HPC-Cluster-Dokumentation hpc:scheduling https://hpc.tu-berlin.de/ 2026-04-28T16:07:28+00:00 HPC-Cluster-Dokumentation https://hpc.tu-berlin.de/ https://hpc.tu-berlin.de/lib/tpl/bootstrap3/images/favicon.ico text/html 2025-03-24T11:08:12+00:00 Anonymous (anonymous@undisclosed.example.com) Access to Cluster https://hpc.tu-berlin.de/doku.php?id=hpc:scheduling:access&rev=1742810892&do=diff Access to Cluster The following groups are eligible for scientific computing on the HPC cluster: * Scientists at the TU Berlin * Students writing their master's/bachelor's thesis at the TU Berlin The HPC cluster is not intended for personal projects or educational purposes (e.g. exercises, labs). Use of the cluster is therefore linked to a chair or research project at the TU Berlin. text/html 2025-01-16T15:31:18+00:00 Anonymous (anonymous@undisclosed.example.com) Fairshare https://hpc.tu-berlin.de/doku.php?id=hpc:scheduling:fairshare&rev=1737037878&do=diff Fairshare Slurm Fairshare Policy is a feature of the Slurm Workload Manager that allows for the fair distribution of computing resources among different users or groups based on their fairshare score. The fairshare score is calculated based on factors such as past resource usage, job priorities, and user/group weights. text/html 2025-01-31T17:16:59+00:00 Anonymous (anonymous@undisclosed.example.com) Profiling and Resource Monitoring https://hpc.tu-berlin.de/doku.php?id=hpc:scheduling:profiling_and_resource_monitoring&rev=1738340219&do=diff Profiling and Resource Monitoring Introduction This documentation is designed to provide a comprehensive guide on monitoring and managing resources on our cluster. It covers the use of profiling tools, strategies for monitoring memory, and best practices for resource allocation to optimize job scheduling and performance. By following these guidelines, users can enhance their workflow efficiency and overall experience on the cluster. text/html 2024-10-25T15:03:55+00:00 Anonymous (anonymous@undisclosed.example.com) Slurm Commands https://hpc.tu-berlin.de/doku.php?id=hpc:scheduling:slurm_commands&rev=1729861435&do=diff Slurm Commands Start Jobs srun This is the simplest way to run a job on a cluster. Initiate parallel job steps within a job or start an interactive job (with --pty). salloc Request interactive jobs/allocations. When the job is started a shell (or other program specified on the command line) it is started on the submission host (Frontend). From this shell you should use srun to interactively start a parallel applications. The allocation is released when the user exits the shell. text/html 2025-01-16T15:32:52+00:00 Anonymous (anonymous@undisclosed.example.com) Scheduling https://hpc.tu-berlin.de/doku.php?id=hpc:scheduling:start&rev=1737037972&do=diff Scheduling Index scheduling index Introduction Job scheduling is done by Slurm Workload Manager.