Integrated Circuit Self-Learning Operating Point Control

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current solutions for optimizing chip performance and power consumption rely on guardbanding, which is inefficient as they do not consider the entire operational state of integrated circuits, including frequency, voltage, temperature, and manufacturing variability, leading to suboptimal performance and power usage.

Innovation Solution

A system that uses health sensors to monitor the operational state of integrated circuits, generating a summary of data which is then analyzed by a control unit to adjust operating parameters such as frequency and voltage, allowing for real-time optimization based on the current operational conditions, thereby removing unnecessary guardband and enhancing performance and power efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If guardbanding is used to guarantee correct hardware operation under worst-case noise events, then reliability is improved, but performance is reduced

Engineering Contradiction:

Improvecorrect hardware operationVSAvoidperformance

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system dynamically adjusts operating parameters (frequency, voltage) based on real-time monitoring of actual noise events and operational conditions, transitioning from static guardbanding to adaptive operation. The control unit modifies parameters on-the-fly to maintain reliability while optimizing performance for current conditions.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes operational parameters (frequency, voltage, power settings) based on monitored noise events and operational state. By adjusting these parameters dynamically rather than maintaining fixed conservative settings, the system achieves both reliability and performance optimization.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If guardbanding is applied in all cases, then reliability is improved, but power consumption increases

Engineering Contradiction:

Improvehardware operation guaranteeVSAvoidpower consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system dynamically adjusts power consumption based on actual operational conditions and monitored noise events. Instead of maintaining high power settings for all cases, the control unit optimizes power usage in real-time, consuming more power only when necessary to handle actual noise events.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system monitors its own operational state and noise events, then autonomously adjusts its power consumption and operating parameters. The chip self-regulates its power usage based on actual conditions without external intervention, achieving energy efficiency while maintaining reliability.

Inventive Principle:

Principle #25Self-service

3Reliability

If voltage droop detectors or power estimation circuitry are used to proactively react to problems, then reliability is improved, but device complexity increases

Engineering Contradiction:

Improveproactive problem reactionVSAvoidcircuitry complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The control unit performs multiple functions: it monitors operational parameters, detects noise events, analyzes operational state, and adjusts operating parameters. By consolidating these functions into a single multi-functional unit rather than separate dedicated circuits for each function, the system maintains reliability while reducing overall complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system implements a feedback loop where the control unit continuously monitors operational parameters and noise events, then adjusts operating parameters based on this feedback. This closed-loop approach provides proactive problem reaction without requiring complex predictive circuitry, as the system reacts to actual measured conditions.

Inventive Principle:

Principle #23Feedback

4Device complexity

If current solutions do not take into account all effects of operational state (frequency, voltage, temperature, manufacturing variability), then device complexity is reduced, but measurement precision worsens

Engineering Contradiction:

Improveoperational state considerationVSAvoidoperational state analysis

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The control unit monitors and adjusts multiple operational parameters (frequency, voltage, temperature) simultaneously, considering their interrelationships and effects on noise susceptibility. By tracking and optimizing these parameters together rather than in isolation, the system achieves precise operational state analysis while managing complexity through integrated control.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS7962887B2Self-learning of the optimal power or performance operating point of a computer chip based on instantaneous feedback of present operating environment

Publication Date: 2011.06.14 GLOBALFOUNDRIES US INC

AI summary

Sensors on the integrated circuit are used to detect the current operating state of the chip, such as frequency, voltage, temperature characteristics, or variation in the integrated circuit manufacturing process. In response, the integrated circuit may choose to modify operational parameters (such as frequency, voltage, or power-down states) in order to dynamically and autonomously maintain an optimal performance and/or power-efficiency operational point.