The uninformed complaints about the use of a 64-bit ISA in ARMv8 have been rather irritating, given we didn't know how it would play out in practice. It seems that the performance increase of the 5S is pretty spectacular; the iPhone 5 is already FAST.
I think a lot of people were (a) fuelled by a desire to complain about Apple as much as possible and (b) were assuming that ARMv8 was an AMD64-type transition or even a SPARC64-type transition, rather than a totally new ISA.
That said, from Anandtech's article, it looks like most of the speed boost is from microarchitectural improvements, rather than use of the new ISA; ARMv7 code shows most of the speedup.
