Knowledge Graph Computing Configuration Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI development lacks a systematic framework to consider environmental sustainability, leading to significant energy consumption and carbon emissions, particularly in the training of AI, ML, and DL models, with existing methods focusing on energy efficiency but neglecting broader environmental and social impacts.
Innovation Solution
A computer-implemented method and system that utilize a knowledge graph generated by machine learning models to determine optimal combinations of computing hardware and software based on energy consumption patterns, performing life cycle assessments to identify the most sustainable configurations for executing computing operations, incorporating multi-objective optimization techniques and life cycle assessments to minimize environmental impact.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI training is performed using high-performance computing hardware (GPUs, TPUs, FPGAs, ASICs) to increase computing capabilities, then the computing performance and training speed improve, but the energy consumption and carbon emissions increase significantly
Solution Approach 1:
The system changes multiple parameters simultaneously including hardware selection (CPU, GPU, TPU, FPGA, ASIC), software configuration (frameworks, libraries, compilers), and training parameters (batch size, learning rate, epochs) to optimize the balance between computing performance and energy consumption. This multi-parameter optimization enables achieving high computing capability while minimizing energy usage through data-driven selection of optimal configurations.
Solution Approach 2:
The system performs preliminary analysis and selection of optimal hardware-software-parameter combinations before actual AI training begins. By pre-evaluating different configurations using knowledge graphs and historical data, the system identifies the most energy-efficient setup that meets performance requirements, avoiding trial-and-error approaches that would consume additional energy and time.
2Object-affected harmful factors
If environmental sustainability considerations are integrated into AI development process, then the carbon footprint and energy waste are reduced, but the complexity of the development process increases
Solution Approach 1:
The system introduces an intermediary framework that acts as a bridge between AI development and environmental sustainability goals. This framework includes knowledge graphs that encode environmental impact data, automated evaluation metrics, and optimization algorithms that translate sustainability requirements into actionable configuration recommendations, simplifying the integration of environmental considerations into the development workflow.
Solution Approach 2:
The system implements feedback mechanisms that continuously monitor and evaluate the environmental impact of AI training configurations. By providing real-time or near-real-time feedback on carbon emissions and energy consumption, the system enables iterative optimization of configurations to reduce harmful effects while maintaining development efficiency through automated adjustment recommendations.
3Reliability
If comprehensive life cycle assessment is performed to evaluate environmental impact of AI training, then the sustainability metrics are improved, but the time and computational resources required for configuration selection increase
Solution Approach 1:
The system performs partial life cycle assessments by focusing on the most significant environmental impact factors rather than evaluating every possible parameter. By identifying and prioritizing key metrics (such as energy consumption during training, hardware manufacturing impact, and data center location effects), the system achieves reliable sustainability evaluation in a fraction of the time required for comprehensive assessments, using approximations for less critical factors.
Data Source
AI summary
A system and method for determining optimal computing configuration for executing a computing operation includes defining one or more constrains of a given computing operation to be executed. The method further includes implementing a knowledge graph to determine at least one suitable combination of computing hardware and computing software based on the given computing operation and the defined one or more constrains therefor. The method further includes quantitatively estimating an energy requirement and qualitatively estimating an energy consumption pattern of the determined at least one suitable combination. The method further includes performing a life cycle assessment for execution of the given computing operation utilizing the determined at least one suitable combination of computing hardware and computing software based on the quantitative estimation of the energy requirement and the qualitative estimation of the energy consumption pattern, to determine an optimal combination of computing hardware and computing software therefor.