Bug 16374 - Screen blanks when using DVI at 1920x1200 with an X1300 (PCI ID 1002:7187)
Summary: Screen blanks when using DVI at 1920x1200 with an X1300 (PCI ID 1002:7187)
Status: RESOLVED INVALID
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
: 16472 (view as bug list)
Depends on:
Blocks:
 
Reported: 2008-06-16 00:23 UTC by Bernhard Rosenkraenzer
Modified: 2018-06-12 19:10 UTC (History)
3 users (show)

See Also:
i915 platform:
i915 features:


Attachments
Xorg.0.log (35.40 KB, patch)
2008-06-16 07:28 UTC, Bernhard Rosenkraenzer
no flags Details | Splinter Review
xorg.conf (2.56 KB, text/plain)
2008-06-16 07:28 UTC, Bernhard Rosenkraenzer
no flags Details
testing patch for pll selection (854 bytes, patch)
2009-02-10 09:01 UTC, Alex Deucher
no flags Details | Splinter Review

Description Bernhard Rosenkraenzer 2008-06-16 00:23:09 UTC
When using an X1300 (PCI ID 1002:7187) connected through DVI to an Acer X243w display, at a resolution of 1920x1200, the screen occasionally blanks for no apparent reason and comes back after around 5 seconds. Chances are the DVI signal is dropping for a while, causing the display to enter (and immediately exit) sleep mode.

It doesn't happen if I switch to a lower resolution (e.g. 1600x1200 or 1680x1050).
It happens regardless of whether or not HorizSync and VertRefresh are specified manually in xorg.conf.
Comment 1 Alex Deucher 2008-06-16 07:09:17 UTC
please attach your xorg log and config.
Comment 2 Bernhard Rosenkraenzer 2008-06-16 07:28:17 UTC
Created attachment 17143 [details] [review]
Xorg.0.log
Comment 3 Bernhard Rosenkraenzer 2008-06-16 07:28:36 UTC
Created attachment 17144 [details]
xorg.conf
Comment 4 Bernhard Rosenkraenzer 2008-06-16 08:06:50 UTC
Seems to be some timing issue the display doesn't like, if I use the modeline below instead of the standard 1920x1200, it doesn't seem to happen (I've been running it for around 10 minutes without any "blackouts" so far).

ModeLine        "1920x1200_60.00" 193.25  1920 2056 2256 2592  1200 1203 1209 1245 -hsync +vsync
Comment 5 Bernhard Rosenkraenzer 2008-06-17 23:45:35 UTC
It just did black out even with the modeline, so that modeline makes it happen less frequently, but doesn't make it go away.
Comment 6 Alex Deucher 2008-06-18 09:08:30 UTC
you might try a reduced clock rate. Do any of the following modelines help?

Modeline "1920x1200_50.00_rb"  127.75  1920 1968 2000 2080  1200 1203 1209 1229  +HSync -Vsync
Modeline "1920x1200_55.00_rb"  140.75  1920 1968 2000 2080  1200 1203 1209 1232  +HSync -Vsync
Modeline "1920x1200_58.00_rb"  148.50  1920 1968 2000 2080  1200 1203 1209 1233  +HSync -Vsync
Modeline "1920x1200_59.00_rb"  151.25  1920 1968 2000 2080  1200 1203 1209 1234  +HSync -Vsync
Comment 7 Stephane Glondu 2008-06-23 08:08:27 UTC
*** Bug 16472 has been marked as a duplicate of this bug. ***
Comment 8 Stephane Glondu 2008-06-23 08:09:57 UTC
It seems that I have the same problem (see #16472).

The Modeline 1920x1200_59.00_rb seems to fix the problem here...
Comment 9 Bernhard Rosenkraenzer 2008-06-23 08:10:57 UTC
None of the modelines work here, some show the problem more frequently than others, but they all have it.
Comment 10 Alex Deucher 2008-06-23 12:56:53 UTC
I've just committed some adjustments to the pll selection function.  Can you test again with ati from git master and see if it works any better?
Comment 11 Bernhard Rosenkraenzer 2008-06-23 23:56:33 UTC
Looks good -- 15 minutes without blanking so far.
Comment 12 Bernhard Rosenkraenzer 2008-06-24 02:21:14 UTC
It's much much better than before, but still not perfect - I just had it blacking out once (after a couple of hours of use). Before the patch, it would black out about once in 10 minutes.
Comment 13 Michael Stein 2008-11-14 02:13:47 UTC
Same bug for me.
Running ubuntu 8.10 with version 1:6.9.0+git20081003.f9826a56-0ubuntu2 (Don't ask me about details for that version number :) )
This modeline is autodetected
modeline  "1920x1200@60"       154.0   1920 1968 2000 2080  1200 1203 12 09 1235 +hsync -vsync

How can I force that modelines?
Comment 14 Alex Deucher 2008-11-14 06:48:24 UTC
(In reply to comment #13)
> How can I force that modelines?
> 

http://wiki.debian.org/XStrikeForce/HowToRandR12
see the "Forcing a preferred mode" section

You might also try turning on/off coherent mode:
xrandr --output DVI-0 --set coherent_mode 0
or
xrandr --output DVI-0 --set coherent_mode 1
Comment 15 Alex Deucher 2008-12-03 01:51:24 UTC
you might also try:
Option "DisplayPriority" "HIGH"
Comment 16 nils 2009-02-08 08:08:04 UTC
I've read many bug reports about radeon cards, dvi and flickering/blank screen.

