Articles: Cooling

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 ]

Bonus: Overclocking Intel Core 2 Quad (B3) at Different Core Voltage

Systematically testing various cooling systems, I often have to change not only the frequency of our quad-core CPU but also its voltage due to obvious reasons. Air, liquid or thermoelectric CPU coolers all differ, sometimes radically, in performance. So I decided to examine the overclocking potential of our Intel Core 2 Quad Q6600 (stepping B3) depending on its voltage. The temperature of the CPU was also monitored.

The tests were performed under the conditions described in the Testbed and Methods section of this review. The CPU was cooled with a Thermalright SI-128 that had a 120x32mm 9-blade fan working at 1160rpm (the SilverStone FM122 model). I tried to keep the test conditions as close to reality as possible, so I tested the CPU in a closed system case. To improve the cooling I replaced the system fan on the side panel with a high-performance but relatively quiet Scythe Minebea 4710KL-04W-B29 at 1140rpm.

The default core voltage of our CPU is 1.2875V and that was the starting point of the test. The mainboard’s BIOS Setup allowed to change the core voltage with a step of 0.025V, but the first step was smaller (0.0125V). I also monitored the voltage with CPU-Z which always reported a different value than set in the BIOS. The CPU multiplier was automatically reduced from 9 to 6 in idle mode.

Here is a table with results and a diagram that shows how the CPU frequency depends on its voltage:

It is clear that good air cooling can help achieve the maximum CPU frequency growth (+28%) even without increasing the core voltage. Then, after the voltage was lifted up by 0.3125V, the frequency could only be increased by 15% more. Note that the CPU shows an almost linear dependence of frequency on voltage. Increasing the core voltage above 1.6V did not affect the frequency potential of the CPU.

Here are temperature data for the CPU at different core voltages:

Note that the quad-core CPU is stable at its default frequency even when the voltage is reduced (you cannot set a core voltage lower than 1.1V on the ASUS P5K Deluxe mainboard). The temperature of the CPU is ridiculously low then, 29°C under peak load. This indicates the opportunity to use a passive cooler for the B3 stepping Core 2 Quad. CPUs of the more progressive G0 stepping dissipate even less heat and thus feature higher overclockability.

And then the picture changes to the opposite and the temperature rises to a critical mark which is close to the thermal throttling threshold of our CPU (82°C). The temperature is ready to jump up as soon as you increase the core voltage – you should keep this in mind when overclocking. The CPU temperature is lower in idle mode than the ambient temperature – the monitoring program must be showing an incorrect, too low, value. This bug seems to have been corrected in SpeedFan 4.34 beta 34. On the other hand, SpeedFan 4.33, TAT and ASUS Probe unanimously reported the same CPU temperature. Well anyway, even if the temperature data are not 100% accurate, this test helps determine the cooler’s practical capabilities which are more important than dry temperature data.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 ]


Comments currently: 4
Discussion started: 11/24/07 03:17:08 AM
Latest comment: 11/26/07 09:28:58 PM

View comments

Add your Comment