Overclocking Comparison

The biggest question on everyone's mind: how does the 400 overclock? Taking into heavy consideration that we only have one slot-1 and one socket-370 CPU, there's a lot of room for error. Both of our processors are engineering samples, and thus may not indicative of production sample performance (who knows, maybe Intel sends the best of the batch to manufacturers, maybe they don't).

Our two CPUs performed rather similarly in the overclocking field as well. Since the 370SBA didn't have jumper settings for intermediate frequencies between 66 and 100Mhz, Supermicro was kind enough to add ratio multipliers to the motherboard.

Please remember that production models of both PPGA and SEPP CPUs will be ratio-locked, so that changing the multipliers will have no effect.

We wanted to see what the physical limits of the CPU were, since the chip may fail at 6.0x100Mhz, but succeed at 83.3Mhz, still yielding a net 100Mhz gain.

The Celeron PPGA indeed failed at the 100Mhz bus speed. It also failed at 5.5x100. At 5.0 x 100, a close approximation of 6.0 x 83.3, the PPGA Celeron booted up fine, clocking to a successful 500Mhz. The SEPP package was easier to deal with, since the P6SBA has multiplier jumpers by default. The Slot-1 Celeron also failed 600Mhz and 550Mhz, but stabilized at 500 (also 5.0 x 100). Socket-370 motherboards which support clock frequencies of 66, 75, 83, and 100Mhz will be able to take advantage of these overclocking speeds.

What about using higher voltages? Without a Socket-370 motherboard capable of adjusting voltage settings, we were limited in this test to the Slot-1 chip. We then took the Slot-1 Celeron and placed it into an Abit BH6, the current choice of 2.2v Celeron overclockers. At 2.1v, we hit success at 550Mhz (5.5 x 100). Ah, the freedom of multipliers. Without the cheat of an engineering sample, this speed wouldn't be possible on the 400, as it sits squarely between 75Mhz and 83Mhz. At 2.2v, the CPU would not POST.