Overclocking
We started out by running benchmarks at a stock speed of 2.8GHz with a 133MHz front-side bus, using a retail Intel heatsink. Once that was done, we shot for the highest overclock possible with the stock heatsink, and wound up with a 151MHz front-side bus using a CPU core voltage of 1.825V. Given the 2.8GHz P4's multiplier of 21, that works out to 3171MHz. In case you're wondering, both of the overclocked speeds required locking the PCI bus at 33MHz, which is easily accomplished via a setting in the BIOS of the IT7-MAX2.

However, completing benchmarks does not a stable processor make, and there is no way I would run the CPU at this speed, with this heatsink, in day-to-day use. I encountered some random errors during testing that led me to believe the stability was less than perfect. Moreover, this processor was just running hot. I mean damn hot. I couldn't hold my finger on the side of the heatsink for more than a second or two

Next up was the Vapochill install, which you've already read about. Once that was done, we ran another set of tests at stock speeds, to test a curious statement in the Vapochill manual:

An issue not often realized is that a CPU at low temperatures not only will be able to run at higher clock frequencies. It will also have an increased performance compared to an identical CPU running at the same frequency, but where heat is removed with traditional passive cooling.

An interesting statement, no? We'll explore it more in the benchmark results. Once the second set of stock testing was done, it was time for the Vapochill overclocking experiments.

Cue evil laugh, bolt of lightning in the background.

So how did we do? Well, not bad. We topped out at a 161MHz front-side bus using a 1.85 core voltage, for a CPU speed of 3381MHz. Personally, I was disappointed, but the top speed had little to do with the Vapochill and more to do with our choice of motherboards. As I mentioned before, we used an Abit IT7-MAX2 for our testing, and it ended up running out of steam before the processor or the Vapochill did.

The problem was voltage. The IT7-MAX2 tops out at 1.85V for the CPU voltage. While this is probably a pretty good limit under normal circumstances (with a traditional heatsink) with the Vapochill you would probably do better with a board that topped out at 1.95 or so. To make matters worse, it seems that the IT7-MAX2 wasn't really giving its all; when set to 1.85V, the PC Health section of the BIOS indicated 1.79V.

Based on the pattern I saw when determining the maximum stable CPU speed, my belief is that the CPU would've gone even faster with more juice. I hope to do a follow-up article and explore things further, either with a different Pentium 4 motherboard or with an Athlon configuration. In the meantime, though, I'd like to point out that the Vapochill had no real trouble maintaining an evaporator temperature of 15 degrees below Celsius, even at 3381MHz with a 1.85V core voltage. I have no doubt the Vapochill could handle a significantly faster/hotter CPU than we presented it with here, and the main reason I'm disappointed is because the core voltage limits kept me from exploring the limits of the CPU. Even with more juice, I seriously doubt I'd really be pushing the Vapochill.

Before moving on, I should point out that we received the Premium Edition of the Vapochill for review. The Premium Edition differs from the Standard Edition only in terms of cooling performance; the Standard Edition can dissipate up to approximately 130 watts of heat energy, while the Premium Edition tops out around 160 watts. You can look at a graph of the performance differences here. The Premium Edition costs approximately $50 more than the Standard Edition.

Our processor testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run at least twice, and the results were averaged.

Our test systems were configured like so:

Thanks to Corsair for providing us with DDR400 memory for our testing. If you're looking to tweak out your system to the max and maybe overclock it a little, Corsair's RAM is definitely worth considering.

The test systems' Windows desktops were set at 1024x768 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.

We used the following versions of our test applications:

• Cachemem 2.65MMX
• SiSoft Sandra Standard 2003 (2003.1.9.3x)
• ZD Media Content Creation Winstone 2002 1.0.1
• POV-Ray for Windows version 3.5
• Sphinx 3.3
• MadOnion 3DMark 2001 SE Build 330
• Unreal Tournament 2003 demo benchmark

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.