Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

damjan · 2009-01-12 16:34:42

When I am doing suspend, the screen is still a mess when I am using libdrm 2.4.3 and xf86-video 2.5.99.2
I have fbcondecor built into the kernel. is that a problem?

fbcondecor as itself is not a problem... but I've noticed that a framebuffer console is a problem wrt suspend/resume. So I use a plain old VGA console now.

I use the xorg-server from Extra.. so nothing special there.

Axell · 2009-01-12 20:58:41

Hello there)
I've never had problem with smoothness in compiz with my i945GM... But 3 things: reflection, blur and water effects always killed it. Even with workarounds... I'm using 2.6.28 kernel now and I don't see anything new with intel video driver. May be I've done something wrong?) Suppose, I have to wait until it's in repo)

BTW, fbcondecor hangs up laptop on Xorg reload sometimes. Like, when I logout(

Cheesebaron · 2009-01-12 21:37:47

Well you have to use the newest version 2.5.99.

Axell · 2009-01-13 10:45:38

Well, I hope to see it working when 2.5.99 are in repo)

phabulosa · 2009-01-14 18:07:57

damjan wrote:

When I am doing suspend, the screen is still a mess when I am using libdrm 2.4.3 and xf86-video 2.5.99.2
I have fbcondecor built into the kernel. is that a problem?
fbcondecor as itself is not a problem... but I've noticed that a framebuffer console is a problem wrt suspend/resume. So I use a plain old VGA console now.
I use the xorg-server from Extra.. so nothing special there.

When I find a way to allow new xf86-video-intel and fbcondecor get along, could you please let us know?

csmart · 2009-01-15 04:32:52

I built all packages (as suggested by mutlu_inek) on my Dell Latitude X1 laptop, but X wouldn't start.
I also built the vesa driver and it did WORK, but using the xf86-video-intel caused a segfault.

The log looked like:

(II) intel(0): Output VGA using monitor section Monitor0
(II) intel(0): Output LVDS has no monitor section
(II) intel(0): I2C bus "LVDSDDC_C" initialized.
(II) intel(0): Attempting to determine panel fixed mode.
(II) intel(0): I2C device "LVDSDDC_C:E-EDID segment register" registered at address 0x60.
(II) intel(0): I2C device "LVDSDDC_C:ddc2" registered at address 0xA0.
(II) intel(0): EDID vendor "LCD", prod id 22528
(II) intel(0): DDCModeFromDetailedTiming: 1280x768 Warning: We only handle separate sync.

If you get this problem, it appears to be caused by the intel driver wanting to enable the VGA port and not the LVDS laptop display.
I fixed this by creating a custom xorg.conf as follows:

Section "Module"
    Load        "glx"
    Load        "dri"
    Load        "v4l"
EndSection

Section "Device"
    Identifier    "Intel Video Card"
    Driver        "intel"
    Option        "Monitor-VGA" "External Monitor"
    Option        "Monitor-LVDS" "Laptop Monitor"
EndSection

Section "Monitor"
    Identifier    "External Monitor"
    Option        "Ignore"    "true"
EndSection

Section "Monitor"
    Identifier    "Laptop Monitor"
    Option        "PreferredMode" "1280x768"
EndSection

Hope this helps.
-c

Last edited by csmart (2009-01-15 05:21:32)

cywhale · 2009-01-15 09:48:10

@csmart: Thank you very much, this helped a lot, new Xserver is starting now without a problem.

With all the new packages from page #1
- glxgears on my travelmate C110/1Ghz Celeron says 61,
- xyz FPS, Tuxracer went up from ~24fps (old system) to ~32fps (new libs, Xorg, driver).
- Compiz Fusion (repo) seems to be a little bit faster than before.

noticed one strange problem - even using the new evdev PKGBUILD there is set a wrong keyboard layout in gdm and X until i start compiz fusion/fusion icon for the first time - after this keyboard layout seems to be DE as it used to be !?

Thank you for this thread.

big_gie · 2009-01-15 14:57:10

xf86-video-intel 2.6 has been released last night!
http://www.phoronix.com/scan.php?page=n … &px=Njk5NQ

agapito · 2009-01-15 15:18:55

2.6 still sucks on my comp, seems stable but 3D performance is horrible on my GMA945

csmart · 2009-01-15 21:38:33

@cywhale
Actually my keyboard didn't work properly either when using evdev. I use wmii and my super key (mod4) wouldn't work. I turned off evdev and it works fine.

Section "ServerFlags"
     Option "AutoAddDevices" "False"
EndSection

Then you'll need to make sure you have entries for your mouse and keyboard (like the old days).

