alexr wrote:But it didnt help with the RTAPI errors and i was working on the assumption of less then 30,000 is "functional". I have never had the other figures in the millions before... even on the roll my own kernels.
Maybe I am reading the figures you posted wrong, but my interpretation is that you asked for a 1,000,000nS servo thread, and the worst-case jitter on that is 5009nS and the last servo period was 998450, ie 1550nS shorter than requested.
I don't think that 998450 is a jitter number.