Re: Upstream Default (FOSS) DDX Driver for NVidia GPUs is not Nouveau (was: Help request installing wi-fi/video-card prop...)
David Christensen composed on 2021-01-20 11:33 (UTC-0800):
> I have a ~2011 Dell Latitude E6520 laptop with Nvidia Optimus graphics.
> It has always required non-free firmware for the Wi-Fi. On Debian 9
> with the free nouveau driver, the graphics work fine; both internal and
> external. I recently beat my head against Debian 10, the free nouveau
> driver, and the non-free nvidia-legacy-390xx-driver; without much
> success [1].
> My ideal is a FOSS distribution that works OOTB with as little
> fiddle-faddeling as possible. I went back to Debian 9. I suggest that
> you try that.
> [1] https://lists.debian.org/debian-user/2021/01/msg00035.html
In 2008 a new technology DDX driver's development was begun:
<https://cgit.freedesktop.org/xorg/driver/xf86-video-modesetting/commit/?id=b212306ae7bd3a79fb290a32ebd3952c2cba846d>
The new technology made it hardware independent, able to be used by any GPU that
has a full-functioning DRM/KMS driver. 2D acceleration depended on Glamor. In
Debian, the xf86-video-modesetting package was named xserver-xorg-video-modesetting.
In 2014, the development stage of xf86-video-modesetting ceased.
<https://cgit.freedesktop.org/xorg/driver/xf86-video-modesetting/>
Its life as a separate package then ceased as well. It was moved into the server
package. I never found an announcement to the effect, but I believe this move in
effect made it the default DDX driver for any version of the Xorg server from
version 1.17.0 and up.
"...GLAMOR acceleration on works on the NVIDIA G80 (GeForce 8) GPUs and newer
where as Nouveau focuses on support going back to the Riva TNT2 days. Also, the
xf86-video-nouveau is designed to be simple and work well..."
<https://www.phoronix.com/scan.php?page=news_item&px=Nouveau-Vs-Modesetting>
In 2014 also, the last official release of the Intel DDX driver was made. Since
then, it continues to receive patches, but the focus of its developers has been on
the Modesetting DDX.
Debian 8 contained server 1.16.4 and the development version of
xf86-video-modesetting. This package disappeared with release of Debian 9, which
runs the 1.19.2 server.
Thus, as of Debian 9, those running the Nouveau DDX are not running the upstream
default, which is named, for better or worse, Modesetting.
Like most other distros, Debian by default installs a metapackage that includes
all optional DDX driver packages, which includes xserver-xorg-video-nouveau. I
cannot provide a reference to this, but IMO the distro's packagers find they must
do this in order to best support at initial installation the remaining antique
GPUs in service by users that the Modesetting DDX driver does not support. A side
effect of this inclusion of the optional DDX drivers is few users are aware that
the Modesetting even exists, much less than it being developers consider it to be
the optimal FOSS DDX.
I have 5 PCs currently running older NVidia GPUs, the newest a Quadro NVS 310 and
a GeForce GT 630. Each is in multiboot with various Debian and other distros, in
current, non-current, and pre-release versions. I also have a GeForce 8600GT on
the shelf that goes in and out of various PCs for testing purposes. All are
normally run on the Modesetting DDX driver, occasionally for testing purposes
only, on the Nouveau DDX. In no case have I found the Nouveau DDX to be better in
any respect than the Modesetting DDX.
I have multiple operating 64bit PCs with Intel IGPs. All that are supported by it
are run on the very same Modesetting DDX driver. The situation is roughly the same
here with AMD APUs, IGPs and GPUs. Some I use with the AMDGPU DDX, the rest, if
supported by it, are all run on the Modesetting DDX.
I have never installed any proprietary graphics driver on any of my own hardware,
so can't offer any guidance on maintenance or performance issues from personal
experience. That said, IMO, anyone who doesn't try the upstream default DDX, if
supported, before installing one of NVidia's own drivers is shortchanging himself
of the knowledge of whether or not maintenance of proprietary drivers when kernels
are changed is worth whatever benefit in performance is actually observable for
his own use case.
--
Evolution as taught in public schools, like religion,
is based on faith, not on science.
Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!
Felix Miata *** http://fm.no-ip.com/
Reply to: