Intel has been granted a patent which ties cooling fan speed to CPU leakage power.
The document, originally filed in September 2008 and approved by the USPTO on March 27, 2012, describes a process to calculate the fan speed that will decrease platform power the most.
The patent specifically refers to CPUs that operate in a low temperature state with a fan running at low speed. According to Intel, the power used to operate a fan can be offset by saved processor power. The process to determine the perfect fan speed would include
- measuring the processor power and the fan power when it is initially determined that the processor is operating in a low temperature state prior to any fan speed adjustments
- calculating the system power prior to any fan speed adjustments
- increasing the fan speed
- measuring the processor power and the fan power for the increased fan speed
- calculating the system power for the increased fan speed
- dynamically adjusting the fan speed based on changes to the system power
- measuring the processor power and the fan power for the dynamically adjusted fan speed
- calculating the system power for the dynamically adjusted fan speed
- selecting the preferred speed when the system power is determined to be substantially at lowest value
It is interesting to note that this patent highlights a scenario in which a fan is not just used to bring the temperature of a CPU down from a certain exceeded threshold, but is leveraged as an almost-always-on option to reduce the power consumption of a CPU platform. The idea obviously refers especially to the problem of leakage power that can be reduced by containing the CPU temperature.