Skip to content

Latest commit

 

History

History
52 lines (38 loc) · 1.84 KB

README.md

File metadata and controls

52 lines (38 loc) · 1.84 KB

⚠️⚠️⚠️WARNING: This project will no longer be updated. If you’re interested, feel free to fork it.

SlurmSlim 💵

Optimize Slurm Job Scheduling with Intelligent Memory Estimation

Overview

SlurmSlim is a lightweight and efficient tool designed to optimize job scheduling in Slurm by accurat ely estimating the required memory for scripts and programs. By leveraging Model Context Protocol (MCP), LLM models, file sizes, and system information, SlurmSlim helps reduce computing costs by preventing over-allocated memory requests.

Features

  • ✅ Intelligent Memory Estimation – Uses MCP and LLM models to predict the optimal memory allocation.
  • ✅ Cost Reduction – Prevents excessive memory requests, lowering overall compute costs.
  • ✅ File & System-Aware – Considers file sizes and system specs for precise estimation.
  • ✅ Lightweight & Fast – Designed for efficiency with minimal overhead.

Installation

git clone https://github.com/JianYang-Lab/SlurmSlim.git
cd SlurmSlim
uv sync  # If applicable

Usage

uv run client.py server.py

Example Output

Estimated Memory: 8.2 GB
Suggested Slurm Command: sbatch --mem=8500M job_script.sh

Why Use SlurmSlim?

  • 🔹 Saves Money – No more over-provisioning, reducing unnecessary cloud or HPC costs.
  • 🔹 Improves Efficiency – Ensures jobs run smoothly without excessive memory requests.
  • 🔹 Seamless Integration – Works directly with Slurm job scripts and scheduling workflows.

Future Work

  • Extend support for CPU & GPU resource optimization
  • Integration with other job schedulers (e.g., PBS, LSF)
  • Advanced machine learning models for prediction

License

📜 MIT License

Contributors