Main Menu
|
Water-cooling for Everyone 3: Swiftech H20-120 Rev. 3 and Thermaltake Bigwater (cont.)Performance: Both water-cooling kits went leak free for 48 hours in the Leak Tester 2000. The green coolants made it especially easy to notice if there were leaks.
Testing was done on an AMD XP 2400+ with an AOpen AK79D-400 Max motherboard. Furthermore, the kits were installed in the Silverstone TJ06 case (as opposed to a "bench" setup) to simulate performance in a standard computing environment. Both kit's radiators were externally installed following their respective guides. As for Swiftech's H20-120, the Radbox was used.
First idle temperatures were obtained by turning on the computer and doing nothing for five minutes. Load temperatures were obtained by running Sandra's Burn-In Wizard (CPU Arithmetic and Multi-Media Benchmarks only) 20 times. The second idle temperature was obtained by leaving the computer alone again for another five minutes. All temperatures are in Celsius and the air temperature entering the case was 27-28C. Thermaltake's Bigwater was tested with the fan controller at the highest and lowest setting. Swiftech's H20-120 was tested with its default fan speed as well as with the fan connected to a modified 3-pin to Molex adapter that adapts the fan to the 5V rather than the 12V, cutting the fan speed roughly in half. The results were:
As the results show, the Bigwater at the high fan speed setting performed superbly. Though Thermaltake reports the high fan speed noise to be 21 dB, it sounds much louder. Turning the fan speed all the way down had a noticeable impact on performance but also on noise; it was practically silent. With its default setup, the H20-120 Rev. 3 also performed superbly. Swiftech reports the H20-120's fan to be 34 dB but it sounds like it is much lower. Surprisingly, cutting the fan speed down resulted in only a very minor temperature increase. With the fan's voltage cut down, the H20-120 also ran practically silent. For overclocking tests, the CPU was able to match its PR speed at 2400MHz (up from the 2000MHz stock) with the Vcore raised to 1.8V (up from stock of 1.65V).
While the stock temperatures showed the two kits to perform relatively equal (at least in high fan speeds), the overclocking results show a clear difference among the Bigwater and H20-120. At both fan speed extremes, the Bigwater showed a load temperature increase of nearly 10 degrees compared to stock temperatures. The H20-120 had about an 8 degrees increase in load, but again the lowered fan speed had only a minor affect on temperatures.
|




