RNS Arithmetic with Dynamic Precision for Lower Power

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Computing devices waste power due to performing unnecessarily precise arithmetic operations, which can be mitigated by dynamically modifying precision to extend battery life.

Innovation Solution

Convert operands to a redundant number system (RNS) and perform arithmetic operations from most significant bit (MSB) to least significant bit (LSB), enabling dynamic precision adjustment based on data type, battery level, and target accuracy, and selectively enabling hardware components to reduce power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If arithmetic operations are performed at higher precision, then computational accuracy is improved, but power consumption increases

Engineering Contradiction:
Improvecomputational accuracyVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent implements dynamic precision adjustment where the arithmetic logic unit can change its operating precision based on runtime conditions such as battery level, data type requirements, and target accuracy specifications. This allows the system to transition between different precision modes (e.g., full precision, half precision, quarter precision) dynamically, optimizing the balance between computational accuracy and power consumption for different operational scenarios.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The invention changes the precision parameter of arithmetic operations based on various conditions including data type characteristics, statistical properties of the data, battery charge levels, and required target accuracy. By adjusting this critical parameter dynamically, the system achieves optimal power efficiency while maintaining sufficient computational accuracy for each specific task.

Inventive Principle:
Principle #35Parameter changes

2Use of energy by moving object

If arithmetic operations are performed at lower precision, then power consumption is reduced, but computational accuracy deteriorates

Engineering Contradiction:
Improvepower consumptionVSAvoidcomputational accuracy
Core Design Contradiction:
Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The system adjusts the precision parameter downward when conditions allow (such as when battery level is low or when the application can tolerate reduced accuracy), thereby reducing power consumption while maintaining acceptable computational accuracy through intelligent parameter selection based on runtime conditions.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The arithmetic logic unit dynamically adapts its precision level based on real-time conditions, switching between different precision modes to optimize the trade-off between power consumption and computational accuracy, ensuring the system operates efficiently across varying power availability and accuracy requirements.

Inventive Principle:
Principle #15Dynamics

3Productivity

If precision is determined at compile time, then code execution efficiency is improved, but adaptability to changing power conditions deteriorates

Engineering Contradiction:
Improvecode execution efficiencyVSAvoidadaptability to power conditions
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent enables dynamic precision determination at runtime rather than static determination at compile time. The system can respond to changing power conditions, battery levels, and application requirements by adjusting precision settings during execution, thereby achieving both efficient code execution and high adaptability to varying operational conditions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system incorporates feedback mechanisms that monitor runtime conditions such as battery charge level, data characteristics, and performance requirements, then use this feedback to dynamically adjust precision settings. This closed-loop approach allows the system to optimize precision dynamically while maintaining efficient code execution.

Inventive Principle:
Principle #23Feedback

4Measurement precision

If full precision arithmetic operations are always performed, then computational accuracy is maintained, but operational time interval of battery is reduced

Engineering Contradiction:
Improvecomputational accuracyVSAvoidoperational time interval
Core Design Contradiction:
Measurement precisionVSDuration of action of moving object

Solution Approach 1:

The arithmetic logic unit dynamically adjusts its precision level based on battery charge level and application requirements, extending the operational time interval by operating at reduced precision when appropriate while maintaining sufficient accuracy for the given task, thereby optimizing battery utilization.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the precision parameter based on battery status and computational requirements, allowing the device to operate longer by using lower precision when full precision is not needed, thus extending the operational time interval while maintaining acceptable computational accuracy.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3529696B1Dynamic variable precision computation
Publication Date: 2025.08.13 ADVANCED MICRO DEVICES INC
  • EP3529696B1 patent drawingFigure 1
  • EP3529696B1 patent drawingFigure 2
  • EP3529696B1 patent drawingFigure 3

AI summary

A conversion unit [115] converts operands [110] from a conventional number system that represents each binary number in the operands as one bit to redundant number system (RNS) operands [210, 215] that represent each binary number as a plurality of bits. An arithmetic logic unit [205] performs an arithmetic operation on the RNS operands in a direction from a most significant bit (MSB) to a least significant bit (LSB). The arithmetic logic unit stops performing the arithmetic operation prior to performing the arithmetic operation on a target binary number indicated by a dynamic precision associated with the RNS operands. In some cases, a power supply [330] provides power to bit slices [311, 312, 313, 314, 315] in the arithmetic logic unit and a clock signal generator [335] provides clock signals to the bit slices.