Processor Temperature Measurement via Median Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing temperature measurement systems, such as Intel's PECI, often provide inaccurate readings due to outlier values, leading to inefficient fan control and increased noise, as they rely on arithmetic means or running averages that are influenced by extreme readings.
Innovation Solution
Oversampling environmental parameter readings, specifically temperature readings, at least three times and using the median average instead of the arithmetic mean to discard outlier values, ensuring more accurate fan control by ignoring the highest and lowest readings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If arithmetic mean or running average methods are used to process temperature readings, then the system can provide continuous temperature data for fan control, but the temperature readings become inaccurate due to influence from outlier values
Solution Approach 1:
The patent extracts and removes outlier temperature readings from the dataset before calculating the average. By identifying and discarding extreme values that deviate significantly from the median, the system prevents these outliers from distorting the temperature measurement, thereby improving accuracy while maintaining reliable fan control
Solution Approach 2:
The patent introduces the median value as an intermediary step in the temperature calculation process. Instead of directly averaging all readings, the system first determines the median to identify outliers, then uses this information to filter the dataset before computing the final average. This intermediary mechanism protects against outlier influence
2Reliability
If fan speed is increased to compensate for inaccurate high temperature readings, then the processor temperature can be maintained in a safe zone, but the system generates increased noise and energy consumption
Solution Approach 1:
The patent converts the potentially harmful effect of outlier readings into a benefit by using the median calculation method. Outlier values that would normally cause erroneous fan speed increases are instead used to identify the true median temperature, allowing the system to maintain appropriate fan speeds and reduce unnecessary noise while still ensuring processor safety
3Speed
If isolated outlier temperature readings are used for fan control, then the system can respond quickly to temperature changes, but the fan control becomes inefficient and generates unnecessary noise
Solution Approach 1:
The patent applies partial action by selectively processing only the relevant temperature readings. Instead of using all readings equally or waiting for extensive data collection, the system takes a partial approach by using the median of a limited set of readings to quickly identify the true temperature trend, enabling efficient fan control without unnecessary responses to outliers
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides more accurate temperature readings and reduces unnecessary fan speed changes, thereby minimizing noise and ensuring effective cooling while maintaining processor temperatures within a safe range.
Implementation Method 1
Devices that monitor temperature and voltage are often included as part of such systems in order to maintain the integrity of the system components... Temperature-to-digital conversion for IC-based temperature measuring solutions is often accomplished by measuring a difference in voltage across the terminals of a diode when different current densities are forced through the PN junctions of the diode. The resulting change (ΔVBE) in the base-emitter voltage (VBE) between the diodes is generally proportional to temperature.
Data Source
AI summary
Temperature readings obtained within a computer system from the location of monitored circuit elements may be oversampled at least three times, and a median average of the three parameter readings rather than the arithmetic mean may be used for controlling a device, e.g. a fan, configured to regulate the environmental parameter, e.g. temperature, a the location of the monitored circuit elements. For example, when a CPU temperature reading is requested by the system comprising the CPU, a thermal monitoring system may acquire at least three consecutive temperature readings of the CPU, discard the highest temperature reading and the lowest temperature reading, and return the median reading to be used in controlling a fan configured to regulate temperature at the location of the CPU, resulting in more accurate temperature readings and more accurate fan control. In various implementations, more than three readings may be considered at a time, and running averages based on median values may be computed in a variety of ways to obtain a temperature control value to control the fan.