X86 and x86-64 are pretty much the same, so any byte-code optimisation will be the same for either. X86 extensions, like SSE, are also another major area of optimisation, with modern cpus obviously having more extensions.
This isn't quite true; x86_64 has more general registers, which are helpful in writing VMs. At one time, Apple JavascriptCore was 60% faster on x86_64 Safari than on x86 Safari on the same machine, though I think the gap has reduced.
