#TIL (and wish I hadn't) that in #x86-land there are at least 11 different documented #NOP instructions, all different lengths, so you can use them as padding to align loops, that NOP in 64 bit mode behaves differently to NOP in 32 bit mode, and that there's an instruction for a floating point NOP. #Linux has a subroutine #optimize_nops. Can we please go back to abacuses and cuneiform.