RPC and Tensorflow conflict with Portenta

I am working on a AI project (echo location).
My goal is to run a Tensorflow model on the Portenta H7 with data provided by the Portenta M4. The M4 is in charge of doing the 360 ° echo scans.
My tensorflow lite model is running well on the Portenta and I wanted to add a RPC connection betwwen the 2 cores.
Unfortunatly a soon as I tried to add RPC the microcontroller looks on the RPC1.begin(), the red light flasing 3 times very fast and 4 times slower and the USB port is unrecongnized by the host. I need to push twice on the reset button to recover the USB.
In order to be sure this issue is not due to my code I reproduced the same error with 2 provided example that I tried and know for sure that the run properly. I uploaded “RPC_m1” wihout any modification indise the M4. I modified slightly the tensorflow “hello_word” : added the #include “RPC_internal.h” and at the begining of the setup() RPC1.begin(); LL_RCC_ForceCM4Boot();
I got excatly the same issue with USB lost and red LED flashing.
I guess there is a conflict access to the “Serial.print” with the TF_LITE_REPORT_ERROR.
It seems that make sense to use H7 for tensorflow predictions and M4 for getting data from sensors and use RPC for that.
Does someone have an idea about that problem ?
Thanks for you help

RPC_m4.zip (1.61 KB)

This topic was automatically closed 120 days after the last reply. New replies are no longer allowed.