Taken from a website :-) Most consumers set out to buy a new PC and mistakenly assume that Intel’s Pentium 4 is faster than the Athlon because its clock speed is so much higher. In fact, the Pentium 4 needs those higher clock speeds just to offer comparable performance to an Athlon as a result of the P4’s extremely long data pipeline. As a CPU processes data, it chops them into packets and sends them through a type of assembly line, called a pipeline. The more steps in the assembly line, the faster each step, or stage, can be completed because less work needs to be performed in each stage. In a longer pipeline, executed data often has to be thrown out and reprocessed, and it takes more clock cycles to refill with data when that happens. Data has to be purged from a pipeline because modern CPUs are many times faster than the system memory that sends the data to them, so processors just guess which pieces of data will be sent next. This process of guessing data is a complicated practice called branch prediction. The Pentium 4’s 20-stage pipeline allows the chip to hit very high frequencies, but it also results in a higher performance penalty when a missed prediction occurs. The Athlon 64 features a 12-stage pipeline, and so suffers less of a performance hit when it predicts incorrectly, because there isn’t as much data in the pipeline to recompile. In addition, a shorter pipeline can often execute more instructions per clock cycle, since data doesn’t have to pass through so many stages before being fully executed. This is the primary reason why a 2.2-GHz Athlon 64 FX-51 can offer superior performance to a 3.2-GHz Pentium 4. The Pentium 4 streams data along at up to 3.2 GHz, but each piece of data has to pass through 20 steps before it’s completely executed. The Athlon 64 has to pass data through only 12 steps. Branch prediction is just one tool a CPU uses to overcome the slow speed of the system memory to offer higher real-world performance. The large L1 and L2 caches that the Athlon 64 and Pentium 4 use to speed the flow of data from system RAM also complement the branch prediction. However, without knowing anything about pipeline lengths, branch prediction algorithms, and cache sizes, consumers continue to buy PCs based on the clock speed of their CPUs. To fight that, AMD has adopted a model numbering system that roughly, and indirectly, equates the performance of an Athlon with a Pentium 4 at a particular clock speed (i.e. the Athlon 64 3200+ and the 3.2-GHz Pentium 4), but this doesn’t seem to be gaining much traction with PC buyers. As a result, AMD may be forced to engineer its future processors with roughly the same number of pipeline stages as Intel’s chips. Although shorter pipelines may be more efficient and actually yield higher performance, that’s not of much help to AMD if no one recognizes the benefit. With more stages in its pipeline, there’s no reason that the next version of the Athlon shouldn’t match or exceed the clock speed of the Pentium 4 at the high end. Adding stages isn’t all about marketing, though. With branch prediction rates now more than 90 percent accurate for both the Pentium 4 and the Athlon 64, the drawback to using a longer pipeline is decreasing. As missed prediction rates continue to drop, the performance hit to a long pipeline will drop as well

Old song. http://www.overclockers.com/articles942/index.asp

Still interesting though.

See...now ive been looking for something to throw back at my friend who thinks hes all that AND a bag of ships cuz of his new p4 3.0 Old song but its new to me,... so thanks

See it's the Athlon who are the fastest and Intel are just trying to keep up :-) Yup, it's a very old story but still a good one

Sonny, It just doesn't work that way... If you want quality you do a P4 If you want value you do the AMD. Wanna wind me up? Skip

If you want quality you do a P4 ? Seriously i think don't know know what you're talking about. The Athlon64 cpu is by far better than any Intel EE cpu, and i don't think that quality is an issue here.