Hi, I am trying to improve the latency on my Machinekit installation. As hardware I have an ASRock 90-MXB870-A0UAYZ. I have chosen this because I read about it in the linuxcnc forum.
I am using Debian 9.8, with kernel 4.9 rt machinekit@debian:~$ uname -a Linux debian 4.9.0-8-rt-amd64 #1 SMP PREEMPT RT Debian 4.9.144-3.1 (2019-02-19) x86_64 GNU/Linux The latency I have is 50 microsecond, and from what I have read it is a bit too high (it should be between 10 and 30 microsecond) so I started studying about latency in the rt kernel. I have found that exists more scheduling policies and that every process declares its own. Indeed in rt-preempt.c I can see it uses SCHED_FIFO (at least, I think it should) I started latency test and checked, but it showed me it is SCHED_OTHER (TS), with proiority 19, the same of every other normal process. machinekit@debian:~$ ps -e -o pid,cls,pri,comm -H PID CLS PRI COMMAND ... 825 TS 19 sshd 831 TS 19 sshd 832 TS 19 bash 1327 TS 19 latency-test 1344 TS 19 halrun 1404 TS 19 halcmd 1408 TS 19 pyvcp ... 1089 TS 19 uuidd 1393 TS 19 rtapi_msgd 1398 TS 19 rtapi:0 Then I started Machinekit, and found out the same machinekit@debian:~$ ps -e -o pid,cls,pri,comm -H PID CLS PRI COMMAND ... 825 TS 19 sshd 831 TS 19 sshd 832 TS 19 bash 843 TS 19 bash 844 TS 19 linuxcnc 1066 TS 19 linuxcncsvr 1127 TS 19 milltask 1128 TS 19 axis ... 1088 TS 19 rtapi_msgd 1089 TS 19 uuidd 1094 TS 19 rtapi:0 1103 TS 19 io But I think it should be FF (SCHED_FIFO), may it be the cause of the high latency? mngr -- website: http://www.machinekit.io blog: http://blog.machinekit.io github: https://github.com/machinekit --- You received this message because you are subscribed to the Google Groups "Machinekit" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. Visit this group at https://groups.google.com/group/machinekit. For more options, visit https://groups.google.com/d/optout.
