Cache Replacement Using Preuse Distance and Tag-Age Priority
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cache data replacement policies fail to consider the preuse distance of newly incoming data, leading to cache contamination where data is discarded without being reused, and require significant hardware resources for storage.
Innovation Solution
A cache data replacement system that determines a set competition priority value, cache line recency priority value, and tag bit priority value to calculate a first reference value, using a tag-age-based policy to replace cache data considering the preuse distance of incoming data, thereby reducing hardware requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing cache data replacement policies are used, then cache data can be replaced, but cache contamination occurs where data is discarded without being reused and significant hardware resources are required
Solution Approach 1:
The patent applies preliminary action by calculating the preuse distance for incoming data before the replacement decision is made. The preuse distance is computed based on the address of the incoming data and the addresses of data currently stored in cache ways, allowing the system to predict whether the incoming data will be reused before existing cache data is discarded. This preliminary prediction prevents cache contamination by avoiding replacement of data that will be needed soon.
2Reliability
If existing cache data replacement policies are used, then cache data can be replaced, but significant hardware resources are required for storage
Solution Approach 1:
The patent extracts only the necessary information needed for replacement decisions from the cache structure. Specifically, it uses the tag bits and address information already present in the cache ways to calculate preuse distances, rather than requiring separate hardware structures to store replacement metrics. This extraction approach reduces hardware area while maintaining effective data replacement capability.
Solution Approach 2:
The patent makes the existing cache structure multi-functional by using the same hardware components (tag arrays, address decoders) to serve both data storage and replacement decision-making functions. The preuse distance calculation leverages existing address comparison logic, eliminating the need for dedicated hardware resources solely for replacement policy implementation.
3Productivity
If cache data is replaced without considering preuse distance, then replacement can be performed, but cache contamination increases and data is discarded without being reused
Solution Approach 1:
The patent implements feedback by continuously monitoring access patterns and using the preuse distance calculation to adjust replacement decisions. The system feeds back information about incoming data addresses and their potential reuse timing into the replacement logic, allowing dynamic optimization of which data to discard. This feedback mechanism ensures that data with high likelihood of imminent reuse is preserved, improving both replacement speed and data reuse efficiency.
Data Source
AI summary
In an example method of replacing data in a cache memory, input data is received. A set competition priority value is determined based on a set-access-count-since-last-miss (SALM) counter value. A cache line recency priority value is determined based on a preuse distance value. A tag bit priority value is determined based on a first tag-bit-subset-array (TBSA) and a second TBSA. A first reference value is calculated based on the set competition priority value, the cache line recency priority value, and the tag bit priority value. A tag-age-based policy is performed based on the first reference value. The tag-age-based policy represents an operation that replaces first cache data stored in a first way among a plurality of target ways with the target data.