Philippe, ----- Ursprüngliche Mail ----- > Von: "Philippe Gerum" <r...@xenomai.org> >> I did some more experiments. The system was already autotuned. >> But raising the base interval from 1000 (default) to 5000 made the issue >> go away. >> > > Does the issue persist at any freq when resetting the clock gravity for > user-space timers to zero instead of (or after) autotuning? e.g. > > $ echo "0u" > /proc/xenomai/clock/coreclck
Actually with 0u I get the best results. Sometimes it takes up to a few minutes until I get negative values but with values computed autotuned the timings turn negative almost immediately. $ echo 6333u > /proc/xenomai/clock/coreclk # computed by autotune $ ./demo/posix/cyclictest/cyclictest -l10000 -m -Sp99 -i100 -n --nsecs # /dev/cpu_dma_latency set to 0us policy: fifo: loadavg: 0.00 0.00 0.08 1/81 5159 T: 0 ( 5158) P:99 I:100 C: 10000 Min: 0 Act: 2999 Avg:2147483647 Max: -1 T: 1 ( 5159) P:99 I:600 C: 1680 Min: 333 Act: 999 Avg: 1097 Max: 8000 $ echo 0u > /proc/xenomai/clock/coreclk $ ./demo/posix/cyclictest/cyclictest -l10000 -m -Sp99 -i100 -n --nsecs # /dev/cpu_dma_latency set to 0us policy: fifo: loadavg: 0.00 0.00 0.08 1/81 5176 T: 0 ( 5175) P:99 I:100 C: 10000 Min: 6000 Act: 6667 Avg: 9236 Max: 20000 T: 1 ( 5176) P:99 I:600 C: 1667 Min: 6000 Act: 14000 Avg: 7444 Max: 15333 Please note that -i100 and -n --nsecs makes the issue appear much faster than without. Thanks, //richard