Re: Curious about EPICS IOC on RTOS for x86

[Matt Rippa via Tech-talk <tech-talk at aps.anl.gov>]

Thanks John and Han. Very useful collection of information. I'll be parsing and testing these configurations in the coming days.

-Matt

On Tue, Oct 13, 2020 at 3:19 PM Jeong Han Lee <citadel.lee at gmail.com> wrote:

Hi Matt,

  There is one example, which I know, based on PREEMPT_RT Linux.

  That is the quite limited case (Debian 9, CentOS7, and limited
hardwares, ecmc under epics-modules repo), but we (Anders and I)
figured out the best optimal configuration according to our
limitations at the moment.

  https://github.com/icshwi/realtime-config

  The repository is quite dirty, since this was my last minute work
before my leave, but you can see a few interesting test results which
we had in MCAG010 and MTCA directories.

  HTH,
  Han

On Tue, Oct 13, 2020 at 4:07 PM Matt Rippa via Tech-talk
<tech-talk at aps.anl.gov> wrote:
>
> Hello,
>
> Often we start a new epics test or experiment with a Linux target and no real time requirements.
> Now I have a (relatively pedestrian) DAQ system with a sample rate in the less than 4KHz range.
> I have some questions about real-time choices on x86:
>
> Are some folks using an RTEMS-pc386 target with EPICS ?
> Are some using Linux with PREEMPT_RT patches and EPICS?
> Any other variant on x86 compatible with EPICS?
> Any comments on performance preempting kernel threads?
> Has anyone profiled their context switch and interrupt latencies for their system? How did you do this?
>
>
> Thanks for sharing your experiences!
> -Matt