Depends. In the x86 world, apps are crippled severely by the lack of programmer-accessible registers (the registers are actually there, it's just that you can't access them directly; they're used internally by the CPU for register renaming and other fancy tricks -- at least with the PPro and subsequent CPUs). AMD's x86-64 platform, however, makes available a significant number of additional programmer-accessible registers.~loserman~ said:
So on x86, the move to 64 bits does provide major performance benefits for all applications, provided they are recompiled in 64 bit mode. Not because of an inherent speed up in 64 bits, but because of the additional registers.
PowerPC, because it already has a large number of usable registers, doesn't get that boost, so you'll see the performance hit from the greater amount of data transferred come into play. So in the context of PowerPC, the quote above is entirely accurate. In the context of x86, it isn't, but only because there are factors at play other than the size of data the processor can handle.
