xor eax, eax

[wossname]

Why do I keep seeing xor eax, eax everywhere?

I can only imagine that it is a quick way of setting the eax register to 0, but is it faster than mov eax, 0?

Why is this not documented anywhere, and is there a list of other simple optimisation tricks?

Can someone shed any light on this?

[Ghirai]

It's also smaller.

And XOR is fully documented, but some don't see other uses for it :toothy

[wossname]

Ahh ok.  (I meant that this trick wasn't documented rather than the opcode itself ;)).

[liquidsilver]

It's simple why it's smaller: mov eax,0 has to store the full 0 value (4 bytes) and an opcode, whereas xor eax,eax just has to store an opcode. Compile it and then you'll see the the mov instruction is much longer.

[liquidsilver]

Here's another optimization leading from this:
xor eax,eax
inc eax
instead of: mov eax,1
and this works for a while, also think dec>0xFF

[Tedd]

"xor eax,eax; inc eax" takes fewer bytes, but is generally slower than "mov eax,1"

Many of these 'tricks' will come to you as you learn more asm - xor eax,eax is so common that it will not take long before you see it.

One more for you: "test eax,eax" to check eax for zero, instead of "cmp eax,0"

[diablo2oo2]

what about

Code Select Expand

push 1
pop eax

?

[JHER VON ARBANEL]

for example xor eax,eax------->mov eax,0
   or eax,eax------------------------>cmp eax,0
  and eax,0fh--------------------->11111111;eax
                                          00001111
                                         ---------------
                                          00001111; u use it like a mask
test eax,0fh ------------------------->same as and but without eax dont modify
u will dicover more shorcuts while u are programing ...  keep working