Does Cubemx-AI runtime uses Double precision or Single precision FPU on Cortex-M7 ?

[RKopp.12]

..

[Gln]

Hi @RKopp.12​ ,

Currently, the network runtime library only supports 32-bit single precision floating points.

More details can be found in the X-CUBE-AI User Manual UM2526:

https://www.st.com/resource/en/user_manual/dm00570145.pdf

Guillaume

[RKopp.12]

Thank you.