SoC Power Analysis via Abnormality-Triggered Logging Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional power analysis methods for System on Chip (SoC) result in inaccurate power consumption evaluations due to various reasons.
Innovation Solution
A method and system for SoC power consumption analysis that involves monitoring key parameters, detecting abnormalities, calling corresponding profiles, and forming logs to perform power consumption-based analysis, thereby improving accuracy and reducing additional power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional power analysis methods are used for SoC, then the analysis process is simple, but the measurement precision of power consumption is inaccurate
Solution Approach 1:
The patent segments the power analysis process into multiple stages: abnormality detection based on key parameters, profile selection based on detected abnormal types, and targeted log collection. This segmentation allows the system to focus resources on relevant analysis tasks, improving measurement precision while managing complexity through structured division of the analysis workflow
Solution Approach 2:
The patent performs preliminary abnormality detection using key parameters before conducting full power consumption analysis. By detecting abnormal types in advance and selecting corresponding profiles, the system prepares the analysis framework beforehand, which improves subsequent measurement accuracy without requiring complete system complexity from the start
2Reliability
If comprehensive log collection is performed for power analysis, then the analysis coverage is complete, but the additional power consumption increases
Solution Approach 1:
The patent applies partial action by collecting logs selectively based on detected abnormal types rather than collecting all possible logs universally. The system calls specific profiles corresponding to detected abnormalities and collects only relevant parameters, ensuring sufficient analysis coverage for identified issues while avoiding the excessive power consumption of comprehensive log collection across all possible scenarios
Solution Approach 2:
The patent implements dynamic log collection by adapting the profile selection and parameter collection scope based on real-time abnormality detection results. The system dynamically adjusts what logs to collect based on the detected abnormal type, ensuring reliability for the specific issue at hand while minimizing unnecessary power consumption from collecting irrelevant data
3Adaptability or versatility
If fixed profile is used for all abnormal types, then the device complexity is low, but the adaptability to different abnormal types is poor
Solution Approach 1:
The patent creates a universal profile selection mechanism that handles multiple abnormal types through a common detection framework. The key parameter monitoring and abnormal type detection system serves as a universal entry point that adapts to different abnormal types by selecting appropriate profiles, providing multi-functionality without requiring separate fixed profiles for each scenario
Solution Approach 2:
The patent changes the parameter set being monitored and collected based on the detected abnormal type. By selecting different profiles that contain different parameter sets tailored to specific abnormal types, the system adapts its monitoring focus dynamically, achieving high adaptability while managing complexity through parameter-based profile selection rather than structural complexity
Data Source
AI summary
A method for a system on chip (SoC) power consumption analysis, the method including: monitoring a plurality of key parameters of the SoC, wherein the plurality of key parameters are related to power consumption of the SoC; detecting an abnormal type of the SoC based on the plurality of key parameters; calling a profile corresponding to the detected abnormal type among a plurality of profiles; monitoring and collecting each of parameters in a parameter set included in the called profile to form a log; and performing power consumption-based analysis on the SoC based on the log.