Slurm Service for Science

The content of the website is under development

Introduction

The popularity of the HUN-REN Cloud has grown steadily in recent years, with GPU resources receiving particularly high interest. As a side effect of this positive trend, by autumn 2024, the allocatable GPU resources have essentially been exhausted, necessitating a transformation of our previously flexible resource management system. The goal is to introduce a fair GPU capacity allocation system, that makes cloud services accessible to more users annually and results in more efficient GPU resource utilization through the SLURM Job manager.

Key Features

SLURM (Simple Linux Utility for Resource Management) is a highly scalable and modular workload management system. It provides job scheduling and job state monitoring capabilities across compute nodes, with the ability to dynamically allocate resources to users for specific time periods. Users can define and send their computationally heavy jobs against the cluster, which then schedules and executes them in a FIFO logic. It also provides features for high-performance computing environments, including support for GPU scheduling and the ability to handle complex job dependencies. The current configuration supports the utilization of computational resources (GPU) for seven days.

Accessing the Service

Access to the service can be requested during the initial period at info@science-cloud.hu.

---