Variable Rounding Point Floating-Point Processing Circuitry

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing data processing systems for floating point numbers lack flexibility in rounding operations, particularly in specifying a variable rounding point within the significand, which limits their ability to efficiently process and convert between floating point and anchored numbers.

Innovation Solution

The system includes processing circuitry and decoder circuitry responsive to a rounding position input operand to generate control signals for processing a floating point input operand, allowing for rounding at a variable point within the significand, and supports conversion-and-add operations to align floating point numbers with anchored numbers by adjusting their bit significance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If fixed rounding position is used in floating point numbers, then processing speed is improved, but flexibility and adaptability are worsened

Engineering Contradiction:
Improveprocessing speedVSAvoidflexibility in rounding operations
Core Design Contradiction:
SpeedVSAdaptability or versatility

Solution Approach 1:

The patent implements a variable rounding point mechanism that dynamically adjusts the rounding position within the significand based on the rounding position input operand. This allows the rounding point to move between different bit positions (e.g., from bit position 0 to bit position 23 in a 24-bit significand), providing flexibility while maintaining efficient hardware implementation through controlled dynamic behavior rather than complete reconfigurability

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the rounding parameter (rounding position) as an input operand that can be specified programmatically. By allowing the rounding position to be a variable parameter rather than a fixed constant, the system achieves adaptability for different conversion scenarios (floating-point to anchored number conversions with varying bit significances) while keeping the underlying hardware structure relatively simple

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If variable rounding point is implemented, then flexibility and accuracy are improved, but device complexity is worsened

Engineering Contradiction:
Improveflexibility in rounding operationsVSAvoiddecoder and processing circuitry complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent designs a universal rounding mechanism that handles multiple conversion scenarios through a single variable rounding point facility. The same hardware circuitry supports conversions to anchored numbers with different bit significances, eliminating the need for separate dedicated rounding circuits for each conversion type and reducing overall system complexity despite the added flexibility

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

Solution Approach 2:

The rounding position input operand acts as an intermediary that bridges the gap between diverse conversion requirements and the unified processing circuitry. By encoding the desired rounding position in this input operand, the system translates various conversion scenarios into a common operational format that the processing circuitry can handle efficiently without requiring complex case-by-case handling

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If conversion-and-add operations are supported, then conversion efficiency is improved, but processing circuitry complexity is worsened

Engineering Contradiction:
Improveconversion efficiencyVSAvoidprocessing circuitry complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent merges the conversion operation and the addition operation into a single integrated conversion-and-add instruction. This combines multiple processing steps (floating-point to anchored number conversion, rounding, and addition) into one unified operation that executes in a single instruction cycle, improving productivity while managing complexity through functional integration rather than sequential separate operations

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The rounding operation is performed as a preliminary step within the conversion-and-add sequence, preparing the floating-point number for the subsequent addition operation. By pre-calculating the rounded value during the conversion phase rather than as a separate post-processing step, the system optimizes the overall conversion efficiency and reduces the total number of processing cycles required

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3341833B1Floating point number rounding
Publication Date: 2020.01.15 ARM LTD
  • EP3341833B1 patent drawingFigure 1
  • EP3341833B1 patent drawingFigure 2~4
  • EP3341833B1 patent drawingFigure 5

AI summary

Processing circuitry (2) supports execution of program instructions having a rounding position input operand so as to generate control signals (14) for controlling processing circuitry (16) to process a floating point input operand with a significand value to generate an output result which depends upon a value from rounding the floating point input operand using a variable rounding point within the significand of the floating point input operand as specified by the rounding position input operand. In this way, processing operations having as inputs floating point operands and anchored number operands may be facilitated.