compute_ru

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
===== Getting started with the School of Technology computing cluster =====
\\
The cluster compute.ru.is is a linux environment, connect with SSH\\
The easy way to connect and do your work the cluster is with an editor on your local computer, for example {{:language_lab:vsapp.png?nolink&50|}}[[https://code.visualstudio.com/|Visual Studio Code]], but other editors can be used, if they can make a SSH connection to a remote host.\\

To setup and connect to the cluster with Visual Studio Code using SSH extension, see [[Compute:vscodeSSH|how to]].\\

 ==== Terminal connection :====
Open a terminal to log-in '' ssh user@compute.ru.is''\\
In Windows you can for example use [[http://www.chiark.greenend.org.uk/%7Esgtatham/putty/|Putty]] or Powershell {{:language_lab:pwrshell.jpg?nolink&50|}} to connect\\
In Mac{{:language_lab:mac.jpg?nolink&50|}} or Linux{{:language_lab:linux1.jpg?nolink&50|}} search for **terminal**\\

==== About the user environment : ====
Each user has his home folder, for student the path is [/student/<username>]\\
Staff members and researchers are on a different filesystem [/home/<username>]\\

  - The first step for users is to make sure their working environment fits their needs, might have to be reconfigure.\\ User can install different version of Python and all necessary modules for their work.\\ For those who are using Python, we recommend, for ease of use, using a [[compute:venvpython|Python virtual environment]]\\ [[Compute:begin|More detailed instruction and tips for beginners]].\\
  - Each user directory has a few files an readMe.lst file and template (slurm) script file (myJob.sh)\\ Every user needs to use a script file for instructing the cluster what to do\\
  - The cluster gives users access to memory, cpu and GPU. **Slurm** is used as workload manager/scheduler system\\ To execute your job on the cluster, first update myJob.sh to run your code (or make a copy)\\ Then submit your job to the queue with command **sbatch myJob.sh**
  - Please refrain from running heavy interactive jobs on the login node.\\
\\
The template script file looks like this\\
''#!/bin/bash\\
#SBATCH <nowiki>--</nowiki>account=student\\
#SBATCH %%--%%job-name=MyJob\\
#SBATCH %%--%%gpus-per-node=1\\
#SBATCH %%--%%mem-per-cpu=2G\\
#SBATCH %%--%%partition=basic\\
#SBATCH <nowiki>--</nowiki>output=myBatch.log\\
''
\\
This example file is in each user's home folder, to view the content use the command ''cat myJob.sh''\\

====Useful commands====
^cmd^Descr^
|squeue|Show all jobs on the queue|
|sinfo|Information about the cluster|
|sbatch <job.sh>|Submit job|
|sacct|My jobs|
|srun|Run job interactively|
|scancel <id>|kill job with <id>|
===Prefered practices :===

  * What are the steps you job needs, does one part have to finish before another can run etc
  * An estimate of the how many tasks
  * what resource each task needs (CPU, memory, GPU)
  * Does every step require the same resources
  * An estimate of how long the will run
====To request access====
send **email** to (help@ru.is) with your full name\\

==Additional information==
[[https://slurm.schedmd.com|Slurm]]\\
[[https://code.visualstudio.com/|Visual Studio Code]] can also be used through SSH\\
Use this email [[mailto:help@ru.is|help@ru.is]] for access requests, inquiries etc.\\
[[compute:pythonGPU|Using GPU with python]]