AMD Cool n Quiet
Our first test is to demonstrate where Cool ‘n’ Quiet is designed to shine. For our initial measurements, we simply allowed the system to idle inside of GNOME v2.10 at 1280 x 1024 with all other power management features and the screensaver disabled. The idling process had lasted for 30 minutes and after that time had expired, the power consumption and temperature were recorded. The process was repeated for both when Cool ‘n’ Quiet was enabled at 1.00GHz and then again when it was disabled and the CPU running at its stock speed of 1.80GHz. As mentioned on the previous page, the Cool ‘n’ Quiet testing had occurred with two different motherboards using the same components – Tyan’s Tomcat K8E-SLI S2866 and ASRock’s 939Dual-SATA2.
As can be seen from the graphs above, there certainly were differences between Cool ‘n’ Quiet being enabled and then again when it was disabled. The power differences were somewhat respectable at approximately 4-6 Watts, or 5-8% reduction. In addition, there was approximately a 0.9-1.1 °C (33 ~ 35 °F) improvement in the CPU HSF temperature. Of course, the thermal probe was attached to the aluminum fin and hadn't directly come in contact with the CPU core, thus we do not know the actual improvement made. Our next test for investigating AMD’s Cool ‘n’ Quiet Technology is compiling LAME (v3.96.1) immediately after the system was idle. The purpose of this test is to see any power consumption or temperature differences as a result of the CPU load, and to see if truly there are any real-world computing differences as a result of Cool ‘n’ Quiet.
Looking over the power consumption under the CPU load from compiling, the Wattage was relatively constant as well as our thermal probe reading throughout the various tests on both the ASRock and Tyan motherboards. Moving onto the time required to compile LAME v3.96.1 using GCC v4.0.0, our tests has showed the Cool ‘n’ Quiet enabled trial to run marginally slower than the setup with the technology disabled with both the 939Dual-SATA2 and K8E-SLI. However, this is an incredibly small amount of lag and is possibly a coincidence for which we are still investigating. For reference, when overclocking the various power-states/levels are kept constant according to cpufreq and are not based upon a ratio of the front side bus. However, in our tests when we had pushed the K8E-SLI to its limits of a 250MHz FSB using the Athlon 64 3000+ (250MHz x 9 = 2250MHz), and enabling Cool ‘n’ Quiet, the CPU ran at 1250MHz while idling according to cpuinfo; meanwhile, cpufreq had reported the processor to be running at its 1000MHz state. We have additional investigations and Linux technological reports planned for the near future, including thoroughly examining Intel’s SpeedStep Technology, and further investigating AMD Cool ‘n’ Quiet.
If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.