AMD drivers are getting into very good shape and are backed by AMD itself (maybe not as much as it should be, but they keep on improving the closed source ones in the meanwhile).
Only leaves the nvidia drivers. That is a problem. But i don't see any reason for nvidia to support an unused display server just yet. If it became widely used, i have no doubts that nvidia would be quick to support it.
Bottom line, i don't really see a big problem in the driver department.
PM is *here*, its dynamic pm that is not here. Its not fatal.
Intel: PM - full
AMD: PM - manual state switch (high/low)
Nouvau: PM - prior to Fermi - full, Fermi and up - minimal(only low)
Also, follow this advice.
Without opensource drivers, there won't be any wayland, at all!
This is same old BS as with Logitech crap. If Logitech refuses to support Linux, its Logitech sucking, not Linux.
If Nvidia or Catalyst don't support Wayland, its them sucking, not Wayland. Also, Wayland is new technology, so chicken-egg. This problem never exists with opensource!
Damn, here for you
The problem with nouveau is lack of motivation behind its development, due to nvidia not opensourcing anything.
This is especially problematic with reclocking on Fermi and up, so the card is kept in lowest power mode, always.
Last edited by brosis; 06-12-2013 at 03:34 PM.
It certainly is fatal for a lot of laptop users, especially those who want to run a 3D desktop.PM is *here*, its dynamic pm that is not here. Its not fatal.
Even on low power-settings my laptop (and those of many others with AMD chipsets) runs 15°C hotter than on Catalyst, making this driver unusable on the hardware. So what does it matter that there is manual settings if it is still crap? And yes, there is a bug report for it, marked as critical, but nothing happens. I suspect just because AMD is working on power-saving code that maybe will released, maybe not, dependent on what the legal department says.
Catalyst is a must on that machine and if you look at AMD's enormous development speed maybe we will see Wayland support in 3-5 years in Catalyst.
Also, Nouveau (just like any other Gallium driver) only supports OpenGL 3.1. Support is coming slowly but surely.
The problem with nouveau isn't the lack of motivation. It is both the lack of time from current devs (most of us are students) and lack of qualified man power! As for not having docs, it is quite possible that if we had all the docs, we would become bored and wouldn't work on it as much (even though, as far as PM goes, implementing the driver takes me like 0.1% of my time, the rest goes to reverse engineering).
In my case I would love to have some kind of PLL programming guide explaining me stuff I cannot possibly reverse engineer without an electrical engineering degree and an electronic microscope. This part is no fun for me :s
Anyway, I am writing a paper on power management for NVidia GPUs, I'll try to advertise it so as you can have an idea of what full PM means!
Also I'm not saying wayland sucks because the FOSS drivers suck, I'm just saying a good technology that relies on sucky technology will get nowhere until the sucky software is actually the least bit useful.
Last edited by chris200x9; 06-12-2013 at 04:23 PM.
only intel have a good opensource drivers, because the company are really dev the drivers.
radeon makes me crazy on my laptop and i cant use hybrid graphics.
most of users have a laptop and the opensource radeon are not a solution for them
about wayland its more of the same, 5 years and not ready, too much time for some people (people who use laptops) and have hybrid systems who really don t like xorg
but this post of yours seems to say the opposite.
Are there any guides/are you guys writing notes of your work that would enable other volunteers to pick up
some of the work of REing, to speed the whole process up?