Floating Point Power Function Hardware Verification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Hardware designs implementing floating point power functions are challenging to verify due to their complexity, especially when using simulation-based verification methods, which are time-consuming and prone to errors, while formal verification struggles with the complexity of floating point power functions with allowable errors.

Innovation Solution

The method involves formally verifying that the hardware design is recurrent over sets of β input exponents, followed by exhaustive simulation over a reduced simulation range to verify the property, leveraging the recurrent nature of hardware designs to efficiently validate floating point power functions using a combination of formal and simulation-based verification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If simulation-based verification is used to verify hardware designs for floating point power functions, then verification can be performed, but the verification time becomes excessively long (days or weeks) and the process becomes unmanageable

Engineering Contradiction:
Improveverification completenessVSAvoidverification time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The verification process is segmented into two distinct phases: formal verification to prove recurrency properties for all possible inputs, and simulation-based verification only for a reduced set of β input exponents. This segmentation allows the majority of verification to be done efficiently through formal methods while using simulation only where necessary, dramatically reducing total verification time from days/weeks to minutes/hours.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts and proves the recurrency property formally, removing the need to simulate all possible input combinations. By taking out the recurrency verification from the simulation process and handling it separately through formal verification, the simulation scope is reduced to only β input exponents, eliminating the time-consuming exhaustive simulation of all possible inputs.

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If formal verification is used to verify hardware designs for floating point power functions, then verification completeness improves, but the complexity of verification increases due to allowable errors in floating point operations

Engineering Contradiction:
Improveverification completenessVSAvoidverification complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the recurrency property from the complex floating point power function verification and proves it separately using formal verification. This extraction simplifies the overall verification by separating the recurrency aspect (which can be proven formally) from the error bound verification (which uses simulation on a reduced set), making the verification process manageable despite floating point complexities.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the verification approach by introducing the recurrency parameter β (where β is a multiple of the reciprocal of the power function exponent). This parameter allows the verification to be performed on a reduced set of inputs while maintaining completeness, effectively transforming an intractable verification problem into a manageable one by changing how the verification space is parameterized.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If exhaustive simulation is performed to verify all possible inputs, then verification accuracy is ensured, but the number of simulation vectors becomes unmanageable and the process takes days or weeks

Engineering Contradiction:
Improveverification accuracyVSAvoidnumber of simulation vectors
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The verification is segmented into formal verification of recurrency (which covers all inputs theoretically) and simulation of a reduced set of β input exponents. This segmentation reduces the number of simulation vectors from all possible inputs to only β representatives, maintaining verification accuracy through the formal recurrency proof while dramatically reducing the simulation workload.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of simulating all possible inputs directly, the patent uses formal verification to create a proof model that copies the behavior across all inputs satisfying the recurrency property. This allows verification accuracy for all inputs to be achieved by simulating and formally proving properties for only a subset of β input exponents.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3239834B1Verification of hardware designs to implement floating point power functions
Publication Date: 2020.01.08 IMAGINATION TECH LTD
  • EP3239834B1 patent drawingFigure 1
  • EP3239834B1 patent drawingFigure 2
  • EP3239834B1 patent drawingFigure 3

AI summary

A method of exhaustively verifying a property of a hardware design to implement a floating point power function. The method includes, formally verifying that the hardware design is recurrent over sets of β input exponents, wherein β is an integer that is a multiple of the reciprocal of the exponent of the power function; and for each recurrent input range of the hardware design, exhaustively simulating the hardware design over a simulation range to verify the property is true over the simulation range, wherein the simulation range comprises only β input exponents.