High Dynamic Ratio Field Compression With Dynamic Bit Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing compression hardware and software are inflexible and inefficient for handling data with high dynamic ratio fields, particularly in machine learning applications, as they are designed for specific data quantum sizes and struggle with larger data formats like 16-bit and 32-bit floating point data.
Innovation Solution
The method involves dynamically mapping M-bit data units into S data fields based on parameters such as size, type, locality, and estimated compression ratios, using machine learning algorithms and directed search engines to select optimal mappings, allowing for efficient compression and decompression of high dynamic ratio fields.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing compression hardware and software are used with fixed data quantum sizes, then compression efficiency is improved for standard formats, but adaptability to high dynamic ratio fields and large data formats (16-bit and 32-bit floating point) deteriorates
Solution Approach 1:
The patent implements dynamic mapping that adapts the decomposition of M-bit data units into N-bit data fields based on the specific characteristics of the input data. The system dynamically determines the optimal number of data fields and their bit allocations rather than using fixed predetermined mappings, enabling the compression hardware to efficiently handle varying data formats including 16-bit and 32-bit floating point data with high dynamic ratio fields.
Solution Approach 2:
The system changes the parameters of data field decomposition by varying the number of data fields (S) and the bit allocation to each field based on the input data characteristics. This parameter adaptation allows the same compression hardware to optimize its operation for different data types and dynamic ranges, resolving the contradiction between fixed hardware design and variable data requirements.
2Adaptability or versatility
If M-bit data units are decomposed into multiple N-bit data fields, then compatibility with compression hardware quantum sizes is improved, but device complexity increases due to dynamic mapping requirements
Solution Approach 1:
The patent performs preliminary decomposition of M-bit data units into multiple N-bit data fields before the actual compression process. This preliminary action prepares the data in a format compatible with the compression hardware's quantum size requirements, and the decomposition strategy is determined in advance based on data characteristics, reducing the complexity during the compression execution phase.
3Productivity
If dynamic mapping is implemented to optimize compression ratios, then compression performance is improved, but processing time increases due to analysis and selection of mappings
Solution Approach 1:
The system analyzes only the most significant bits of floating point data to determine the optimal mapping configuration, rather than performing exhaustive analysis on all bits. This partial action approach captures the essential characteristics needed for mapping optimization while significantly reducing the computational time and complexity of the analysis phase.
Data Source
AI summary
Various embodiments include methods and devices for implementing compression of high dynamic ratio fields. Various embodiments may include receiving a compression block having data units, receiving a mapping for the compression block, wherein the mapping is configured to map bits of each data unit to two or more data fields to generate a first set of data fields and a second set of data fields, compressing the first set of data fields together to generate a compressed first set of data fields, and compressing the second set of data fields together to generate a compressed second set of data fields.