Other Parts Discussed in Thread: SYSBIOS
Hello,
I'm just testing a little bit SYS/BIOS, IPC and Multicore-Analyzer to get a feeling for the performance of the 8 Cores. I tried a simple example where a timer is triggered every 10 us. This example works fine.
Now setting up some events for every core I just try to trigger one event on CORE1 everytime the timer is triggered (simple with Notify_sendEvent). Setting the period to 50 us, everything is okay. But setting the timer period below 50 us (example given 20 us), CORE0 will not end after a dedicated counter of triggers as before -> CORE0 is hang up in NotifyDriverShm.c on line 415 (coming from CacheProxy_inv in dsp_pe66.c line 28568).
With time measurements (using TSCL etc.), Notify_sendEvent only lasts around maximal 1 us, so I don't expect that the timer will overrun itsself. So I don't know what's the reason for this behaviour and how I can avoid it as it is necessary for our workflow to use events with all cores within a period of around 5 us.
Maybe there exists another way to notify the cores for processing in such a small time period?
Best Regards,
Bernd