Building and running the model on Ethos-U55

Here we are. Just this recipe keeps us from completing this book. All the tools are installed, and the TFLite model is converted to C code, so where does that leave us? We still need to build an application to recognize images with the CIFAR-10 model. Once the application is ready, we need to compile it and run it on Corstone-300 FVP.

Although it seems there is still a lot to do, in this recipe, we will modify a prebuilt sample for the Ethos-U microNPU to simplify all the remaining technicalities.

In this recipe, we will show you how to modify the Ethos-U example available in TVM to run the CIFAR-10 inference. The application will then be compiled with the Makefile and Linker scripts provided in the prebuilt sample and finally executed on Corstone-300 FVP.

The following Bash script contains the commands referred to in this recipe:

• build_and_run.sh:

https://github.com/PacktPublishing/TinyML-Cookbook/blob/main/Chapter08...