ge_mask.c

std::uint32_t duplicate(std::uint32_t o) {
  return ((o & 0x80808080) >> 7) * 0xFF;
}

# compiler suggestion:
duplicate(unsigned int):
        shr     edi, 7
        and     edi, 16843009
        mov     eax, edi
        shl     eax, 8
        sub     eax, edi
        ret

# hand optimized:
duplicate(unsigned int):
        and     edi, 0x80808080
        mov     eax, edi
        shr     edi, 7
        sub     eax, edi
        ret