A Model and Method for Computing Resource Scoring in Supercomputing Centers

This study introduces a novel model and method for scoring computing resources in supercomputing centers, aimed at overcoming the limitations of existing rule-based, heuristic, and intelligent agent-based approaches. Traditional methods struggle with adaptability, real-time performance, and universality, which are critical in dynamic computing environments. The proposed universal computing resource scoring model facilitates intelligent resource allocation, optimizes utilization, enables automated decision-making, and manages task priorities and costs effectively. It employs a sophisticated algorithm that processes real-time data on various computing parameters such as CPU/GPU processing power, utilization, memory capacity, and network performance. This model dynamically assesses and allocates resources based on the specific requirements of computing tasks, enhancing efficiency and effectiveness. The methodology encompasses parameter collection, normalization, weight assignment, and weighted scoring, leading to scenario-specific resource recommendations. This approach addresses computational complexity and uncertainties in real-world environments, offering a scalable and adaptable solution for resource management across diverse supercomputing settings. The model's universality and flexibility mark a significant advancement in computing resource management, with broad applications in data center management, resource allocation, and task scheduling.