Cache Memory Bandwidth Optimization via Reference Size Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing devices face inefficiencies in memory bandwidth usage, leading to increased power consumption when accessing memory for data processing, particularly due to the need for error detection and correction using parity information.
Innovation Solution
Implementing a computing device with a cache memory system that caches data and parity information, allowing for efficient bandwidth utilization by storing and outputting data in reference sizes matching the main memory's communication units, thereby reducing the need for excessive bandwidth usage during write and read operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the computing device accesses the main memory for data processing, then data processing capability is improved, but power consumption increases due to increased bandwidth usage
Solution Approach 1:
The patent divides the memory system into two segments: a main memory for storing data and parity information, and a cache memory for caching frequently accessed data. This segmentation allows the processor to access cached data without utilizing the main memory bandwidth, thereby reducing power consumption while maintaining data processing capability.
Solution Approach 2:
The cache memory pre-stores (caches) data that is likely to be accessed by the processor before actual access occurs. This preliminary action enables the processor to retrieve data from the fast cache memory instead of accessing the main memory, reducing bandwidth usage and power consumption during data processing operations.
2Reliability
If the computing device uses parity information for error detection, then data reliability is improved, but bandwidth usage increases
Solution Approach 1:
The patent combines data and parity information into a single memory structure, allowing the cache memory to cache both simultaneously. This merging enables the system to verify data integrity through parity checking without requiring separate bandwidth for parity operations, as both data and parity are readily available in the cache.
Solution Approach 2:
The cache memory acts as an intermediary between the processor and main memory, storing both data and parity information locally. This intermediary structure allows error detection and correction operations to be performed using cached parity information without accessing the main memory, thereby reducing bandwidth usage while maintaining data reliability.
Data Source
AI summary
A computing device includes a main memory, a processor, and a cache. The main memory stores data and parity, for checking an error of the data, and sends and receives the data and parity with a reference size. The processor accesses the main memory, and the cache memory caches the data. If the processor requests a write operation for current data, the current data are stored to the cache memory and the cache memory changes the stored current data to the reference size and outputs the current data changed to the reference size to the main memory. A size of the current data is smaller than the reference size.


