Scheduled Maintenance: We are aware of an issue with Google, AOL, and Yahoo services as email providers which are blocking new registrations. We are trying to fix the issue and we have several internal and external support tickets in process to resolve the issue. Please see: viewtopic.php?t=158230
xserver-xorg-video-intel driver on newer Intel graphics
xserver-xorg-video-intel driver on newer Intel graphics
Hi,
I checked on the website today, in Debian sid, "xserver-xorg-video-intel" driver on newer Intel graphics cards are not necessary anymore:
https://packages.debian.org/sid/xserver ... ideo-intel
As of the time of writing, it is version "2:2.99.917+git20160522-1". Since I am not always up-to-date with Linux driver development, can somebody explain why this is? Does this mean that the Linux Kernel now contains the full driver of Intel graphics cards produced from year ~2007 onward?
Wirawan
I checked on the website today, in Debian sid, "xserver-xorg-video-intel" driver on newer Intel graphics cards are not necessary anymore:
https://packages.debian.org/sid/xserver ... ideo-intel
As of the time of writing, it is version "2:2.99.917+git20160522-1". Since I am not always up-to-date with Linux driver development, can somebody explain why this is? Does this mean that the Linux Kernel now contains the full driver of Intel graphics cards produced from year ~2007 onward?
Wirawan
- Head_on_a_Stick
- Posts: 14114
- Joined: 2014-06-01 17:46
- Location: London, England
- Has thanked: 81 times
- Been thanked: 132 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
The advice is to drop the Intel DDX driver (xf86-video-intel) and instead adopt the modesetting driver that is included with the Xorg package.
The advantages of this package are Glamor acceleration and DRI3 page flipping as standard and (hopefully) fewer bugs and a more stable performance.
The advantages of this package are Glamor acceleration and DRI3 page flipping as standard and (hopefully) fewer bugs and a more stable performance.
Not quite, the Intel drivers have three parts -- the DDX driver (either xf86-video-intel or the Xorg modesetting driver), the Mesa driver (for 3D acceleration) and the kernel component (the i915 module).Does this mean that the Linux Kernel now contains the full driver of Intel graphics cards produced from year ~2007 onward?
deadbang
Re: xserver-xorg-video-intel driver on newer Intel graphics
Thanks for the explanation.
Can you give an advice what to use on a Debian stable (jessie) platform? I don't tweak that much except that I use kernel 4.6 right now. Is the xserver provided by this distro ready for a test drive? dpkg -l gives this version for package xserver-xorg: 1:7.7+7 .
Wirawan
Can you give an advice what to use on a Debian stable (jessie) platform? I don't tweak that much except that I use kernel 4.6 right now. Is the xserver provided by this distro ready for a test drive? dpkg -l gives this version for package xserver-xorg: 1:7.7+7 .
Wirawan
- Head_on_a_Stick
- Posts: 14114
- Joined: 2014-06-01 17:46
- Location: London, England
- Has thanked: 81 times
- Been thanked: 132 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
Yes, this thread applies to Debian jessie:wirawan0 wrote:Can you give an advice what to use on a Debian stable (jessie) platform?
https://forums.bunsenlabs.org/viewtopic.php?id=1185
deadbang
Re: xserver-xorg-video-intel driver on newer Intel graphics
Thanks for the response. I tested this configuration and the generic modesetting driver worked well.
How (for newbies):
As root, create a file called containing only:
Then restart X. Note that I used kernel 4.6 Jessie backports per bunsenlab forum page listed before. So far it worked well. I will update if I see issues.
There are several issues I observed:
1) xbacklight could not be used anymore, although the display intensity can still be adjusted via hardware key (Fn+F5/Fn+F6). The hardware is Thinkpad T450s.
2) Also on Xorg.0.log it noted the following:
I think this is probably an easy fix? Or no fix yet for this kernel? I checked, however, that: youtube video was smooth, vlc and mplayer can play video well, as well as google maps rescaling of map that used to be a problem under the intel DDX--all were gone. But using software rendering, that seems to be an issue.
How (for newbies):
As root, create a file called
Code: Select all
/etc/X11/xorg.conf.d/20-use-generic-modesetting.conf
Code: Select all
Section "Device"
Identifier "Intel Graphics"
Driver "modesetting"
EndSection
There are several issues I observed:
1) xbacklight could not be used anymore, although the display intensity can still be adjusted via hardware key (Fn+F5/Fn+F6). The hardware is Thinkpad T450s.
