There is no such "IPC" value. Idiots say "Intel lowered the IPC so they could increase the MHz" when Intel simply made their CPU differently so that it runs existing code (AKA, INSTRUCTIONS) less efficiently (AKA, less PER) yet scales to much higher speeds (AKA, CLOCK cycles). If you code specifically for the P4 using SSE2 and now Hyperthreading, and you make sure your compiler ignores old Penmtium Pro-Pentium III optimizations, you'll get much better performance. It doesn't make up for the deep pipeline, but you get the idea. AMD obviously chooses to continue optimizing their chips for last generation's code. Both are nice strategies.