Inconsistent FFT Results Across Different Optimization Levels on ESP32 Platform

Hey guys,

I use the Microfft library on the ESP32 (xtensa) platform, but get different values depending on the compiler optimization level. Have already addressed the problem within the 'esp-idf-hal' project. Maybe there are some ideas here too. The problem also affects the RealFFT/RustFFT library.

References:

• https://github.com/esp-rs/esp-idf-hal/issues/262
• https://github.com/ramtej/esp32-compiler-marvels-std-rs

Thank you, Jiri