HANDS ON GUIDE

Step 5: How stable is stable?

Although my Q6600 was now happily running in Windows at 2.7GHz, coping with the desktop isn't an indication of stability. At each stage of overclocking, it's important to stress-test the system. The CPC Media Benchmarks (www.custompc.co.uk/benchmarks) are a good all-round system test, but to really push a CPU, you need to max out every core. To do this, I used Orthos (http://sp2004.fre3.com/beta/beta2.htm). Orthos can only run on two cores, so you need to run two copies, and use Task Manager to manually set its affinities to Core 0/1 and Core 2/3 to make sure all four cores on a quad-core chip are in use.

While Orthos was running, I also had CoreTemp on the go to check my chip's temperature (www.thecoolest.zerobrains.com/CoreTemp). The BIOS - and most of the software supplied by the motherboard manufacturer - only has access to the CPU's thermal diode, which measures the temperature of the CPU's IHS (Integrated Heatspreader) at its geometric centre. However, this value depends on how the BIOS is calibrated to read the sensor, and some BIOSes are pretty inaccurate. CoreTemp collects data from a Digital Thermal Sensor (DTS) in each CPU core, which is factory-calibrated and, since it doesn't rely on the motherboard or BIOS, it's more accurate. Under load, at 2.7GHz my Q6600 was only just breaking 60˚C, which is well below the T-junction MAX (the shutdown temperature) of 100˚C for the G0 Q6600.

Latest Hands on guides

These are the most popular features articles over the past three months.

Mobile Broadband

Compare prices

Fastest, cheapest 3G mobile broadband dongles from 3, Vodafone, T-Mobile and Orange
from just £10/month