2) Also on Xorg.0.log it noted the following:
Code: Select all
[ 81.929] (II) AIGLX: Screen 0 is not DRI2 capable
[ 81.929] (EE) AIGLX: reverting to software rendering
[ 81.956] (II) AIGLX: Loaded and initialized swrast
[ 81.956] (II) GLX: Initialized DRISWRAST GL provider for screen 0
- Head_on_a_Stick
- Posts: 14114
- Joined: 2014-06-01 17:46
- Location: London, England
- Has thanked: 81 times
- Been thanked: 132 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
I don't think this is an issue -- the modesetting driver uses DRI3 page flipping by default as per modesetting(4)wirawan0 wrote:Also on Xorg.0.log it noted the following:
Code: Select all
[ 81.929] (II) AIGLX: Screen 0 is not DRI2 capable [ 81.929] (EE) AIGLX: reverting to software rendering [ 81.956] (II) AIGLX: Loaded and initialized swrast [ 81.956] (II) GLX: Initialized DRISWRAST GL provider for screen 0
Check hardware rendering using `glxinfo|grep rendering` from mesa-utils
deadbang
Re: xserver-xorg-video-intel driver on newer Intel graphics
Update on this issue.
1) xbacklight not working was annoying, but not the worst problem.
2) The worst problem is the one I described in this bug that nobody responded to (so far):
https://mail.google.com/mail/u/0/#searc ... 6135e640b4
In brief, the X display will become unresponsive when certain legacy X software was invoked. Some examples shown in the bug report include:
a) running osd_cat plus gtk showing a pop-up notification on the screen corner
b) running xpdf to display a corrupt PDF file
After that, the display would become many seconds late to update the content, which mainly noticed with text editing (like, characters not showing up in GUI terminal until 3-5 seconds). Actually the bug report title was misleading. Instead of X keyboard becoming slow & responsive, the display itself becae very unresponsive after one of certain events was triggered--see the bug report for detail. These should not have happened at any time, otherwise it is like a time bomb to the X display. The problem would persist until I restart the X display. I wonder if anyone of you ever experienced problem like what I described in this bug report.
I was using Backport kernel (4.3, 4.6, or 4.7 from jessie-backports), otherwise the system is using jessie (stable) distribution.
Wirawan
1) xbacklight not working was annoying, but not the worst problem.
2) The worst problem is the one I described in this bug that nobody responded to (so far):
https://mail.google.com/mail/u/0/#searc ... 6135e640b4
In brief, the X display will become unresponsive when certain legacy X software was invoked. Some examples shown in the bug report include:
a) running osd_cat plus gtk showing a pop-up notification on the screen corner
b) running xpdf to display a corrupt PDF file
After that, the display would become many seconds late to update the content, which mainly noticed with text editing (like, characters not showing up in GUI terminal until 3-5 seconds). Actually the bug report title was misleading. Instead of X keyboard becoming slow & responsive, the display itself becae very unresponsive after one of certain events was triggered--see the bug report for detail. These should not have happened at any time, otherwise it is like a time bomb to the X display. The problem would persist until I restart the X display. I wonder if anyone of you ever experienced problem like what I described in this bug report.
I was using Backport kernel (4.3, 4.6, or 4.7 from jessie-backports), otherwise the system is using jessie (stable) distribution.
Wirawan
- gradinaruvasile
- Posts: 935
- Joined: 2010-01-31 22:03
- Location: Cluj, Romania
- Contact:
Re: xserver-xorg-video-intel driver on newer Intel graphics
This again. On Skylake Intel the modesetting driver is bad even on testing compared to intel's DDX - tearing everywhere, slow updates etc.
Intel works pretty much ok, you have tearfree option that works etc.
Another thing that maybe is overlooked by people advocating modesetting is that on Jessie (with backports or not) the X server is old compared to testing/sid. Why is this important? Modesetting (which is itself a module in the X server) uses GLAMOR acceleration that is again a module INCLUDED in the X server, not like SNA externally provided by Intel DDX. The glamor/modeset code is actively developed and debugged on current X server versions but there is NO newer X server on jessie (or -backports).
