edgeimpulse / example-standalone-inferencing-zephyr

17 stars 7 forks source link

Hardfault with non-EON model #22

Closed janjongboom closed 3 years ago

janjongboom commented 3 years ago

This model hard faults on nRF52 and I'm not sure why: no_eon_int8.zip

The hard fault occurs in micro_allocator.cc in StartModelAllocation when InitGraphAndContextTensorData is invoked. I'm expecting some stack corruption or something similar, because when I start adding printf() statements around the code it starts to pass.

@VojislavM @MarkoSagadin