I\'m implementing binary translation and have to deal with sequences of NOPs (0x90) with length about 16 opcodes. Is it better for performance to place JMP (to the end) at s
If the NOPs are to align the stream, then they have more value than just being a NO OP. if your concerned with pure speed, see Agner Fog's Optimization Manuals Vol. 4.