[Linux-aus] strange ATI video issue

[James Cameron]

That's surprisingly similar to my week, but the details don't match up
entirely.

At a telescope we had a performance problem with a four headed
workstation, which uses XDMCP to host an X11 display from an
instrument computer, then uses x11vnc to provide that to multiple
remote observers.  It was limited CPU vs GPU bandwidth.  Once we had
enough display updates happening, the TCP queues for the X server
would begin to grow, and interactive latency for the VNC users would
climb from 250ms to about 10 seconds.  I reproduced it with many
instances of `xclock -update 1`.

What looked like your scenario for me was that an strace of Xorg
showed bursts of -ERESTARTSYS in response to a
DRM_IOCTL_NOUVEAU_GEM_PUSHBUF as the problem grew, then ioctl's
began pausing entirely.

So, entirely different graphics driver, just a similar behaviour
suggesting difficulty handling the rate of buffer pushes.

Try using strace (over SSH) to check for what code path is giving
grief.

In the end, I ran out of time and was urged to switch to a
proprietary driver.  I was able to move from 96 xclocks with some lag,
to 192 xclocks with no visible lag, filling all four displays.

Kernel 5.4.0 plus Ubuntu 20.04 patches.