no floating-point libraries

Eliminate floating-point overhead by enforcing fixed-point arithmetic and poisoning float types in embedded toolchains.

Avoiding floating-point libraries is a critical optimization for resource-constrained hardware like the ARM Cortex-M0+ or MSP430. Many low-power microcontrollers lack a dedicated Floating Point Unit (FPU), forcing the compiler to link bulky software emulation libraries that inflate code size and spike execution time. By adopting fixed-point arithmetic (using Q-format notation) and enforcing strict compiler pragmas like #pragma GCC poison float double, developers ensure deterministic performance and bit-exact results. This approach is standard in high-integrity fields like automotive safety and financial auditing, where the rounding errors of IEEE 754 standards are unacceptable and every byte of Flash memory counts.

https://mcuoneclipse.com/2024/03/13/how-to-make-sure-no-floating-point-code-is-used/