Compressed Domain Processing for Direct Operations on Encoded Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data compression methods are either computationally expensive or fail to achieve high compression ratios with high throughput and low latency, and typically require decompression before processing.
Innovation Solution
The development of Compressed Domain Processors (CDPs) that enable real-time operations on compressed data without decompression, using techniques like SigBits, SigBytes, and Residue Number Systems (RNS) for efficient encoding and decoding, allowing for high compression ratios, high throughput, and low latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional data compression methods are used, then compression ratio is improved, but processing speed and throughput deteriorate due to decompression requirements
Solution Approach 1:
The patent introduces compressed domain processors as an intermediary system that operates on compressed data representations without requiring full decompression. These processors serve as a mediator between compressed data storage and processing operations, enabling direct manipulation of compressed data through specialized instructions that interpret and operate on the compressed format natively.
Solution Approach 2:
The patent segments the processing architecture into distinct components: conventional processors for control and uncompressed operations, and compressed domain processors for operating directly on compressed data. This segmentation allows different parts of the system to operate in their optimal modes, with the compressed domain processor handling data manipulation tasks while maintaining compression benefits.
2Quantity of substance
If conventional compression methods are used, then data size is reduced, but energy consumption increases due to decompression and recompression operations
Solution Approach 1:
The compressed domain processor acts as an intermediary that eliminates the energy-intensive decompression-recompression cycle. By providing specialized hardware or software instructions that operate directly on compressed data representations, it mediates between storage and processing needs without requiring full data expansion, thereby significantly reducing energy consumption.
Solution Approach 2:
The patent replaces the mechanical decompression-compression process with a more efficient computational approach. Instead of physically expanding and compressing data through sequential processing steps, the system substitutes this with direct operations on compressed representations using specialized processor instructions, eliminating redundant mechanical operations.
3Productivity
If compressed domain processing is implemented, then processing efficiency is improved, but device complexity increases
Solution Approach 1:
The compressed domain processor is designed with multi-functionality to handle various compressed data formats and operations through a unified architecture. By creating a universal processing unit that can operate on different compression schemes and perform multiple types of data manipulation, the patent reduces the need for separate specialized hardware for each compression format, thereby managing complexity while maintaining versatility.
Solution Approach 2:
The patent implements compressed domain processing capabilities as software instructions or microcode that can be copied and executed on existing processor architectures. This approach allows the complex processing logic to be replicated through software layers rather than requiring complete hardware redesign, enabling efficient compression operations while maintaining compatibility with existing processor structures.
Data Source
AI summary
Compressed domain processors configured to perform operations on data compressed in a format that preserves order. The Compressed domain processors may include operations such as addition, subtraction, multiplication, division, sorting, and searching. In some cases, compression engines for compressing the data into the desired formats are provided.