asetek Vapochill Extreme Edition
How can cooling be such a hot topic?

Written by: Robert Maloney
September 29, 2003

How We Configured Our Test System:

To help fully explain the scores we listed in some of the following benchmarks, we felt it was necessary to explain how this system was set up before running the benchmarks. Our normal practice is to start off by manually optimizing the BIOS settings to the most aggressive system options available to us.  This meant that the memory frequency was manually set to DDR400 with the CAS timings set to 2-5-2-2.  Since the Epox 4PDA2+ v2 came with Accelerated Memory Mode, a form of memory enhancement, we opted to enable this as well, which was then set to the "MAX" performance level.

Once the BIOS options were set, the hard drive was formatted as a FAT32 partition, and Windows XP Professional with Service Pack 1 was installed. After the Windows installation completed, we installed the latest Intel chipset drivers and upgraded to DirectX 9.0a.  We then installed the drivers for the rest of the components, using drivers supplied on Epox's driver CD, except for the ATi Radeon 9700 Pro video card.  For the 9700 Pro, we installed the latest (at the time of testing) ATi Catalyst drivers, version 3.4.  Auto-Updating, Hibernation, and System Restore were disabled, and then we set up a 768MB permanent page file. We set the visual effects to "best performance" in system performance to limit any effects these settings would have on the benchmarks.   Satisfied that every thing was set up correctly, we installed all of the benchmarking software, defragged the hard drive, and rebooted one last time.  One complete set of temperatures and benchmarks were then completed using a stock Intel air-cooled heatsink, in a standard, enclosed chassis with no other case fans or cooling options.

We then set to building our Vapochill unit, as described on the previous page.  Completing that task, we booted into Windows XP and ran a few quick tests to see that Windows was running stable, and that the Vapochill unit was maintaining the default temperature mandated by the ChillControl unit.  After a quick reassembly of the CPU-Unit, we attached the monitoring cable to an open serial port, and booted into the ChillControl software.  We manually set the HOLD temperature to -20 degrees Celsius, and left the Start PC temp at -5 degrees.  To ensure proper cooling, we set both Fan Speed settings to 85%, which we felt was a decent compromise between speed and noise control.  Finally, we manually entered the CPU speed (2400MHz), saved the settings, and went to testing.  After we got our initial set of results, we went back into the ChillControl, and set the hold temperature to -30 degrees Celsius, which ramped up the compressor output, hovering at about 3500rpm.  We re-ran the our tests to see how this would affect the temperature readings.

HotHardware Test Setup Can the Vapochill XE keep its cool?

CPU Temperatures and some overclocking - Updated Now with even more power!

The first, and most obvious, test was to compare the difference in the temperature of our Pentium 4 2.4GHz CPU after fifteen minutes at idle, and then again when placed under load.  To simulate the load, we replayed the CPU tests from 3DMark03 repeatedly for another fifteen minutes.  Temperatures were then recorded using CPU Diode temperatures as reported in the BIOS.  Unfortunately, the Epox BIOS did not give us the correct information we needed, so in these graphs the temperature shown was taken from the an additional temperature sensor, which was connected directly to the Pentium 4 heat spreader..

Once the Vapochill got running, it was a steady decline down to the Hold temperature of -5 degrees Celsius for the evaporator, after which Windows began its booting process.  The temperature continued to drop, eventually bottoming out at -26 degrees.  In time, it gained a couple of degrees back, finally settling at a cool -24.5 degrees Celsius.  Meanwhile, the additional diode we had connected was reading at -9.5 degrees Celsius at 2000 rpm, and reached as low as -13 degrees C at 3500rpm.  Compared to our air-cooled P4, this was a difference of 51.5 degrees.  Putting the system under some stress naturally caused temperatures in both setups to rise.  While the P4 with the stock Intel HSF jumped up 11 degrees, a 29% increase in temperature, the ChillControl prevented the Vapochill cooled system from rising above -2 degrees C in our original test, and -9C when fully cranked.

Convinced that the CPU-Kit was installed correctly and performing well, we then set out to better our original overclocking results with the Epox 4PDA2+ v2.  If you recall, in that review we had hit a high stable overclock at 281MHz.  Secure with this knowledge, we immediately set our system up to at least match this speed, and had no problems whatsoever.  In fact, we let the system run overclocked for an additional 15 minutes and checked the temperature to see how the Vapochill would fare at a much higher core speed and voltage for the CPU.  As you can see, the Vapochill was unfazed as the temperature only managed to rise two and a half degrees to -7 degrees Celsius at 2000rpm or by half a degree to -12.5 degrees Celsius at 3500rpm.  By comparison, at the same overclocked speed with normal cooling measures, the CPU was running at 49 degrees C.

        
WCPUID at Stock and Overclocked Speeds
(click on image to enlarge)

Since we obviously couldn't (or wouldn't) stop there, we continued to increase the front side bus until we reached the highest speed that Windows XP would open under, running at a 310MHz FSB.  While we were able to get into Windows, the system was highly unstable, even with increased VCore adjustments.  We slowly backed off the FSB speeds and found 303MHz to give us the most stable operating system.  Thus, we were able to gain an additional 22MHz on the front side bus over an overclocked system using air-cooled methods.  In total, this resulted in running the system 103MHz over normal speeds, a 51.5% increase in the FSB, with our modest P4 2.4GHz CPU now running at an utterly superb 3.63GHz.

Benchmarks and final rating