Floating-Point Adder Segmentation for Mixed-Sign Hardware Constraints

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In designing hardware for floating-point addition in constrained devices like mobile devices, there is a trade-off between power consumption, processing performance, and size, where improvements in one factor often result in worsening of the other factors, and existing adder circuits do not efficiently handle sets of floating-point numbers with mixed signs.

Innovation Solution

The method involves identifying and adding together floating-point numbers with the same sign using same-sign adders, which are implemented in fixed function circuitry, and then using mixed-sign adders for final summation, optimizing the circuitry to reduce power consumption and silicon area while maintaining processing performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a hardware implementation of floating-point addition is used to improve processing efficiency, then power consumption and latency are reduced, but the hardware size increases

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidhardware size
Core Design Contradiction:
ProductivityVSArea of stationary object

Solution Approach 1:

The adder circuit is segmented into two specialized functional units: a same-sign adder for adding numbers with identical signs and a mixed-sign adder for adding numbers with different signs. This segmentation allows each unit to be optimized for its specific function, reducing the overall hardware complexity compared to a general-purpose adder while maintaining high processing efficiency

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The circuit performs preliminary classification of input numbers by sign before addition. The classification logic identifies whether inputs have the same or different signs and routes them to the appropriate adder unit, enabling optimized processing paths to be activated in advance and reducing the computational overhead during the addition operation

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If a general-purpose floating-point adder is used to handle all cases, then versatility is maintained, but hardware complexity and size increase

Engineering Contradiction:
Improvehandling capabilityVSAvoidcircuit complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system dynamically selects between different addition strategies based on the sign relationship of input numbers. The classification logic adaptively routes operations to the most appropriate adder unit, allowing the circuit to optimize its behavior for each specific case while maintaining the ability to handle all possible input combinations

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The invention changes the operational parameters of the adder circuit based on the sign bits of the input numbers. By detecting sign parameters and adjusting the circuit configuration accordingly (routing to same-sign or mixed-sign adder), the system achieves versatility without requiring a single complex general-purpose adder design

Inventive Principle:
Principle #35Parameter changes

3Use of energy by stationary object

If circuitry is optimized for same-sign addition only, then power consumption and area are reduced, but the ability to handle mixed-sign operations is lost

Engineering Contradiction:
Improvepower consumptionVSAvoidsign handling capability
Core Design Contradiction:
Use of energy by stationary objectVSAdaptability or versatility

Solution Approach 1:

The classification logic unit serves multiple functions: it identifies same-sign pairs, identifies mixed-sign pairs, and routes inputs to the appropriate adder unit. This multi-functionality allows the system to maintain comprehensive sign handling capability while using specialized, energy-efficient adder units for each specific operation type

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3767454B1Apparatus and method for processing floating-point numbers
Publication Date: 2021.06.23 IMAGINATION TECH LTD
  • EP3767454B1 patent drawingFigure 1~2
  • EP3767454B1 patent drawingFigure 3
  • EP3767454B1 patent drawingFigure 4

AI summary

Adder circuits and associated methods are disclosed, for processing a set of at least three floating-point numbers to be added together. The method comprises identifying (606), from among the at least three numbers, at least two numbers that have the same sign - that is, at least two numbers that are both positive or both negative. The identified at least two numbers are added together (608) using one or more same-sign floating-point adders (120, 220a, 320, 420). A same-sign floating-point adder comprises circuitry configured to add together floating-point numbers having the same sign and does not include circuitry configured to add together numbers having different signs.