We see a similar situation during the Everest stress test, during which both CPU cores were maxed out at 100% utilization. The LCLC and Silverstone NT-06 at 2640RPM keep pace with each other and produce the best thermal result, at 45C. This is significantly better than the nearly 60C that the Intel stock cooler was able to achieve.
While 60C is hardly a dangerous temperature level for a Core 2 Duo, 45C is much better. At 45C, all systems should have perfect stability. The lower temperature achieved by the LCLC will also extend the life of the processor compared to the Intel stock cooler. During the Everest stress test, the GPU temperature rose from an idle temperature of 56.3C to an average of 58.6C. This is because the GPU block receives the heated exhaust water from the CPU block before it reaches the heat exchanger to be cooled.
Note that the LCLC has a greater thermal delta between idle and load than the stock cooler. This is because the stock cooler's fan speed is thermally controlled and increases in speed when the graphics card is under load. On the other hand, the LCLC's GPU block is cooled by the heat exchanger. The fan we used on the heat exchanger is not thermally controlled and remains at a constant speed. This means the stock cooler has greater cooling ability while under load, while the LCLC's cooling ability remains constant. Hence the higher thermal delta. This can be remedied with a thermally controlled heat exchanger fan.
The graphics cooling assembly fan is powered by the graphics card via a 4-pin connector. However, only 2 of the 4 pins are being used and the fan is not thermally controlled which means it stays at a constant speed. This helps explain the temperature delta difference between the LCLC and the stock cooler.
Acoustic Performance: We did not take detailed acoustic measurements for this review since the LCLC does not come with a fan for the heat exchanger, therefore the noise level of the system will depend on the fan used. However, the pump and graphics assembly fan both produce noise but they are extremely quiet. The noise level of both the pump and the graphics assembly fan were below the ambient noise floor of our test lab which means they are effectively 'silent'. We were not able to detect the sound of the pump or graphics assembly fan over the noise produced by the power supply fan, from 1 meter away. In comparison, the GeForce 8800 GTX stock cooler is clearly audible.
Throughout our tests we used a Scythe S-FLEX SFF21D 120mm fan to cool the LCLC's heat exchanger. In this configuration, the system was extremely quite, and the only noise we were able to detect from 1 meter came from the S-FLEX, power supply and hard drive (while seeking). We believe it would be possible to use an even quieter fan and still maintain respectable thermal performance, however a fan is necessary. Without a fan, CPU and GPU temperatures both rose beyond 80 degrees Celsius over a period of 10 minutes.
The Intel stock cooler's fan was much louder than the S-FLEX. The Silverstone NT-06 comes with a speed adjustable, extra-deep fan (35mm instead of the standard 25mm). At 750 RPM it is nearly silent and only slightly more audible than the S-FLEX. At the max speed of 2640 RPM, the Silverstone fan reaches mini-vacuum cleaner status, both in the volume of air being moved and noise level. The fact that the LCLC was able to outperform the Silverstone with a much smaller fan is testament to its superb cooling efficiency.