Asrock Q1900M: Unexpected Realtime Delay, but Good Latency Values

That makes sense.

Do you have any idea on how i could maybe debug what its stalling at?

Well i figured this out, pretty strange but it turns out that there is a watchdog messege on dmseg that occures exactly every 1 minute and i had dmesg ouput on the serial port. Well apearantly this superio chip stalls when the serial port is used as anytime anything is written to that serial port there is a latency spike. Simply removeing the serial port console= line from the kernels cmdline fixed the issue.

So it would seam that on this machine writeing to the serial port blocks the parallel port for a bit, thats not too bad i can just use netconsole to monitor the system

I had an asrock Q1900B-ITX and it liked to have two identical memory modules. May also help with latency.

Test results (copied from here: forum.linuxcnc.org/18-computer/29224-asrock-q1900b-itx#128103 ):
Quick ~30-60min test with 6x glxgears and a hd video, only memory module(s) were changed between tests.
servo / base latency, ram
11035 / 11234 2x 8G Hynix HMT41GS6BFR8A-PB 1600 CL11
9380 / 11482 2x 8G Corsair CMSO8GX3M2A1333C9 1333 Mhz CL9
10747 / 12206 8G Micron MT16KTF1G64HZ-1G6E1 + 8G Samsung M471B1G73QH0-YK0
9886 / 17090 4G Micron MTBKTF51264HZ-1G6E1 + 4G Samsung M471B5173QH0-YK0
11618 / 17726 1x 8G Samsung M471B1G73QH0-YK0
11898 / 20126 1x 2G Ramaxel RMT3190ME76F8F-1600
11693 / 22918 1x 4G Micron MTBKTF51264HZ-1G6E1
10440 / 23242 1x 8G Hynix HMT41GS6BFR8A-PB 1600 CL11
11627 / 24188 1x 8G Micron MT16KTF1G64HZ-1G6E1
11257 / 28173 1x 4G Samsung M471B5173QH0-YK0 PC3L-12800S-11-12-B4
10107 / 28223 1x 8G Corsair CMSO8GX3M2A1333C9 1333 Mhz CL9