It's not becuase the CPU supports 64-bit addressing that they're faster clock-for-clock than, say, a Pentium 4. If you're running a 32-bit OS and apps (you are) then 64-bit addressing does nothing for performance or capability.
It's due to the fact that the Pentium 4 has very long pipelines to allow such high clock speeds, while it takes longer to fill that many pipelines, you lose IPC (
Instructions executed
Per
Clock). Throw in the on-die memory controller (64-bit on socket 754, 128-bit on socket 939) and stellar raw FPU performance of AMD's silicon and it's easy to compensate for the difference in clock speed. AMD's made a positive model with their performance rating based on performance rather than high clock speed. These days, when buying a CPU, never look at the fact that you get a 2.2GHz Athlon 64 CPU for the price of a ~3.xGHz Pentium 4. Look at the performance rating, it's pretty accurate with A64.
Not to say I don't love my P4. It's done everything I've thrown at it with absolute ease, and I wouldn't trade HT for the world (well, maybe a true dual-core CPU
).
