Memory Power Model Using Data Address Status Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current power modeling in EDA tools for integrated circuits is not precise enough, as it relies solely on operating modes, leading to pessimistic or optimistic estimates of power consumption, especially within the same mode, due to variations in current consumption that are not accounted for.
Innovation Solution
A power consumption model is developed using characteristic vectors that combine operating mode statuses with variations in data/address port statuses, allowing for more accurate power estimation by analyzing these vectors and constructing a lookup table for power consumption based on specific conditions, such as the write enable, chip enable, and Hamming distance of access information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If power consumption is modeled by separating operating modes only, then the model structure is simple, but the power estimation precision is insufficient
Solution Approach 1:
The pin set is segmented into multiple groups (first group, second group, third group) with different status definitions. The first group pins have binary status (0 or 1), the second group pins have varying status values, and the third group pins have binary status. This segmentation allows the model to capture different types of pin behaviors and their impact on power consumption, thereby improving estimation precision while maintaining manageable model structure.
Solution Approach 2:
Different status definitions are applied to different pin groups based on their specific characteristics. The first group uses simple binary status, the second group uses variation-based status (Hamming distance), and the third group uses binary status. This local differentiation allows each pin group to be modeled with the most appropriate status type, improving overall power estimation accuracy without uniformly increasing complexity.
2Productivity
If power consumption is estimated using traditional operating mode separation, then computational resources are saved, but the estimation accuracy becomes pessimistic or optimistic
Solution Approach 1:
The model introduces Hamming distance as an additional parameter to characterize pin status variations within the same operating mode. By incorporating this parameter, the model can distinguish between different activity levels and access patterns, leading to more accurate power estimation. The status definitions are designed to capture the essential variations in power consumption without requiring exhaustive simulation of all possible states.
Data Source
AI summary
A power consumption model for a memory device is provided. According to each characteristic vector, a corresponding power lookup table is built. Each characteristic vector comprises an operating mode and a variation of the data and/or address status of the memory device.