VibeCodeHPC: An Agent-Based Iterative Prompting Auto-Tuner for HPC Code Generation Using LLMs
Graduate School of Informatics, Nagoya University, Aichi, Japan
arXiv:2510.00031 [cs.SE], (26 Sep 2025)
@misc{hayashi2025vibecodehpcagentbasediterativeprompting,
title={VibeCodeHPC: An Agent-Based Iterative Prompting Auto-Tuner for HPC Code Generation Using LLMs},
author={Shun-ichiro Hayashi and Koki Morita and Daichi Mukunoki and Tetsuya Hoshino and Takahiro Katagiri},
year={2025},
eprint={2510.00031},
archivePrefix={arXiv},
primaryClass={cs.SE},
url={https://arxiv.org/abs/2510.00031}
}
We propose VibeCodeHPC, an automatic tuning system for HPC programs based on multi-agent LLMs for code generation. VibeCodeHPC tunes programs through multi-agent role allocation and iterative prompt refinement. We describe the system configuration with four roles: Project Manager (PM), System Engineer (SE), Programmer (PG), and Continuous Delivery (CD). We introduce dynamic agent deployment and activity monitoring functions to facilitate effective multi-agent collaboration. In our case study, we convert and optimize CPU-based matrix-matrix multiplication code written in C to GPU code using CUDA. The multi-agent configuration of VibeCodeHPC achieved higher-quality code generation per unit time compared to a solo-agent configuration. Additionally, the dynamic agent deployment and activity monitoring capabilities facilitated more effective identification of requirement violations and other issues.
October 5, 2025 by hgpu
Your response
You must be logged in to post a comment.
