Cooling Performance
Before we move on to the benchmarks proper, let's analyse what extra cooling performance we get from the VapoChill. While it's this that we're trying to essentially benchmark, it's still one of the hardest parts of the article to quantify. With overclocking entirely subjective and so many components able to have an effect on things, it's not exactly scientific sometimes. To that end, the 3.2EE was tested in the same motherboard using a Thermaltake AX478 cooler with a 25CFM fan, to see where air cooling would take the CPU compared to VapoChill XE.The EPoX motherboard has a maximum Vcore setting of 1.60V in the BIOS, running at 1.52/1.53V actual voltage. So pretty much stock voltage (hence the Vcore modification to the board from the factory) for the CPU.
XE Quiet means 2500rpm condenser fan, 40% intake fan speed. XE means 3500rpm and 100% respectively.
| Cooling | Clock (MHz) |
| Thermaltake AX478 | 3600 |
| VapoChill XE | 3840 |
| VapoChill XE Quiet | 3825 |
While it seems like 3.6 might have been the limit with any cooling, given the voltage available from the board, the VapoChill shows us that there's still some impressive headroom when using phase change cooling. A stable 3.84GHz from only 1.52V is impressive from both the processor and the cooling. What's also impressive is that from the same voltage and 'Quiet' VapoChill settings, the following was possible, although not 100% benchmark able.
That's a boot into Windows speed, and very stable for just browsing around and doing light tasks. Anything heavy for the CPU to do would result instability however. Look at the voltage in the top right hand corner. 4GHz at stock voltage from the 167 million transistor behemoth of a processor, shows that the XE is more than capable of handling a regular Pentium 4 at the same speed. Only voltage held it back, with the CPU needing a little more than 1.52V for total stability.
The volt modification on the 4PDA2+ wasn't usable due to BIOS issues, stopping me from benchmarking well into the low 4GHz range. My conclusions were vindicated, any consumer Pentium 4 is easy meat for the XE.
At the benchmarked speed, 3.825GHz (I prefer the Quiet settings I usually use, the louder settings give little extra performance with the test CPU), evaporator load temperatures never rose above -20c, with the idle evaporator temperature at a chilly -33c for testing. That combined to give a CPU diode load temperature of no more than 3c, depending on the benchmark.
The latest firmware and Window ChillControl control software allow you to set less than -30c hold temperature for the ChillControl, something the DOS control software couldn't. Perfect for P4's it seems, given that testing with the Barton Athlon XP never let the evaporator temperature drop that low. Here's a screenshot of the WinCC settings used for all the testing.
Note that for all testing, the CPU was set to 15x multiplier. The 3.2EE sample (and hopefully all retail examples) was fully unlocked, all multipliers were available, right up to 28x. Seeing it set to 28 x 200 in the BIOS on first boot is enough to scare the daylights out of any P4 owner, VapoChill XE or not. Thankfully that wasn't a proper boot speed, just a selection peculiarity with the 4PDA2+ BIOS and the 3.2 Extreme Edition. 15x multiplier with the bus speed range I was working in, combined with the XMS3200 memory, gave excellent performance when set to 5:4 memory ratio. At 255MHz front side bus clock, that's 202MHz memory clock and the possibility of setting tight memory latencies. CL2, Tras=5, Trcd=2 and Trp=2 were used.
On to the benchmarks.
