The architecture of the chips, and the length of the pipeline (and the amount of cache) means that the newer chips can do more operations per clock cycle than their previous brethren - meaning a 2.2GHz 64bit chip from AMD or Intel will be "faster" than, say, a 3.2GHz P4 or a similarly clocked AMD K7 chip. That is correct, a dual-core chip has 2 actual CPU dies on one PCB, and both cores share the same L2 cache. Note that two of the major differences between 64bit CPUs and 32bit CPUs is the fact that 64bit CPUs have double the registers as a 32bit chip (they can only be used by 64bit software, however), making them much faster and capable of doing more when running a 64bit OS and software. Also, a 64bit OS running on a 64bit CPU can access far more memory than a 32bit OS running on a 32bit or 64bit CPU - 32bit CPUs can access at most 32GB of RAM (64GB with some expensive hardware trickery) and are limited to 2 or 3GB of virtual address space for applications (unless they support trickery, like AWE/PAE) and 2GB or 1GB for kernel address space, as well as having additional kernel pool memory constraints in Windows (256MB of kernel nonpaged pool, ~530MB kernel paged pool for 32bit, 48MB maximum session view size). 64bit CPUs can access up to 16 EXAbytes of RAM (theoretically - Windows limits you to a max of 8TB right now), as well as having 128GB of nonpaged and paged pool available, and 8TB of virtual address space for both usermode and kernel applications, natively. There's more to it than that, but those are the major points most people would be concerned about.