Memory Optimization in Partitioned Systems with Limited MPU Registers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In partitioned systems with multiple memories and a limited number of registers in the memory protection unit (MPU), existing methods for memory optimization often result in local optimality, failing to achieve overall system efficiency and leading to potential data overwriting and access collisions.
Innovation Solution
A computer-implemented method that calculates run-time changes of data based on access statistics and determines optimal data placement in multiple memories, while assigning MPU registers to define memory areas and ensure secure data access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If memory optimization is performed after MPU configuration in partitioned systems, then local optimality is achieved, but overall system efficiency deteriorates and data access collisions occur
Solution Approach 1:
The patent performs memory optimization calculations before finalizing the MPU configuration. By calculating run-time changes and determining optimal data placement in advance, the system establishes a memory distribution plan that accounts for multiple tasks and security requirements, preventing data overwriting and access collisions before they occur.
Solution Approach 2:
The patent implements a feedback mechanism where access statistics are collected and analyzed to calculate run-time changes. This feedback loop allows the system to adjust memory allocation based on actual access patterns, iteratively improving both local and overall optimality while maintaining security constraints.
2Manufacturing precision
If unlimited registers are available in MPU, then global memory optimality can be achieved, but device complexity and hardware costs increase
Solution Approach 1:
The patent applies partial action by using only the necessary number of registers required for the actual memory allocation needs. Instead of allocating unlimited registers, the system calculates the optimal memory distribution first, then assigns only the minimum required registers to describe the boundaries of memory areas, avoiding unnecessary hardware complexity.
Solution Approach 2:
The patent changes the approach from hardware-centric (providing unlimited registers) to software-centric optimization. By using algorithms to calculate optimal placement and dynamically managing the limited registers, the system achieves global optimality through intelligent parameter management rather than brute-force hardware expansion.
3Reliability
If memory areas are strictly separated for security, then data protection is improved, but memory access efficiency deteriorates
Solution Approach 1:
The patent segments memory into distinct areas with clear boundaries defined by MPU registers, ensuring security isolation between different tasks and applications. Each memory area is carefully delimited to prevent unauthorized access and data overwriting, while the segmentation is optimized to minimize access overhead and maintain efficiency.
Data Source
AI summary
A computer-implemented method for optimizing the memory of a partitioned system including multiple memories, at least one processing core, and at least one memory protection unit (MPU), each MPU including multiple registers. The method includes calculating run-time changes of each piece of data of a multitude of data which are to be processed by the processing core, with the respective piece of data being placed in each memory of the multiple memories based on access statistics for the respective piece of data, each piece of data of the multitude of data being assigned to one rights area or multiple rights areas; determining a placement of the data in the memories based on the calculated run-time changes; and allocating the multiple registers of the MPU for the certain placement of the data in the multiple memories, one register of the multiple registers identifying a memory area of the multiple memories.