Intel Pentium III 550

In terms of overclocking, the 550 definitely has the ability to run at much higher frequencies than the 500 could. While most 500MHz Pentium IIIs maxed out at around 560MHz (112MHz x 5.0), the 550 can easily hit 616MHz (112MHz x 5.5) with L2 cache enabled. After reaching 616MHz the ability to run the 550 any higher begins to be dependent on the quality of and the cooling applied to (if any) the processors L2 cache. At 616MHz, the L2 cache operates at 308MHz, over 10% above its rated operating frequency. The next possible setting is 644MHz (117MHz x 5.5) which was stable with some added cooling, although your success will vary even more from chip to chip than the ability to hit 616. At 644MHz, the L2 cache runs at 322MHz, around 17% above the specified operating frequency. After 644MHz it quickly becomes very difficult to overclock the 550 with L2 cache enabled.

So far, overclocking has been mentioned under the explicit conditions that L2 cache remains enabled, however what happens when L2 cache is disabled? If you disable the L2 cache of the Pentium III processor, youre essentially testing the potential of Intels core and you end up removing the L2 cache as a bottleneck for overclocking. Luckily, AnandTechs Pentium III 550 processor was an engineering sample which did not have the clock multiplier implemented, meaning the clock multiplier could be adjusted in 0.5x increments. An interesting phenomenon occurred after manipulating the clock multiplier of the processor, the L2 cache was disabled. This allowed the core of the Pentium III 550 to truly be put to the test, and after booting into Windows 98 at 100MHz x 6.5 (650MHz) the next step was running at the big 700 (100MHz x 7.0). Surprisingly enough, the sample hit 700MHz without any added cooling (other than the retail heatsink/fan) although its stability was next to nonexistent. Added cooling did help increase the stability of the Pentium III 550 at 700MHz, however for the most part supercooling would be necessary to hit 700MHz on the 550.

So the Pentium III 550s core can hit 700MHz, so what? What this is showing is that Intels manufacturing process is at a point where a few tweaks in the core would allow for speeds greater than 550MHz to hit the streets today, if the need were to rise. From Intels standpoint, the Pentium III 550 doesnt need a 600MHz partner just yet since their closest competitor, AMD, has yet to even release a 500MHz K6-3 part.

The physical limit of the Pentium IIIs current manufacturing process happens to be around the 700MHz mark, meaning that it wouldnt be too far fetched to "hypothesize" the 0.18 micron Coppermine processors that are due out later this year would feature clock speeds of around 700MHz+. It also wouldnt be surprising to see the "mysterious" arrival of a 600MHz Pentium III (not Coppermine, current technology  0.25 micron) if AMD does indeed ship their upcoming K7 at 550MHz as planned. Intel wont stand idle as AMD takes the lead in the clock frequency race, and the potential of the overclocked 550 is proof in itself that Intel could make the move to 600MHz and beyond if the need were to come around.

This is unfortunate for AMD seeing as they do not have the same luxury as Intel in terms of the quality of their yields as well as the number of fab plants they have. Not only must AMD have more lenient QA testing but they must also work with the fact that they dont have as many fabrication plants as Intel. Whereas Intel can crank out a 600MHz Pentium III now and possibly a 650MHz part soon enough, AMD must do a little more work to get 550MHz and 600MHz parts shipping before they can concentrate on raising the bar any more. Luckily for AMD, they will be able to compete with Intel quite well at 550MHz so they wont have to rely on a clock frequency lead to gain the edge in performance when the K7 hits the streets in June/July.