[Ovmsdev] PSRAM issue, unicore build

Michael Balzer dexter at expeedo.de
Sat Oct 19 14:28:09 HKT 2019


Update: unicore mode is unusable for us.

Further tests showed CAN processing times get really bad (i.e. milliseconds, not microseconds). This affects especially live control and monitoring via OBD2 /
CANopen, also the RX callbacks get processed with a too high delay to be able to overwrite control frames that need fast responses (like charge current control
on the Twizy).

Also I got the same bug event as before while running in unicore mode, so either this bug isn't related to the PSRAM issue or unicore mode is not sufficient as
a workaround.

I'm going back to dual core.

Regards,
Michael


Am 18.10.19 um 21:04 schrieb Michael Balzer:
> Everyone,
>
> the PSRAM issue still isn't solved, users reported new events and I had two events myself today. So toolkit -98 also doesn't fix the issue completely for us.
>
> Following the findings of @neoniusTR (Thomas Rogg), I have enabled building the OVMS in FreeRTOS unicore mode, i.e. FreeRTOS and all tasks only running on core 0.
>
> My previous load measurements showed we're only using max. ~20% of the CPU, and my test runs with a unicore build don't exhibit any slowness or sluggishness.
>
> To use unicore mode, build with CONFIG_FREERTOS_UNICORE=Y.
>
> I've just released the unicore build to my edge branch on dexters-web.de so will be able to gather some user feedback.
>
> Regards,
> Michael
>

-- 
Michael Balzer * Helkenberger Weg 9 * D-58256 Ennepetal
Fon 02333 / 833 5735 * Handy 0176 / 206 989 26





More information about the OvmsDev mailing list