Variable-Size Cache via Item Selection Vector
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional caching methods result in unnecessary data traffic and energy consumption due to caching entire blocks of predetermined size, even when not all data items within the block are accessed.
Innovation Solution
Implementing a cache system that selectively caches only some data items within a block based on usage patterns, using an item selection vector to determine which data items to load or store, thereby reducing data communication and optimizing cache storage capacity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If entire cache blocks of predetermined size are cached, then cache storage capacity is utilized, but unnecessary data traffic and energy consumption occur
Solution Approach 1:
The patent segments the cache block into individual data items that can be independently selected for caching. Instead of treating the entire cache block as a single unit, the system divides it into smaller granular elements (data items) and uses an item selection vector to identify which specific items should be cached, thereby reducing unnecessary data transfer while maintaining cache effectiveness.
Solution Approach 2:
The patent applies local quality by allowing different data items within the same cache block to have different caching statuses. The item selection vector enables selective caching where only the relevant data items (those with local quality of being accessed or needed) are transferred to cache, while other items remain in memory, optimizing both data traffic and cache performance.
2Speed
If entire cache blocks are transferred, then data access speed is maintained, but energy consumption increases
Solution Approach 1:
The patent extracts only the necessary data items from the cache block for caching, rather than transferring the entire block. The item selection vector identifies and extracts specifically the data items that need to be cached, reducing the volume of data transferred and thereby lowering energy consumption while maintaining fast access speed for the required data.
Solution Approach 2:
The patent applies partial action by transferring only the portion of the cache block that is actually needed (as indicated by the item selection vector) rather than the entire block. This partial transfer reduces energy consumption associated with data movement while ensuring that all necessary data items are available for fast access.
3Device complexity
If fixed-size cache blocks are used, then cache management is simplified, but cache storage capacity is wasted
Solution Approach 1:
The patent introduces dynamics into the cache management system by allowing the effective cache block size to vary based on the number of data items selected for caching. While the physical cache block structure remains fixed for simplicity, the logical cache block size becomes dynamic, adjusting to contain only the necessary data items identified by the item selection vector, thereby improving storage capacity utilization without significantly increasing management complexity.
Data Source
AI summary
A computing system, method and apparatus to cache a portion of a data block. A processor can access data using memory addresses in an address space. A first memory can store a block of data at a block of contiguous addresses in the space of memory address. A second memory can cache a first portion of the block of data identified by an item selection vector. For example, response to a request to cache the block of data stored in the first memory, the computing system can communicate the first portion of the block of data from the first memory to the second memory according to the item selection vector without accessing a second portion of the block of data. Thus, different data blocks in the first memory of a same size can be each cached in different cache blocks of different sizes in the second memory.