Bug#461649: Follow-Up with pictures of black flickering bands

[Alex Deucher]

On Wed, Mar 19, 2008 at 5:10 PM, Stephen Benoit (Linux)
<linux at magnumline.com> wrote:
> Hi, Brice.
>
>  I am still using the unstable xorg with the xf86-video-ati from GIT
>  and the suggested xorg.conf changes mentioned before.
>
>  I am taking another stab at my shaking X in Etch.  To summarize:
>  Any change on screen (e.g. scrolling) causes a flicker (either a
>  thin black band (one or more rows of pixels) or shifting of a
>  row of pixels (the shaking effect).
>
>  BTW, with Option "NoAccel" "true", the shaking stops, but so does
>  my productivity ;)
>
>  This time, I am including some photos :) to help out.
>
>  The first two the upper-left corner of my typical desktop when
>  things are quiet versus when something is scrolling on the screen
>  and the display shakes.
>
>  The other three show a plain "X :0 &; sleep 10; aterm -display :0 &"
>  scenario while scrolling text in the aterm window.  Horizontal
>  black bands appear randomly over the screen.
>
>  It reminds me of a microcontroller project I worked on a few years
>  that generated real-time signals with sofware interrupts.  If the
>  CPU was busy with an uninterruptable operation, sync pulses would
>  be skipped or delayed.  I wonder if the ATI card gets busy with AGP
>  bus I/O and loses track of sending out the rasters on time.
>
>  If you can think of any options I could try, including kernel modules
>  (buggy intel_agp or radeonfb?), please let me know!

Does adjusting the DisplayPriority option help (see radeon man page)?

Some other options to try:
Option "BusType" "PCI"
Option "AGPMode" "X" where X is 1,2,4, or 8 (depending on your chipset)
Option "DRI" "False"

Alex