But i use Testing and modesetting on Skylake isn't that good on it either.
BTW Intel's DDX is developed (the version number actually contains the day the git snapshot was taken (which is 20161206, 06 december 2016, packaged on 14 December 2016 on Testing and something similar on backports).
These being said have to say Intel's video drivers are inferior to AMDs opensource drivers as far as stability and hardware acceleration is concerned. My Skylake laptop (I7-6820HQ quad core/8thread CPU, Intel 530 GPU) on Steam performs worse on it than my age old desktop AMD IGP (r600g/Radeon HD 8570D) on Testing with the same mesa/Xserver version (despite having nominally support for OpenGl 4.5 vs 4.1 on AMD).
Intel works pretty much ok, you have tearfree option that works etc.
Another thing that maybe is overlooked by people advocating modesetting is that on Jessie (with backports or not) the X server is old compared to testing/sid. Why is this important? Modesetting (which is itself a module in the X server) uses GLAMOR acceleration that is again a module INCLUDED in the X server, not like SNA externally provided by Intel DDX. The glamor/modeset code is actively developed and debugged on current X server versions but there is NO newer X server on jessie (or -backports).
But i use Testing and modesetting on Skylake isn't that good on it either.
BTW Intel's DDX is developed (the version number actually contains the day the git snapshot was taken (which is 20161206, 06 december 2016, packaged on 14 December 2016 on Testing and something similar on backports).
These being said have to say Intel's video drivers are inferior to AMDs opensource drivers as far as stability and hardware acceleration is concerned. My Skylake laptop (I7-6820HQ quad core/8thread CPU, Intel 530 GPU) on Steam performs worse on it than my age old desktop AMD IGP (r600g/Radeon HD 8570D) on Testing with the same mesa/Xserver version (despite having nominally support for OpenGl 4.5 vs 4.1 on AMD).
Re: xserver-xorg-video-intel driver on newer Intel graphics
Sounds like we are really left in the cold for Intel graphics hardware.
Where is Intel here? Why are we left with buggy drivers for a long time?
I am surprised that nobody else observed issues that I list below.
Below are the list of choices that we have
* GOING WITH INTEL DDX, using "uxa" acceleration
Tested this with my laptop...Google maps causes X server processor usage
to shoot to 100%; and page scrolling on Firefox was extremely slow.
* GOING WITH INTEL DDX, using "sna" acceleration
This seems to be the best as far as stability, but it has issue of screen
corruptions at several known cases. See here for detail
https://bugs.debian.org/cgi-bin/bugrepo ... bug=851067
* GOING WITH MODESETTING DDX
As @gradinaruvasile mentions, the modesetting DDX has its own set of
problems, and probably the issue is worse than Intel/SNA combination.
Here is my prime example of why we don't want to use modesetting DDX on
Intel graphics:
https://bugs.debian.org/cgi-bin/bugrepo ... bug=835123
So, what are our options, other than what mentioned above?
I can't wait for Debian 9 (stretch) to come out & stabilize; but I need to
take time to migrate all my stuff to that OS, and there will be more new
set of issues to face.
I tested drive Debian 9 and found out that the graphics with modesetting
works fine. But the Intel DDX (+SNA) is still corrupt as ever.
Sounds like, backporting X server is a good workaround here, so
modesetting would work fine.
Wirawan
Where is Intel here? Why are we left with buggy drivers for a long time?
I am surprised that nobody else observed issues that I list below.
Below are the list of choices that we have
* GOING WITH INTEL DDX, using "uxa" acceleration
Tested this with my laptop...Google maps causes X server processor usage
to shoot to 100%; and page scrolling on Firefox was extremely slow.
* GOING WITH INTEL DDX, using "sna" acceleration
This seems to be the best as far as stability, but it has issue of screen
corruptions at several known cases. See here for detail
https://bugs.debian.org/cgi-bin/bugrepo ... bug=851067
* GOING WITH MODESETTING DDX
As @gradinaruvasile mentions, the modesetting DDX has its own set of
problems, and probably the issue is worse than Intel/SNA combination.
Here is my prime example of why we don't want to use modesetting DDX on
Intel graphics:
https://bugs.debian.org/cgi-bin/bugrepo ... bug=835123
So, what are our options, other than what mentioned above?
