Suggestion: Compile with flags that optimise for the processor in the PI3/PI4.

[g8kig]

I have tested and it seems faster to me without any new problems.

gcc -Wall -Wextra -O2 -march=armv8-a -mfloat-abi=hard -mfpu=neon-fp-armv8 -g -o

Of course the -Wall -Wextra might be a bit much but I have updated my fork to clean up all the errors.