-c

Last edited by csmart (2009-01-15 21:45:36)

Ronin-Sage · 2009-01-15 21:59:30

I wonder, is it worth scrapping my downgraded packages in favor of xf86-video-intel 2.6(+ associated packages)? I have an Intel 950GMA, and am wondering if it's possible to get appropriate performance with kernel 2.6.28 and the new libgl, mesa, etc. packages.

mutlu_inek · 2009-01-16 07:21:25

Ronin-Sage wrote:

I wonder, is it worth scrapping my downgraded packages in favor of xf86-video-intel 2.6(+ associated packages)? I have an Intel 950GMA, and am wondering if it's possible to get appropriate performance with kernel 2.6.28 and the new libgl, mesa, etc. packages.

I have to agree with agapito, the GMA950 and similar cards are horribly slow with the new mesa/x server/intel 2.6.0 on EXA and UXA, compared to previous XAA performance. I hope this improves soon.

microcz · 2009-01-16 09:55:13

hope intel will do something for people with gma950 and gmaX3100 cards ..the peformance is more than horrible, since i upgraded to intel from depreceated i810 (which worked fine) i have had only problems with acceleration, i can't run almost nothing, even tuxracer or other opengl programs. My friends who has same cards report same problems.

Dheart · 2009-01-16 10:44:03

gma X3100 works fine with newest driver, intel-dri, mesa and 2.6.28.
Tuxracer framerate is 24-28, Frozen Throne - 24-30
Complete playable. I get more than twice the framrate I got with 2.4.0 and older Xorg stuff (before the big upgrade that brought in device autodetection and ruined intel cards.)

cywhale · 2009-01-16 20:54:10

Made some tests with the new intel driver (2.6.0) using kernel 2.6.28 from extra and the packages listed on page #1 of this thread.
As a result 2D/3D performance seemes to be improved quite a bit using UXA without Compiz Fusion, tuxracer actually is playable @1024x768 (1Ghz Celeron ULV, Intel 855) now.

Using Compiz there are no noticable improvements anymore (tuxracer went a little bit faster), 30fps CF-benchmark and really slow Firefox scrolling. Window movement, cube, fade and minimize effects seem to be quite smooth now.

Compiled packages
------------------------------------------------------------

libpthread-stubs
libdrm
libxcb
libx11
dri2proto-git
xf86driproto-git
libgl-git
mesa-git
inputproto
pixman
randrproto
xorg-server
intel-dri
xf86-video-intel
xf86-input-keyboard
xf86-input-mouse
xf86-input-synaptics

EXA benchmarking
------------------------------------------------------------
GtkEntry - time: 0,08
GtkComboBox - time: 2,44
GtkComboBoxEntry - time: 1,78
GtkSpinButton - time: 0,60
GtkProgressBar - time: 0,56
GtkToggleButton - time: 0,65
GtkCheckButton - time: 0,42
GtkRadioButton - time: 0,64
GtkTextView - Add text - time: 1,48
GtkTextView - Scroll - time: 0,75
GtkDrawingArea - Lines - time: 2,15
GtkDrawingArea - Circles - time: 4,69
GtkDrawingArea - Text - time: 13,00
GtkDrawingArea - Pixbufs - time: 0,54
---
Total time: 29,77

glxgears: 401
tuxracer: 21-29

EXA/Compiz benchmarking:
------------------------------------------------------------

GtkEntry - time: 0,14
GtkComboBox - time: 2,97
GtkComboBoxEntry - time: 2,17
GtkSpinButton - time: 0,67
GtkProgressBar - time: 0,64
GtkToggleButton - time: 0,73
GtkCheckButton - time: 0,46
GtkRadioButton - time: 0,69
GtkTextView - Add text - time: 1,58
GtkTextView - Scroll - time: 0,86
GtkDrawingArea - Lines - time: 9,84
GtkDrawingArea - Circles - time: 28,58
GtkDrawingArea - Text - time: 18,42
GtkDrawingArea - Pixbufs - time: 1,19
---
Total time: 68,96

glxgears/compiz fusion: 381
tuxracer/compiz fusion: 18-23
Compiz Fusion benchmark: 35

UXA benchmarking
------------------------------------------------------------

GtkEntry - time: 0,06
GtkComboBox - time: 3,69
GtkComboBoxEntry - time: 2,36
GtkSpinButton - time: 1,22
GtkProgressBar - time: 1,38
GtkToggleButton - time: 1,15
GtkCheckButton - time: 0,56
GtkRadioButton - time: 0,71
GtkTextView - Add text - time: 0,76
GtkTextView - Scroll - time: 0,53
GtkDrawingArea - Lines - time: 2,29
GtkDrawingArea - Circles - time: 3,86
GtkDrawingArea - Text - time: 1,59
GtkDrawingArea - Pixbufs - time: 0,28
---
Total time: 20,45

glxgears: 61
tuxracer: 25-45

UXA/Compiz Fusion benchmarking
------------------------------------------------------------

