Automating HPC Software Compilation, Deployment, and Error Resolution through an LLM-based Multi-Agent System

High-performance computing (HPC) systems rely on complex software configurations that are traditionally managed through manual processes, leading to inefficiencies and increased risk of errors.In this paper, we present an LLM-based multi-agent system designed to automate the compilation, deployment, and error resolution of HPC software.Our approach leverages state-of-the-art language models to generate, refine, and iteratively improve build scripts, thereby reducing reliance on static documentation and manual intervention.Extensive experiments on a benchmark cluster demonstrate a 97% success rate in autonomously building over 200 commonly used HPC software packages.Moreover, our agents effectively identify and rectify build and test failures and provide detailed descriptions with actionable recommendations when manual intervention is required.This advancement significantly reduces administrative overhead, enabling HPC administrators to reallocate resources to other critical tasks and paving the way for more resilient and efficient system operations.