Floating Point Multiplier-Accumulator with Dynamic Precision Adjustment

Solution Overview

Problem

Existing integer architectures face challenges in performing scalable and efficient dot product computations between large floating-point multiplicands, particularly due to the high complexity and power consumption of signed floating-point multiplication operations, and the need for precise accumulation without requiring full precision on intermediate results.

Innovation Solution

A pipelined floating-point multiplier-accumulator system with sign, exponent, and mantissa processors that dynamically adjusts precision based on exponent differences, using a maximum exponent finder and Pad, Complement, and Shift stages to optimize multiplication and accumulation operations, allowing for variable precision and reduced power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If full precision is used for all intermediate accumulation results, then measurement precision is improved, but device complexity and power consumption increase

Engineering Contradiction:

Improveaccumulation precisionVSAvoidarchitecture complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements dynamic precision adjustment by monitoring the exponent difference between the accumulated result and incoming multiplicands. When the exponent difference exceeds a threshold, the system automatically reduces the number of bits used in accumulation operations. This dynamic adaptation allows the architecture to maintain full precision when needed while reducing complexity and power consumption when high precision is not required, directly resolving the contradiction between measurement precision and device complexity.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the precision parameter (number of bits) of the accumulator based on the exponent difference between input values. By adjusting the accumulation precision parameter dynamically rather than using fixed full precision, the patent reduces device complexity and power consumption while maintaining sufficient precision for the actual computational needs, thereby resolving the contradiction between precision and complexity.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If full precision is used for all intermediate accumulation results, then measurement precision is improved, but power consumption increases

Engineering Contradiction:

Improveaccumulation precisionVSAvoidpower consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent implements dynamic precision adjustment by monitoring the exponent difference between the accumulated result and incoming multiplicands. When the exponent difference exceeds a threshold, the system automatically reduces the number of bits used in accumulation operations. This dynamic adaptation allows the architecture to maintain full precision when needed while reducing complexity and power consumption when high precision is not required, directly resolving the contradiction between measurement precision and device complexity.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the precision parameter (number of bits) of the accumulator based on the exponent difference between input values. By adjusting the accumulation precision parameter dynamically rather than using fixed full precision, the patent reduces device complexity and power consumption while maintaining sufficient precision for the actual computational needs, thereby resolving the contradiction between precision and complexity.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If high precision multiplication is performed, then measurement precision is improved, but speed decreases due to increased complexity

Engineering Contradiction:

Improvemultiplication precisionVSAvoidcomputation speed

Core Design Contradiction:

Measurement precisionVSSpeed

Solution Approach 1:

The patent implements dynamic precision adjustment by monitoring the exponent difference between the accumulated result and incoming multiplicands. When the exponent difference exceeds a threshold, the system automatically reduces the number of bits used in accumulation operations. This dynamic adaptation allows the architecture to maintain full precision when needed while reducing complexity and power consumption when high precision is not required, directly resolving the contradiction between measurement precision and device complexity.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS12106069B2Power saving floating point multiplier-accumulator with precision-aware accumulation

Publication Date: 2024.10.01 CEREMORPHIC INC

AI summary

A floating point multiplier-accumulator (MAC) multiplies and accumulates N pairs of floating point values using N MAC processors operating simultaneously, each pair of values comprising an input value and a coefficient value to be multiplied and accumulated. The pairs of floating point values are simultaneously processed by the plurality of MAC processors, each of which outputs a signed integer form fraction and a maximum exponent. A range estimator forms a possible range of values from the exponent differences and determines an adder precision. The integer form fractions are summed using the adder precision, a sign bit is extracted, and a floating point value is output. Each MAC processor provides its integer form fraction with a precision determined by the MAC processor's exponent difference.