GNU compiler flags for generating FPU assembly instructions for nrf5340 app core

[kpreiss]

Hi,

We are building an application on the nrf5340 application core.  We noticed that library calls are being made to perform double precision floating point arithmetic which are NOT utilizing the FPU.  What libraries should we be linking with to take advantage of the FPU?

Thank you.

[Jørgen Holmefjord]

Hi,

Can you try to enable the following Kconfigs in your prj.conf file?

• CONFIG_FPU
• CONFIG_FPU_SHARING

Best regards,
Jørgen

[kpreiss]

Thanks Jorgen,

We are using CMake and the GNU compiler but not the Zephyr build system.  Could you please confirm the actual arguments passed into the GCC compiler and what libraries to link with?

Best,

Kurt

[Jørgen Holmefjord]

Hi,

See this file.

For Cortex M33 with single-precision FPU, the GCC flags should be:

-mfpu=fpv5-sp-d16
-mfloat-abi=hard

Best regards,
Jørgen