Chip Internal Voltage Prediction Model for Early Load Spike Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to timely detect abnormal chip internal voltage changes caused by sudden increases in load, leading to functional failures due to voltage drops, as interventions are often made too late.
Innovation Solution
A method for generating a chip internal voltage prediction model using load and characterization data sets, calibrated and trained to predict voltage changes, allowing for early detection and timely intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If voltage monitoring is performed at multiple internal nodes of the chip, then voltage prediction accuracy is improved, but power consumption increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing voltage characteristics for multiple internal nodes during chip design. The voltage prediction model is built in advance using simulated data from these nodes, allowing the system to predict voltages without performing real-time measurements or calculations during chip operation, thus avoiding the power consumption associated with active monitoring.
Solution Approach 2:
The patent uses copying by creating a virtual voltage prediction model that replicates the behavior of multiple internal nodes without physically monitoring them. The model copies the voltage characteristics through simulation and storage, enabling prediction of node voltages without the need for actual voltage sensing hardware or continuous measurement operations that would consume power.
2Measurement precision
If voltage monitoring is performed at multiple internal nodes, then voltage prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing voltage characteristics for multiple internal nodes during chip design. The voltage prediction model is built in advance using simulated data from these nodes, allowing the system to predict voltages without performing real-time measurements or calculations during chip operation, thus avoiding the power consumption associated with active monitoring.
Solution Approach 2:
The patent uses copying by creating a virtual voltage prediction model that replicates the behavior of multiple internal nodes without physically monitoring them. The model copies the voltage characteristics through simulation and storage, enabling prediction of node voltages without the need for actual voltage sensing hardware or continuous measurement operations that would consume power.
3Measurement precision
If voltage monitoring is performed at multiple internal nodes, then voltage prediction accuracy is improved, but design time increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing voltage characteristics for multiple internal nodes during chip design. The voltage prediction model is built in advance using simulated data from these nodes, allowing the system to predict voltages without performing real-time measurements or calculations during chip operation, thus avoiding the power consumption associated with active monitoring.
Solution Approach 2:
The patent uses copying by creating a virtual voltage prediction model that replicates the behavior of multiple internal nodes without physically monitoring them. The model copies the voltage characteristics through simulation and storage, enabling prediction of node voltages without the need for actual voltage sensing hardware or continuous measurement operations that would consume power.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Embodiments of the present disclosure provide a generation method for a chip internal voltage prediction model, a prediction method, and related apparatuses. The generation method for the chip internal voltage prediction model includes: obtaining a load data set and a relevant index data set of a chip respectively; calibrating the relevant index data set according to a calibration cycle number and obtaining a calibrated relevant index data set; and training the chip internal voltage prediction model based on the load data set and the calibrated relevant index data set, and obtaining the chip internal voltage prediction model with completed training. The method provided by the embodiments of the present disclosure can predict the situation of the chip internal voltage in advance, so that timely intervention can be made in the operating status of the chip.