AArch64 Increment is quite inefficient
Today I stumbled across the following assembler from the AArch64 NCG (apparently implementing a ticky counter increment):
0x00000000015fa85c <+20>: mov w15, #0x1 // #1
0x00000000015fa860 <+24>: add x17, x17, x15
As far as I can tell, this could be instead:
add w15, w15, 1