The manual documents movsd in SSE2 as 'transfers a double precision value between source and destination operand (only the low quad word is trasferred). At least one of the operands have to be a SSE register, the second one can be also a SSE register or 64-bit memory location.' However I thought movsd was a zero parameter op (usually prefixed by rep) which moved [esi] to [edi].

Is this a typo or are you saying if it has (two) parameters, one of them must be an SSE register?

Same question for cmpsd

Some of the mnemonics for opcodes overlape. When this happens it depends on the context (ie the source and destination).

When there's movsd xmm0,xmm1 it's obviously not the mov string dword opcode but the mov scalar double opcode.

Thanks,

What I was actually doing was writing a syntax file for my editor, and, since my m/c does not support SSE2, colour SSE2 instructions different. movsd/cmpsd were the only cases I found. On reflection I guess there is no harm overloading the occasional op like this as any block of SSE2 code will no doubt contain other SSE2 ops. Not that I've actually seen any code that uses these yet

Pete