Processor Power Measurement Calibration via Test Routine
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current processor power measurement techniques are inaccurate due to lossy current sensors, leading to inefficiencies in data center power management and workload distribution.
Innovation Solution
A system and method that calibrate processor power measurements by executing a test routine to determine impedance and current, using a calibration factor to correct real-time current measurements, thereby increasing accuracy and enabling more efficient power management in data centers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If lossy current sensors are used for power measurement, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent applies preliminary action by performing calibration measurements before actual power monitoring. A calibration routine is executed to establish a relationship between voltage and current measurements, creating a calibration factor that compensates for sensor inaccuracies. This pre-calibration step enables subsequent power calculations to achieve high precision even when using lossy current sensors.
Solution Approach 2:
The patent implements feedback through an iterative calibration process. The system measures voltage and current, calculates power consumption, compares it against known reference values, and adjusts the calibration factor accordingly. This feedback loop continues until the measured power matches the expected power within an acceptable tolerance, thereby compensating for sensor losses and improving measurement precision.
2Measurement precision
If calibration routines are executed to improve measurement accuracy, then measurement precision improves, but loss of time increases
Solution Approach 1:
The calibration routine is executed as a preliminary action during system initialization or setup phase, before normal power monitoring begins. By performing the time-consuming calibration process in advance, the patent ensures high measurement precision for all subsequent operations without causing time delays during actual power monitoring and workload management operations.
Solution Approach 2:
The patent applies partial action by executing the calibration routine only when necessary - such as during initial system setup, after hardware changes, or when measurement accuracy requirements change. Rather than continuously calibrating, the system performs calibration selectively, reducing time loss while maintaining measurement precision when needed.
Data Source
AI summary
A system can include a processing core to execute machine readable instructions. The system can also include a memory accessible by the processor core. The memory can include preprogrammed test data that characterizes one of an impedance of a processor and a current output to the processor during execution of a test routine. The processor can include the processing core and the one of the impedance of the processor and the current output to the processor is based on a power measurement taken during execution of a test routine. The power measurement can be taken with a current sensor that is at least one of lossy or at least about 98% accurate.