HPC I/O Performance Analyzer

This project analyzes HPC I/O data, generates LLM-based optimization suggestions, and creates structured CSV reports.

---

Setup

1. Clone the repository/Organize files: Ensure your project structure matches:

   agent_io/
   ├── code_v2/
   │   ├── main.py
   │   ├── agent.py
   │   ├── data_loader.py
   │   ├── llm_api.py
   │   ├── utils.py # New file
   │   ├── config.yaml
   │   └── data_v2/
   │   └── output/
   └── ...
   

2. Install dependencies:

   pip install pandas pyyaml

---

How to Run

Navigate to the code_v2 directory in your terminal.

Step 1: Generate LLM Suggestion Text Files (.txt)

Run the script for each analysis pipeline. This calls the LLM and saves its full suggestions as .txt files in output/. You must run each of these at least once:

python main.py darshan_shap
python main.py raw_darshan
python main.py shap_only