Megahertz sells.....and so do those "Intel Inside" logo's; however, lower prices sell too!
In marketing there is a concept known as Good, Better, Best. Given the current state of affairs, AMD is squarely in the "Better" catagory. Given this scenario with pricing scaling linearly between the groups, consumers most often choose "Better" and save a little money. Some go the bargin side with "Good" while a few lean toward "Best".
Reality is that AMD's "Better" solution at 1.2Ghz is priced below Intels "Good" solution (PIII 1Ghz). This is a premo position for AMD to be in going into the holliday season.
The Celeron and Duron systems would have to be placed in a different catagory all together called "Value". Yes, cheep, but get the job done, line of thinking.
The bad thing for Intel is that right now, they can produce tons of PIII's, but only a relatively small amount of P4's. The introduction of P4 will only push the price that Intel can charge for PIII down and hurt their bottom line.
Faced with this delema, it is rumored that Intel will release a 1.3Ghz P4 to compete with AMD in the "Better" catagory.
Intel's strategy is full of holes and they know it. There is no marketing for P4, and a whole new flurry of PIII advertising. This further outlines the weakness in Intel's marketing plan. Remember the shortages last year of high end Intel parts? Imagine the same again next year.
Intel will be attempting another roll-over in their plants to get to P4 production. On the .18um process, every P4 made will take up the production capacity equivelent to 3 PIII's thus lowering Intel's production capability. In the middle of next year, Intel will attempt to roll out their .13um copper process in order to increase their production capacity back to where it was and raise speed ......more delays.
AMD on the other hand has .18um copper down cold. I speculate that their new core is simply a re-design of the layout of the previous core to take advantage of the copper interconnets along with a better branch prediction unit and a better cache pathway (I doubt the rumors about adding to the pipline since this would really be a tear up of the core where the other mods are auxillary functions).
According to some sources, this will result in a core that can scale to 2Ghz on .18um copper!
And
Here is the link.