Underneath the x86/x64 ISA is a really different architecture. So to reach the backwards compatibility it needs a x86/x64 decoder and that is not a significant

Underneath the x86/x64 ISA is a really different architecture. So to reach the backwards compatibility it needs a x86/x64 decoder and that is not a significant portion of the silicon as far as I know. Those instructions might be somewhat inefficient, but this is not really an issue when comparing, any ARM chip emulating will have to tackle the same problem. If Apples chip does not need this backwards compatibility, they essentially have to do it in software when emulating. They do this with a combination of recompilation at install time and some emulation at runtime. I suspect the compilation step is one of the main contributors to this boost in speed and efficiency. Windows could do the same, optimize the executable to better fit the underlying chip.