Is there any chance to solve this problem?
I've tried:
- every options in "man radeon"
- different modelines
- debian lenny radeon driver and the git master from today

After some hours of testing:
I discover that these two lines in xorg.conf reduces the frequency of blanking:

Modeline "1920x1200R" 152.00 1920 1968 2000 2080  1200 1203 1209 1240 -hsync +vsync
Option "PreferredMode" "1920x1200R"
   
The idea is changing the polarity and reducing the frequency from 60 to 58.

Everything is working fine with the VGA-port or in windows with VGA or DVI-port. 

I use a radeon 9200 SE and a Samsung 2433 with a resolution of 1920x1200 pixel. I don't see any conspicuous line in /var/log/Xorg.0.log.

Which further information are needed to fix this bug? 
Comment 17 Michael Stein 2009-02-09 00:40:30 UTC
I solved the problem. But it is long time ago and I forgot the commands. (Than my brother gave me a nvidia-card - this was the final resolution...)

So ati:
The Problem was that the radeon-driver detects the maximum refresh-rate of the display which something like 150Mhz. The graphiccard can't run this frequency at the resolution 1920x1200.
So, I installed the ati-driver. He knows more about the card-specs and I figured out the working refresh-rate with xrandr.

I finished with:
Section "Monitor"
        Identifier      "Configured Monitor"
        Vendorname      "Generic LCD Display"
        Modelname       "LCD Panel 1920x1200"
        Horizsync       30-83
        Vertrefresh     56.0 - 76.0
        modeline  "1920x1200@60" 138.2 1920 1968 2000 2080  1200 1203 1209 1235 +hsync -vsync
EndSection

Section "Device"
        Identifier      "Configured Video Device"
        Boardname       "ATI Radeon (vesa)"
        Vendorname      "ATI"
        Driver  "radeon"
        Option "UseModeline" "1"
        Option     "IgnoreEDID" "on"
        Option "PanelSize" "1920x1200"
        Option "PreferredMode" "1920x1200@60"
        Option "DDCMode" "on"
        Option "DynamicClocks" "on"
EndSection
Comment 18 nils 2009-02-09 11:07:47 UTC
Okay, I should buy a new graphic card ;-)

But - Why does the card work with windows? It can't be a hardware limitation.
Comment 19 Alex Deucher 2009-02-09 11:25:36 UTC
(In reply to comment #18)
> Okay, I should buy a new graphic card ;-)
> 
> But - Why does the card work with windows? It can't be a hardware limitation.
> 

The monitor doesn't like something about the timing.  It could be a bad combination of pll dividers or something with the crtc timing parameters.  I'm sure the windows driver has tons of quirks that we don't have yet.  Try slightly modified modelines until you find one that works and we can try and then we can try and come up with a quirk to deal with it properly.
Comment 20 Michael Stein 2009-02-09 23:50:07 UTC
Try out the fglrx-driver and see how the tune the card. (That was my way)
Than tune yours.
I think the it is running under windows, because they use that driver. It know more about the cards - maybe more undocumented ...
Comment 21 Alex Deucher 2009-02-10 09:01:34 UTC
Created attachment 22758 [details] [review]
testing patch for pll selection

Try this patch.  This should help narrow down if it's a pll problem.  Uncomment a different pll_flags setting for each try and see which if any help with the problematic monitor.
Comment 22 nils 2009-02-10 10:00:33 UTC
I test the patch and no line solved the problem. I see the worst result with:
pll_flags |= RADEON_PLL_PREFER_LOW_FB_DIV;

In a second run I vary the modelines, but only the values in X,Y and polarity.

Modeline "NAME" X 1920 1968 2000 2080  1200 1203 1209 Y -hsync +vsync

I could reduce the blanking, but not solve the problem. 

Does it makes sense to vary the other values?
Comment 23 Alex Deucher 2009-02-10 10:28:02 UTC
This is the mode from the edid on my 1920x1200 monitor.  It works fine with my x1300.

Modeline "my1920x1200"  154.00  1920 1968 2000 2080  1200 1203 1209 1235 +hsync -vsync
Comment 24 Alex Deucher 2009-10-21 13:49:13 UTC
can you try with ati from git master?  Specifically commit 66b194a78c470cb3978f310828dd96c3f3e96944.
Comment 25 Alex Deucher 2009-12-09 10:06:44 UTC
Can you try with the latest code from xf86-video-ati git master?  I just committed some new PLL code that might help.
Comment 26 nils 2009-12-09 10:53:55 UTC
I can't do anything in the moment, because the machine is broken. 
Comment 27 Alex Deucher 2010-10-19 16:32:10 UTC
Any updates?
Comment 28 Adam Jackson 2018-06-12 19:10:44 UTC
Mass closure: This bug has been untouched for more than six years, and is not
obviously still valid. Please reopen this bug or file a new report if you continue to experience issues with current releases.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.