Floating-Point Exceptional Value Encoding with Partial Exponent Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional encoding of exceptional values in floating point number formats requires reading all exponent bits, leading to increased circuit complexity, power consumption, and delay in floating point unit (FPU) operations.
Innovation Solution
A normalized floating point format is introduced, where exceptional values are encoded using a subset of exponent bits within an unused range, allowing detection without reading all exponent bits, and optionally using additional bits for type indication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional encoding of exceptional values is used (requiring all exponent bits to be read), then reliability of exceptional value detection is maintained, but device complexity and power consumption increase
Solution Approach 1:
The patent extracts the exceptional value indication from the full exponent field by using only a subset of exponent bits (specifically, the most significant bits) to encode exceptional values. This extraction principle reduces the number of bits that need to be read and processed, thereby reducing circuit complexity and power consumption while maintaining reliable detection of exceptional values such as infinities and NaNs.
2Measurement precision
If all exponent bits are read to detect exceptional values, then measurement precision of exceptional value detection is ensured, but loss of time increases due to longer processing delay
Solution Approach 1:
The patent applies the extraction principle by removing the requirement to read all exponent bits and instead using only a subset of the most significant exponent bits to detect exceptional values. This reduction in the number of bits to be read directly decreases the processing delay while maintaining detection precision, as the most significant bits contain sufficient information to identify exceptional values.
3Adaptability or versatility
If a normalized floating point format with extended exponent field is used, then adaptability for representing normal and subnormal values on common exponential scale is improved, but device complexity increases
Solution Approach 1:
The patent applies the universality principle by designing the normalized floating point format to handle multiple types of values (normal values, subnormal values, and exceptional values) using a unified encoding scheme. The extended exponent field with e+1 bits provides a common exponential scale that can represent all these value types without requiring separate handling logic, thereby improving adaptability while the subset encoding for exceptional values helps mitigate the complexity increase.
Data Source
Figure 1
Figure 2~3A
Figure 4~5
AI summary
The discloses relates to a processor which is able to re-encode a normal value of a first floating point format to a normalized floating point format, and re-encode a subnormal value of the first floating point format to the normalized floating point format. According to the disclosure, a subset of the exponent bits of the exponent field of a normalized floating point format is used to encode a predetermined indication of being an exceptional value, wherein a combination of bit values used to encode this predetermined indication does not occur within a range of the exponent field of the normalized floating point format used to represent any possible non-exceptional, normal and subnormal values.