Dynamic Cache Fill Adjustment for Image Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing motion image decoding technologies face inefficiencies in bandwidth reduction and cache memory utilization due to fixed cache configurations that do not adapt to the varying features of the stream, leading to unnecessary data reads and increased power consumption.

Innovation Solution

An image processing device that dynamically adjusts the cache fill size based on the feature analysis of the target image, such as intra macro block ratio or motion vector variance, to optimize cache memory utilization and reduce unnecessary data reads.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If a large capacity memory is used to hold image data of decoded pictures, then the bandwidth for accessing memory increases, but the power consumption and cost increase

Engineering Contradiction:
Improvememory capacityVSAvoidpower consumption
Core Design Contradiction:
Quantity of substanceVSUse of energy by moving object

Solution Approach 1:

The patent divides the memory system into two segments: a large capacity main memory for storing decoded picture data and a smaller cache memory for storing reference image data. This segmentation allows the system to reduce the bandwidth requirement for the main memory while maintaining high access speed for reference images through the cache memory, thereby reducing power consumption without sacrificing memory capacity.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If a fixed cache configuration is used, then the device complexity is reduced, but the cache hit ratio decreases due to varying stream features

Engineering Contradiction:
Improvecache configuration complexityVSAvoidcache hit ratio
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent implements a dynamic cache configuration system that automatically adjusts cache parameters (such as cache size, block size, and replacement policy) based on the analysis of stream features including intra macro block ratio and motion vector variance. This dynamic adaptation allows the cache to optimize its performance for different video content characteristics, maintaining high cache hit ratios without requiring manual intervention or complex fixed configurations.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes cache parameters dynamically based on extracted feature amounts from the video stream. When the intra macro block ratio or motion vector variance exceeds certain thresholds, the cache configuration parameters are adjusted accordingly. This parameter change mechanism enables the cache to adapt to varying video content features, improving cache hit ratio while keeping the overall device complexity manageable through automated control.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If cache fill size is increased to improve cache hit ratio, then more unnecessary data is read, but the power consumption increases

Engineering Contradiction:
Improvecache hit ratioVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent dynamically adjusts the cache fill size parameter based on the extracted feature amounts from the video stream. When motion vector variance is low or intra macro block ratio is high, indicating limited reference image usage, the cache fill size is reduced. Conversely, when motion vector variance is high indicating extensive reference image usage, the cache fill size is increased. This parameter adaptation ensures that the cache fills only the necessary amount of data, improving cache hit ratio while minimizing unnecessary data reads and reducing power consumption.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3051816B1Cache fill in an image processing device
Publication Date: 2019.04.10 RENESAS ELECTRONICS CORP
  • EP3051816B1 patent drawingFigure 1
  • EP3051816B1 patent drawingFigure 2
  • EP3051816B1 patent drawingFigure 3~4

AI summary

In an image processing device, a motion image decoding processing unit extracts a feature amount of a target image to be decoded from an input stream, and changes a read size of a cache fill from an external memory to a cache memory, based on the feature amount. The feature amount represents an intra macro block ratio in, for example, one picture (frames or fields), or a motion vector variation. When the intra macro block ratio is high, the read size of the cache fill is decreased.