Realistically, even though we've only had an opportunity to sample three Celeron 366 CPUs over the past two weeks, here's how Sharkyextreme feels that the overclocking success percentages map out for the Intel retail boxed versions of the Slot-1 Celeron 366 CPU:
Combo Speed Success Rate
5.5 x 66MHz = 366MHz 100%
5.5 x 75MHz = 413MHz 100%
5.5 x 83MHz = 456MHz 100%
5.5 x 100MHz = 550MHz 65%
5.5 x 103MHz = 567MHz 30%
5.5 x 112MHz = 616MHz* 1%
5.5 x 124MHz = 682MHz* .01%
5.5 x 133MHz = 732MHz* .0001%
*Supercooling required
As we test and evaluate more of the new Celeron CPUs, we'll update this list appropriately. For now, on an average 440BX mainboard with a proper Celeron CPU cooling fan/heatsink (two fans minimum) we're prepared to estimate that 65% of all retail boxed Celeron 366 CPUs can achieve a stable overclocked speed of 550MHz at a voltage rate of 2.0v. The 65% figure goes up slightly if the user owns an Abit BX6 or BH6 mainboard, which allows up to 2.3 volts to be sent to the CPU. (Or more if you know the BIOS flash trick). Figure between a 70 and 75% success for Abit mainboard owners looking to hit 550MHz.
These numbers contrast to the previous overclocking king, the Celeron 300A, which was capable of being successfully overclocked to 450MHz roughly 85% of the time in the retail boxed version with little voltage manipulation. The extra 100MHz of the 550MHz overclock adds tremendous stress on the integrated L2 cache of the Celeron 366, that's probably the biggest obstacle now with consideration to Celeron CPU yield quality.
It still takes a lot of luck to get just the right CPU for this type of experimenting, but buyers should plan on a worst case scenario Celeron 366 overclocking to 456MHz (5.5 x 83MHz) very easily, and with tremendous stability. Of the three Celeron 366s at Sharkyextreme, two could be overclocked and benchmarked to 550MHz, the third wouldn't boot into Win95 OSR 2.1 without its L2 cache disabled. All three hit the 456MHz overclock with their L2 cache enabled, and without any other problems.
