Entropy-Based Cache Block Selection for Memory Throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing systems face challenges in managing cache memory efficiently, particularly in high-throughput environments like cloud computing and web servers, leading to bottlenecks and increased costs due to the inability to effectively handle high volumes of data.
Innovation Solution
A system that calculates entropy levels for storable objects and memory blocks, selects the block with the lowest entropy for erasure, and performs an erase operation to optimize cache management and improve performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional cache management methods are used, then system simplicity is maintained, but cache efficiency and throughput are limited
Solution Approach 1:
The patent transforms the cache eviction decision-making process from simple LRU/FIFO algorithms to an entropy-based selection mechanism. By calculating entropy values for each memory block based on access patterns and evicting blocks with highest entropy (most uncertain/least valuable), the system dynamically adapts to varying workload characteristics, thereby improving throughput while maintaining manageable complexity through a clear mathematical criterion
Solution Approach 2:
The system continuously monitors access patterns of stored objects and uses this feedback to calculate entropy values for memory blocks. This feedback loop enables the cache management mechanism to adapt to changing data access behaviors in real-time, optimizing cache hit rates and throughput without requiring complex manual configuration or intervention
2Productivity
If more cache memory is allocated to handle high volumes of data, then throughput capacity increases, but system cost increases
Solution Approach 1:
The cache management mechanism autonomously identifies and evicts the least valuable memory blocks based on entropy calculations without external intervention. This self-service approach ensures that the limited cache memory capacity is continuously optimized for current workload requirements, maximizing throughput capacity from available memory resources without requiring additional hardware investment
3Measurement precision
If entropy calculation is performed for all memory blocks, then cache optimization accuracy improves, but processing overhead increases
Solution Approach 1:
The system calculates entropy values for memory blocks selectively rather than continuously updating all blocks at every access event. By performing entropy calculations at appropriate intervals and only when necessary for eviction decisions, the system achieves sufficiently precise eviction accuracy while minimizing the processing overhead and energy consumption associated with continuous entropy computation across all memory blocks
Data Source
AI summary
A computing system includes: a memory storage unit, having memory blocks, configured as a memory cache to store storable objects; and a device control unit, coupled to the memory storage unit, configured to: calculate an entropy level for the storable objects based on an eviction policy; calculate a block entropy for each of the memory blocks based on the entropy level of the storable objects in the memory blocks; select an erase block from the memory blocks, wherein the erase block is an instance of the memory blocks with the lowest value of the block entropy; and perform an erase operation on the erase block.