dave.

[FireBurn]

How about configuring outputs on a single GPU, so it can be used on multi-seat setup? I skimmed through the patch, but didn't saw such configuration options.

I believe that should already be possible

[Drago]

How about configuring outputs on a single GPU, so it can be used on multi-seat setup? I skimmed through the patch, but didn't saw such configuration options.

[phoronix]

Phoronix: Provider Object Support For RandR 1.5

The latest work by David Airlie for improving the X.Org infrastructure to handle modern GPU features (e.g. multi-GPUs, Optimus-like capabilities, etc) comes in the form of a proposed RandR protocol update...

http://www.phoronix.com/vr.php?view=MTExNzI