Fixed-Length Posit Floating-Point Encoding for Simpler Arithmetic Circuits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Operation devices that support Posit floating-point numbers are larger in circuit scale and have higher power consumption compared to those supporting typical floating-point formats like IEEE 754.
Innovation Solution
A storage medium and operation device utilizing a floating-point number data structure with a fixed combined code length, where the exponent part is encoded using variable-length coding, allowing for a fixed correspondence between digit and bit positions, reducing the need for complex shift processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Posit floating-point format with variable-length exponent coding is used, then dynamic range and numerical accuracy are improved, but circuit scale and power consumption increase significantly
Solution Approach 1:
The patent segments the floating-point number into three distinct parts: sign part, exponent part, and significand part. This segmentation allows each part to be processed independently, simplifying the circuit design. The variable-length exponent is handled through dedicated decoding logic that converts it to a fixed-length format, enabling modular processing and reducing overall circuit complexity while maintaining the benefits of variable-length encoding for improved numerical accuracy.
2Measurement precision
If Posit floating-point format with variable-length exponent coding is used, then dynamic range and numerical accuracy are improved, but power consumption increases significantly
Solution Approach 1:
The patent employs dynamic conversion of the variable-length exponent to a fixed-length exponent format during processing. This dynamic approach allows the system to adapt to different exponent lengths without requiring dedicated hardware for each possible length, thereby reducing power consumption. The conversion logic dynamically adjusts the processing based on the actual exponent length, maintaining numerical accuracy while minimizing energy usage compared to static designs that must accommodate the maximum possible exponent length.
3Adaptability or versatility
If variable-length coding is used for the exponent part, then larger dynamic range is achieved, but complex shift processes are required
Solution Approach 1:
The patent introduces a fixed-length exponent part as an intermediary between the variable-length exponent and the significand. This intermediary serves as a bridge that translates the variable-length exponent into a standardized fixed-length format, eliminating the need for complex shift processes. The fixed-length exponent acts as a mediator that simplifies the interaction between different parts of the floating-point number, maintaining the large dynamic range benefits of variable-length encoding while reducing circuit complexity.
Data Source
AI summary
A structure of floating-point number data stored in a storage medium according to an embodiment is provided with a first partial code obtained by encoding all or part of an exponent of a floating-point number using variable-length coding, and a second partial code including a significand of the floating-point number. The length of the combined code of the first partial code and the second partial code is fixed, and the end bit of the first partial code and the least significant hit of the second partial code are adjacent to each other.