RE: How to include math CMSIS DSP library in NRF9160?

Sigurd — Mon, 02 Sep 2019 09:30:18 GMT

Hi,

[quote userid="72754" url="~/f/nordic-q-a/51656/how-to-include-math-cmsis-dsp-library-in-nrf9160"]1-st question - how can i set preprocessor defines using SES and NRF connect options?[/quote]

You can set it in the CMakeLists.txt with add_definitions()

[quote userid="72754" url="~/f/nordic-q-a/51656/how-to-include-math-cmsis-dsp-library-in-nrf9160"]ARM_MATH_CM3 define in arm_math.h file. Include lines with sources in CMakeLuists.txt[/quote]

The nRF9160 uses a ARM Cortex-M33 CPU. The CMSIS DSP Library is only for Cortex-M7/Cortex-M4/Cortex-M3/Cortex-M0.

[quote userid="72754" url="~/f/nordic-q-a/51656/how-to-include-math-cmsis-dsp-library-in-nrf9160"]How should i correctly include and use math functions in NRF9160?[/quote]

In order to use math.h , set CONFIG_NEWLIB_LIBC=y in your prj.conf

EDIT 26.05.2020:

Starting from nRF Connect SDK v1.3.0, there is support for CMSIS Version v5. A arm_math.h library with support for ARM Cortex-M33 / Armv8-M is included there.

Set these config's in you project to use it:

CONFIG_CMSIS_DSP=y
CONFIG_NEWLIB_LIBC=y
CONFIG_FPU=y
//Set CONFIG_NEWLIB_LIBC_FLOAT_PRINTF if  printf should be able to print float
CONFIG_NEWLIB_LIBC_FLOAT_PRINTF=y

And use

#include <arm_math.h>

in the application code to start using CMSIS DSP functions.