ML-DSA on a Cortex-M33: From 140 ms to 27 ms, One Table at a Time
TL;DR I needed post-quantum signatures (ML-DSA, FIPS 204) on an STM32U5A5, a Cortex-M33 at 160 MHz. I started from mldsa-native, a portable, formally verified C implementation, in its low-stack co...