Quote Originally Posted by brosis View Post
Bug fix speed for Intel is ~14 days.
Annoying bugs always take longer https://bugzilla.kernel.org/show_bug.cgi?id=49571

Seriously, intel has all sorts of problems with their driver unless you use lts kernels and software that has been tested for some months.
From 3.6 you get these hangs, in 3.8 rc I think I had one variation reported and it got fixed quite quickly I think, perhaps it fixed some others too. I still sometimes can trigger it, maybe once every 20 minutes when playing nitronic rush. Now however it can recover after a few seconds without losing 3d acceleration so it's not that bad anymore.

Currently in 3.9 rc and also in 3.8 it sometimes gets stuck when trying to from/to a tty to/from X.
Code:
[11538.531257] [drm:atom_op_jump] *ERROR* atombios stuck in loop for more than 5secs aborting
[11538.531262] [drm:atom_execute_table_locked] *ERROR* atombios stuck executing D46E (len 62, WS 0, PS 0) @ 0xD48A
Quote Originally Posted by brosis View Post
They are the driving force behind opensource and especially behind opengl. They work very hard with community and drivers are superb.
Yes, they do. But in my experience AMD does too. Try reporting a bug in the open source radeon driver and you have a good chance that in less than a day someone will have looked at it and posted a response or a patch to test.

Quote Originally Posted by brosis View Post
Nvidia is virtually bug-free. They also don't need community since their driver is superb.
I think they got better recently. A year ago or so I wasn't even able to rotate the screen without getting rendering glitches or X crashes in ubuntu.

Quote Originally Posted by brosis View Post
opensource HD7xxx - fail. Hardware is available since over a year.
Apparently s3tc is missing from radeonsi. That's not very good. On the other hand, many things render just fine, i.e. xonotic on "ultra" details. Performance could be better but is ok for now.