GtkEntry - time: 0,09
GtkComboBox - time: 3,30
GtkComboBoxEntry - time: 2,65
GtkSpinButton - time: 1,52
GtkProgressBar - time: 1,70
GtkToggleButton - time: 1,36
GtkCheckButton - time: 0,67
GtkRadioButton - time: 0,91
GtkTextView - Add text - time: 1,14
GtkTextView - Scroll - time: 1,06
GtkDrawingArea - Lines - time: 12,92
GtkDrawingArea - Circles - time: 32,41
GtkDrawingArea - Text - time: 4,59
GtkDrawingArea - Pixbufs - time: 0,84
---
Total time: 65,16

glxgears/compiz fusion: 61
tuxracer/compiz fusion: 22-45
Compiz Fusion benchmark: 30

Last edited by cywhale (2009-01-16 21:43:58)

woifi · 2009-01-18 11:52:22

lo there,

does anyone know when xf86-video-intel 2.6 and the whole other stuff will come to the testing repo?

tia,

woifi

Ronin-Sage · 2009-01-18 20:13:29

So the issues seem to be limited to 945GM and lower chipsets, then?

cywhale · 2009-01-18 21:17:12

After 24h testing I got strange distortions (char 'e') using the new packages, don't know if it's Xorg, Intel driver or another package causing this, posted a Screenshot in my blog post (including the screenshot here would look quite ugly because of white background and shadow, sorry for direct linking). Did anyone observe something similar?

Last edited by cywhale (2009-01-18 21:33:49)

stefanwilkens · 2009-01-18 21:38:38

http://www.phoronix.com/scan.php?page=a … _uxa&num=1

Somehow things do seem promising

Dheart · 2009-01-18 22:34:48

I've been running the upgraded packages (old xorg-server, though) for a half a month already, and I have yet to notice graphical corruption. (Except in Frozen Throne direct3d mode (which noone uses, because opengl works fine))

dakor · 2009-01-21 15:17:01

I upgraded to the Intel 2.6 driver and compiz gets a way better framerate, but movie playback is horrible.

Question, now with the 2.6 intel driver and exa is the "MigrationHeuristic" "greedy" option still needed?

Edit: Also scrolling in Firefox with uxa is choppy

Last edited by dakor (2009-01-21 15:19:26)

b9anders · 2009-01-23 13:08:11

Tried building the packages today. Performance with exa feels somewhat better than xorg 1.4. Uxa is slightly worse and freezes the system if I try the no tiling option recommended.

Last edited by b9anders (2009-01-23 13:49:29)

mutlu_inek · 2009-01-29 06:13:50

Regarding issues (read: SLOWness) with GMA945 and earlier, there is good news (well, patience is still needed), see: http://virtuousgeek.org/blog/index.php/ … and_tiling

olovram · 2009-02-04 02:37:41

Amm, when was thiss going t extra?

baze · 2009-02-04 06:08:54

olovram wrote:

Amm, when was thiss going t extra?

1.6 final isn't out yet and it's not even in testing. considering unusable 3d performance on pre-965 chips due to no tiling in stable releases (requires patched libdrm, intel driver from git and a patched kernel) this should make it to testing first anyway.
so i'd say xorg 7.5 in extra is still a bit away...

Arch Linux

#76 2009-01-12 16:34:42

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#77 2009-01-12 20:58:41

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#78 2009-01-12 21:37:47

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#79 2009-01-13 10:45:38

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#80 2009-01-14 18:07:57

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#81 2009-01-15 04:32:52

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#82 2009-01-15 09:48:10

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#83 2009-01-15 14:57:10

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#84 2009-01-15 15:18:55

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#85 2009-01-15 21:38:33

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#86 2009-01-15 21:59:30

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#87 2009-01-16 07:21:25

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#88 2009-01-16 09:55:13

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#89 2009-01-16 10:44:03

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#90 2009-01-16 20:54:10

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#91 2009-01-18 11:52:22

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#92 2009-01-18 20:13:29

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#93 2009-01-18 21:17:12

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#94 2009-01-18 21:38:38

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#95 2009-01-18 22:34:48

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#96 2009-01-21 15:17:01

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#97 2009-01-23 13:08:11

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#98 2009-01-29 06:13:50

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#99 2009-02-04 02:37:41

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

#100 2009-02-04 06:08:54

Re: Intel i965 on 2.6.28 kernel --- results with xf86-video-intel 2.5.99.

Board footer