I can't wait for Debian 9 (stretch) to come out & stabilize; but I need to
take time to migrate all my stuff to that OS, and there will be more new
set of issues to face.
I tested drive Debian 9 and found out that the graphics with modesetting
works fine. But the Intel DDX (+SNA) is still corrupt as ever.
Sounds like, backporting X server is a good workaround here, so
modesetting would work fine.
Wirawan
- gradinaruvasile
- Posts: 935
- Joined: 2010-01-31 22:03
- Location: Cluj, Romania
- Contact:
Re: xserver-xorg-video-intel driver on newer Intel graphics
I run Testing and Intel is far better than modesetting. Tried a few times after X was updated but nothing really changed.
- stevepusser
- Posts: 12930
- Joined: 2009-10-06 05:53
- Has thanked: 41 times
- Been thanked: 71 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
Well, jessie-backports already has some backports of the two base parts that take quite a lot of resources to compile, which are newer llvm and Mesa packages. In comparison, the xorg backports should be very quick to build in comparison, if you'd like to give them a try. There's just a lot more individual packages to do.
My Skylake 520 graphics seem to be fine in Firefox, and with no artifacts seen in video playback, with this 20-intel.conf file in /etc/X11/xorg.conf.d:
inxi -G output:
This is on the MX 16 Jessie-based distro which has its own llvm and Mesa backporter--though I remember that the Mesa 13.0.2 backport did create some artifacts for me in video playback in the openGL video output when the tearfree option was enabled, but Mesa 13.0.3 cleared it up for me.
My Skylake 520 graphics seem to be fine in Firefox, and with no artifacts seen in video playback, with this 20-intel.conf file in /etc/X11/xorg.conf.d:
Code: Select all
Section "Device"
Identifier "Device0"
Driver "intel"
BusID "PCI:0:2:0"
Option "AccelMethod" "sna"
Option "DRI" "true"
Option "TearFree" "true"
Option "DRI" "3"
EndSection
Code: Select all
Graphics: Card-1: Intel HD Graphics 520
Card-2: NVIDIA Device 179c
Display Server: X.Org 1.16.4 driver: intel Resolution: 1920x1080@60.01hz
GLX Renderer: Mesa DRI Intel HD Graphics 520 (Skylake GT2) GLX Version: 3.0 Mesa 13.0.4
MX Linux packager and developer
- Head_on_a_Stick
- Posts: 14114
- Joined: 2014-06-01 17:46
- Location: London, England
- Has thanked: 81 times
- Been thanked: 132 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
For my Haswell system under stretch, the modesetting DDX driver has finally cured my tearing problems and I no longer need a special driver option or a compositor to manage the issue.
deadbang
Re: xserver-xorg-video-intel driver on newer Intel graphics
I tried past day modesetting-glamor with DRI3 and kernel 4.10.1, and it feels a litle bit faster and more unstable than intel-sna,
Modesetting gives Tears in videos. (No TearFree option available )
Modesetting gives Tears in videos. (No TearFree option available )
bester69 wrote:STOP 2030 globalists demons, keep the fight for humanity freedom against NWO...
- Head_on_a_Stick
- Posts: 14114
- Joined: 2014-06-01 17:46
- Location: London, England
- Has thanked: 81 times
- Been thanked: 132 times
Re: xserver-xorg-video-intel driver on newer Intel graphics
Try it under Wayland instead — I can recommend the GNOME & stretch combination.bester69 wrote:Modesetting gives Tears in videos.
deadbang
- gradinaruvasile
- Posts: 935
- Joined: 2010-01-31 22:03
- Location: Cluj, Romania
- Contact:
Re: xserver-xorg-video-intel driver on newer Intel graphics
I would probably used modesetting too, but the no TearFree option + slower playback refresh rate are dealbreakers.
Re: xserver-xorg-video-intel driver on newer Intel graphics
Head, I tried weston-wayland in stretch and videos still got tears..Head_on_a_Stick wrote:Try it under Wayland instead — I can recommend the GNOME & stretch combination.bester69 wrote:Modesetting gives Tears in videos.
bester69 wrote:STOP 2030 globalists demons, keep the fight for humanity freedom against NWO...