AMD Cool n Quiet
Similar to Intel's SpeedStep Technology, which originally came about to prolong the battery life for mobile users through reducing the CPU clock frequency and voltage, AMD's Athlon 64 processors support the use of Cool 'n' Quiet Technology. Advanced Micro Devices Cool 'n' Quiet Technology is targeted for both consumer and commercial use through reducing the heat output, noise level, and power consumption for systems running Athlon 64 processors. To not infer with the computing experience, Cool 'n' Quiet dynamically clocks the CPU to lower the power consumption while delivering performance-on-demand so the system will not run slow when under heavy usage (i.e. gaming, encoding, and compiling). In fact, AMD Cool 'n' Quiet Technology was recognized with Energy Star recognition in 2005 from the United States (EPA) Environmental Protection Agency. Unlike Intel's SpeedStep Technology that is limited to desktop utilization on the Pentium 600 and 800 (excluding 820) and newer series, Cool 'n' Quiet is supportive by all Athlon 64-based systems even going back to Socket 754 systems. In addition, Cool 'n' Quiet is compatible with AMD's Athlon 64 X2 dual-core processors. However, to have the technology enabled the motherboard, BIOS, software driver, and CPU cooler must be supportive of this technology. Before Cool 'n' Quiet, it could be found in AMD notebooks under the name of PowerNow! Technology. As previously mentioned, Cool 'n' Quiet Technology works by reducing the CPU clock speed and processor voltage when full processing power is not needed (i.e. idling, surfing the Internet, and reading messages). Unlike Intel's SpeedStep that drops the CPU multiplier down to 14x, and then reduces the voltage, Athlon 64 processors have multiple levels of operation. These levels of operation range depending upon the specific CPU. As shared in one of AMD's technical demonstrations, the Athlon 64 3500+ has four power states/levels at which the CPU can operate -- 2.2GHz @ 1.5V, 2.0GHz @ 1.4V, 1.8GHz @ 1.3V, and 1.0GHz @ 1.1V. Of course, the faster the processor the greater number of power states available. The same philosophy of these different "states" is similar to that of ATI's PowerPlay for use on the graphical front. AMD also mentions that Cool 'n' Quiet Technology have the ability to adjust these performance levels by up to 30 times per second to ensure responsiveness and minimal CPU stalling when switching from a non-demanding task to a strenuous environment, and vice-versa. In addition to appropriately adjusting the frequency and voltage, the CPU heatsink fan can adjust rotation speeds depending upon load. AMD does produce a great deal of software for Microsoft Windows XP and 2000 to utilize Cool 'n' Quiet Technology from Dashboard software to AMDClock and Power Monitor, however, as of right now AMD only has an Athlon 64 Cool 'n' Quiet driver (v1.50.03) for Linux. This driver is for use in conjunction with Linux 2.6.10+ kernels and cpuspeed or cpufreq. Most modern day Linux distributions should have no problems utilizing Cool 'n' Quiet with an out-of-the-box configuration on most supported systems. In an effort to seek the solid advantages of using Cool 'n' Quiet on a desktop Athlon 64, as well as to discover any disadvantages, we recently ended with an array of in-house tests. To begin this expedition, below are the hardware and software components used as the basis of testing.
|Processor:||AMD Athlon 64 3000+ (Winchester)|
Tyan Tomcat K8E-SLI
|Memory:||2 x 1GB OCZ PC-4000|
|Graphics Card:||Leadtek PX6600GT 128MB|
|Hard Drives:||Seagate 200GB SATA NCQ|
|Cooling:||AMD Stock CPU HSF|
|Power Supply:||SinTek 500SLI 500W|
|Operating System:||Fedora Core 4|
|Linux Kernel:||2.6.15-1.1830_FC4 (x86_64)|
|GCC - GNU Compiler:||4.0.0|
|Graphics Driver:||NVIDIA 1.0-8178|
For accurate testing, we had used the ASRock 939Dual-SATA2 motherboard as well as the Tyan Tomcat K8E-SLI S2866. The two motherboards were tested separately to ensure precise measurements Of course, both motherboards support Cool 'n' Quiet and it allows the technology to be enabled/disabled from their respective BIOS. Unfortunately, an all too common occurrence at Phoronix has been discovering the latest AMD motherboards not offering detection with the latest version of LM_Sensors (v2.9.2) or alternatively, the hardware sensors reporting inaccurate measurements. However, the next version of LM_Sensors (v2.10.0) is scheduled for an official launch in the second week of February, and the changes do make it look to be a very promising release. Seeking to attain the most accurate numbers, as well as the lack of LM_Sensors detection, our Cool ‘n’ Quiet testing was limited to using hardware equipment. Our key instrument monitoring the Wattage of the system was the Seasonic PowerAngel as it was connected to the power supply. For our temperature tests, we had attached a metal thermal probe to an inside fin of the CPU heatsink. The thermal probe was connected to a WinMax Global WIN8696 digital multimeter with temperature reading capabilities in both Celsius and Fahrenheit. Unfortunately, due to the present state of Linux sensors with the latest-and-greatest in enthusiast hardware, we were unable to use LM_Sensors to properly report the CPU die temperature. The PowerAngel and thermal probe were used to measure the temperature and power consumption differences between Cool 'n' Quiet and then again when the technology was disabled. Outside of our direct measurements, we measured compilation time immediately after idling to see if truly there are any real-world performance differences when it comes to the computing speed. Of course, during this entire experiment the testing ambient room temperature was maintained as well as all other constants to ensure accuracy. The CPU heatsink is included with retail Athlon 64 models and the thermal paste used was Arctic Silver 5.