Welcome to our new forum
All users of the legacy CODESYS Forums, please create a new account at account.codesys.com. But make sure to use the same E-Mail address as in the old Forum. Then your posts will be matched. Close

multiple questions regarding program speed

Anonymous
2020-01-17
2020-02-17
  • Anonymous - 2020-01-17

    Originally created by: phenixrb

    Hello everyone,

    As noted, originally CODESYS on raspberry is not real time,
    in order to make programs on raspberry faster,
    what is the best option?

    I saw that it is possible to do it by adding "rt-preempt" in rasbian, but with a decrease of the CPU.
    "rt-preempt" supported by OSADL,
    and CODESYS contributes to the OSADL group.
    which for me, assumes that the group is serious.

    but is it necessary or appropriate to add "preempt-rt" to be in real time?
    in my case I am not a raspbian _ ....
    how not to make a mistake?

    or then, does the software "Control for Raspberry Pi MC SL"
    with a good definition:
    attributions on the core.
    priority of tasks.
    type of task
    cycle time.
    enough to approach "real time"?

    and what is a good task configurator setting?
    group POUs, visualization, ....... by type, size ......

    for me there is so much possibility that i don't see how to optimize this.

    if you have ideas or tutorials, please let me know.

    IMG: MAINTASK.jpg

     
  • sgronchi - 2020-02-17

    First, realtime means "with predictable latency". Usually realtime is at odds with throughput, i.e. realtime mode actually maximum throughput (except if the CPU-intensive task has non-realtime priority and Codesys has, but you can work around it playing with niceness if you want). A definition that is not totally right but gets the idea is that it reduces the deviation from the average throughput.
    PREEMPT_RT is a kernel patch, you could follow the tutorial at https://lemariva.com/blog/2018/07/raspb ... nel-4-14-y to compile it (adjust the kernel version).

     

Log in to post a comment.