After enabling FPU in Kconfig, the computation speed remains unchanged. Why?

[crm]

HI,

I’m using nRF Connect SDK (NCS) v2.8.0 with the blinky example.nrf54l15dk, In the project, I configured the following in Kconfig:

kconfig

CONFIG_FPU=y  
CONFIG_FPU_SHARING=y  

In the main function, I performed floating-point calculations with the following code and measured the execution time (~2.7 ms) by toggling a GPIO pin.

However, when I disabled the FPU by setting:

kconfig

CONFIG_FPU=n  
CONFIG_FPU_SHARING=n  

and recompiled, the measured execution time remained unchanged (still ~2.7 ms).

Question:
Why is there no performance difference? Are there additional configurations required to properly enable/disable the FPU?

Thanks!

[Turbo J]

Check the resulting assembly code - the optimizer will completely eliminate the floating point calculation in the loop. It could actually remove for loop entirely.

[crm]

I set the optimization level to 'Optimize for Debugging.' Does this need to be changed to the default setting?"