I cannot recall when was the last time we were able to write an overclocking article based on an AMD processor (thank you, archive). Since the demise of the Athlon64 architecture there hasn't been much to celebrate on this front for AMD CPUs.

The original Phenom B2 stepping was truly stuck on its default speed, and we are glad this has taken a turn with AMD's latest offering.

The B3 Phenom stepping was a bit better, though it was only the “Black Edition” processors with the unlocked multiplier that were any good. Then you had the power consumption issue, where feeding the original Phenoms even more power was never a good idea.

We were not entirely surprised when early reports claimed that the Phenom II X4 940 could reach 4.40GHz when overclocked and using water-cooling. The new Phenom II X4 processors are very efficient and can use less power than a Core 2 Duo. Sadly, using an air-cooler we were unable to get anywhere near that target, falling short at 3.60GHz with our sample processor.

Various reports seem to indicate that outgoing retail versions of this processor overclock better than the sample processor we have. Although our sample would post at 3.90GHz, it failed to load Windows Vista at that speed. Entering the operating system was possible at 3.75GHz, but it wasn't 100% stable either.

In order to overclock our sample processor we used the simple multiple method, leaving the FSB intact. To reach a stable 3.60GHz the multiplier was set to 18x (default is 15x) and the voltage increased to 1.500v.

Using the standard voltage of 1.350v we reached 3.50GHz, so it was disappointing that such a large voltage was needed to stabilize the processor for a mere 0.1GHz speed bump.

Looking around the web we found a variety of results posted, anywhere from 3.50GHz to 3.90GHz using the reference air cooler seems realistic. Nevertheless, we will show you what a 20% increase in clock speed can achieve in terms of pure performance boost, so that should help you determine if overclocking a Phenom II X4 is worth the hassle or not.