Cache Prediction via Hash Indexing to Reduce Collisions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional cache systems face low prediction accuracy and high power consumption due to their inability to distinguish between different access paths for addresses with the same tag fields and their reliance on advance data access for computation acceleration, which leads to collisions and inefficient power usage.

Innovation Solution

A cache system that utilizes a hashing module to generate a hash value from a target address, using this hash value as an index to store or retrieve way information in a prediction table, thereby improving prediction accuracy and reducing power consumption by avoiding collisions and accurately recording access paths for each address.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the way prediction table utilizes tag fields of addresses as index, then the cache system can store way information, but addresses with same tag fields collide and prediction accuracy decreases

Engineering Contradiction:

Improveprediction accuracyVSAvoidaddress mapping complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the address fields into tag field and index field, using the index field to access the prediction table while the tag field is used for cache way selection. This segmentation allows different addresses with same tag to have different prediction entries through unique index values.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary prediction table structure that mediates between the address and cache way selection. The prediction table stores predicted ways indexed by address fields, acting as an intermediary layer that resolves collisions before final cache access.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Speed

If advance data access is performed to accelerate computation, then execution speed improves, but power consumption increases due to unnecessary accesses

Engineering Contradiction:

Improvecomputation execution speedVSAvoidpower consumption

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The patent implements feedback mechanisms where the prediction table stores historical access patterns and uses them to guide future advance data accesses. The system learns from actual access behavior and adjusts prediction to avoid unnecessary advances, reducing power consumption while maintaining speed benefits.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent changes the parameter used for prediction from simple tag field to a combination of tag field and index field, enabling more precise prediction that adapts to actual access patterns. This parameter change allows the system to distinguish between useful and useless advance accesses.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS8180965B2System and method for cache access prediction

Publication Date: 2012.05.15 REALTEK SEMICON CORP

AI summary

A cache system includes a cache having a plurality of cache units, a prediction table and a hashing module. The prediction table is utilized to store way information of at least one cache unit corresponding to at least one accessing address, and the hashing module generates a hashing value corresponding to a target accessing address and reads way information from the prediction table or writes the way information to the prediction table by using the hashing value as an index.