"That’s nonsense. There are serious performance gains by going 64-bit. Addressing more than 4 GB of memory is not the only advantage."
The performance gains are only there assuming that the operations at 64-bit require the same number (or less) of clocks as the same operations would in 32-bit and also that the app uses data in such a way that it is operating on data that is naturally 64-bit wide.
In actual real-world use jumping up to 64-bit and keeping everything else the same doesn't really buy you much of anything for most apps. In the move to x64, everything else didn't stay the same, the number of registers was increased, a newer version of SSE was introduced, etc. So x86->x64 did see a pretty good speed-up.
It looks like some of this is also true with the A7 (more registers, etc), so there will probably be a similar speed advantage, but the core statement as made is slightly off base, IMO.
In any case, no code running on iOS will take advantage of these new 64-bit features until developers are able to create 64-bit builds of their software, though I don't fault Apple for releasing the hardware first since either way there's a chicken and egg issue and the sooner phones start moving to 64-bit the better, even if the on-board memory they have isn't quite near the 4GB level.
