Pseudo-LRU Cache Line Replacement for Skewed-Associative Caches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing least-recently-used (LRU) replacement technique for skewed-associative caches becomes prohibitively expensive in terms of area, speed, and complexity as the number of ways increases, due to the exponential scaling of entries in the search table.
Innovation Solution
A pseudo-LRU mechanism that uses a hierarchy of lookup tables, where indices generated from cache line addresses are used to determine the least-recently-used way through logical or mathematical operations, reducing the complexity and area requirements by scaling linearly with the number of ways.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional LRU replacement technique is used in skewed-associative cache, then accurate least-recently-used identification is achieved, but the search table size scales exponentially with the number of ways, leading to prohibitive area and complexity
Solution Approach 1:
The patent segments the search table into multiple smaller sub-tables, where each sub-table corresponds to a specific way in the skewed-associative cache. Instead of using a single large search table that scales exponentially with the number of ways, the system divides it into N sub-tables for N ways, where each sub-table has a manageable size. This segmentation reduces the area and complexity while maintaining accurate LRU identification through coordinated access to the appropriate sub-table based on the way selector.
2Measurement precision
If traditional LRU replacement technique is used in skewed-associative cache, then accurate least-recently-used identification is achieved, but the access time increases due to large search table traversal
Solution Approach 1:
By segmenting the search table into smaller sub-tables, the patent reduces the traversal time required to access LRU information. Each sub-table can be accessed more quickly than a single large table, and the segmented structure allows parallel or selective access to the relevant sub-table based on the way selector, thereby reducing overall access time while maintaining accurate LRU identification.
Solution Approach 2:
The patent performs preliminary organization of LRU information into way-specific sub-tables, so that when a cache access occurs, the system can directly access the pre-organized sub-table corresponding to the relevant way. This preliminary structuring eliminates the need to traverse entire large tables during cache operations, significantly reducing access time while preserving LRU accuracy.
3Quantity of substance
If the number of ways in skewed-associative cache is increased to improve cache capacity, then cache performance is improved, but the LRU replacement complexity increases exponentially
Solution Approach 1:
The patent applies segmentation by creating a separate sub-table for each way in the skewed-associative cache. When the number of ways increases to improve cache capacity, the LRU mechanism scales linearly with the number of sub-tables rather than exponentially with a single large table. Each sub-table manages LRU information for its corresponding way independently, reducing the overall complexity of the LRU replacement mechanism while maintaining the ability to handle increased cache capacity.
Data Source
AI summary
Embodiments of the present invention provide a system that replaces an entry in a least-recently-used way in a skewed-associative cache. The system starts by receiving a cache line address. The system then generates two or more indices using the cache line address. Next, the system generates two or more intermediate indices using the two or more indices. The system then uses at least one of the two or more indices or the two or more intermediate indices to perform a lookup in one or more lookup tables, wherein the lookup returns a value which identifies a least-recently-used way. Next, the system replaces the entry in the least-recently-used way.