Articles: Cooling
 

Bookmark and Share

(25) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 ]

Performance

So, the testbed with the overclocked to 4GHz processors was placed inside a thermal chamber with the constant temperature maintained at 30o C. Before the CPU temperature got stable the system was idling for a while. After the temperature in idle mode got stable we measured it (idle). Then we ran the S&M utility version 1.7.3 with 100% workload and had it running until the CPU temperature stopped rising (burn mode). The temperature changes were all checked and the following diagram was built from the results:

You can see very well how the temperature inside the thermal chamber was rising and then leveling out. Slight deviations from the level can be explained by the way the thermal chamber works: it turns on and off to make sure that the temperature inside is 30±0.5o C (29.5-30.5o C). The temperature in idle mode varied between 41.5o C and 43.5o C resulting into the average of 43o C. In burn mode under the workload created by the S&M utility the temperature rose to 52.5-53.5o C, 53o C on average. Everything seemed to have gone perfectly well, and I could have ended the tests here, however the Scythe Ninja cooler that was the best in that previous test session could only cool the CPU down to 47o C in idle mode and 60o C in burn mode.

I started looking for differences between the testing and measuring conditions of our today’s test session and the previous one. And I did find them! I used the RightMark CPU Clock Utility version 1.8 for throttling management, and this time I had a newer version of this program. So, I repeated my tests again.

If we disregard the preheating period for the thermal chamber that was absent during the second round of tests, the results shown by the Tuniq Tower 120 were absolutely the same: 43o C and 53o C in idle and in burn modes respectively. The conclusion was as follows: there must be some other small differences between the test systems that I accidentally omitted. Otherwise, how could I explain that an unknown cooling solution easily outperforms the best coolers out there? And the advantage is really tangible! I couldn’t think of anything else that could have been different, so there was only one way-out left: I had to test the other coolers again, too.

Well, since I have just revealed such remarkable results of the new Tuniq Tower 120, why not pay some extra attention to the thermal paste that comes with it? However, I regretted this brave decision: sticky, thick thermal paste of dark gray color would want to evenly smear over the CPU cover. It would stick to anything possible but for the processor. After some vain efforts to make it look decent, I decided to put up with a few chunks in the very middle and set the cooler on top of the CPU. Actually, it would solely be Tuniq’s fault if the results worsen, because of this stubborn and hard-to-smear thermal solution.

But, the situation didn’t get any worse at all. With Tuniq TX-1 thermal paste the results remained the same: 43o C in idle mode and 53o C in burn mode. It indicates not only that the thermal conductivity of the Tuniq paste is similar to that of Zalman paste, but also that the differences between various kinds of thermal paste are negligible because of high processor heat dissipation.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 ]

Discussion

Comments currently: 25
Discussion started: 02/17/06 07:26:23 PM
Latest comment: 01/10/10 03:37:48 AM

View comments

Add your Comment