http://gfi.wiki.uib.no/api.php?action=feedcontributions&feedformat=atom&user=Ngfihgfi - User contributions [en]2026-11-10T11:03:01ZUser contributionsMediaWiki 1.46.2http://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2058System monitoring and best practice2026-07-03T07:31:23Z<p>Ngfih: </p>
<hr />
<div><br />
Active users share cyclone computing resources (CPU and memory). With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are some user recommendations to make sure that everybody can use it more efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, top ('top -e -m'): Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to sixteen (16) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mysmem='smem -t -r | numfmt --invalid=ignore --header=1 --field=4,5,6,7 --from-unit=1Ki --to=iec --format="%.1f"'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2038Using cyclone2026-04-09T19:37:30Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* PanoplyJ module is currently not working, rather use this alias for starting Panoply 5.9.1<br />
*:alias panoply='/vol/vb_apps/software/PanoplyJ/5.9.1-Java-25.36/panoply.sh'<br />
*:For MacBook users it may be necessary tor run this XQuarts config via terminal (while XQuarts completely off):<br />
*:defaults write org.xquartz.X11 enable_render_extension 0<br />
* NCL module is currently not working, rather use the alias below for NCL 6.6.2:<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code> <br />
*:alias ncl-activate='conda activate /Data/gfi/users/local/share/envs/ncl_stable'<br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2037Using cyclone2026-04-09T19:36:17Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* PanoplyJ module is currently not working, rather use this alias for starting Panoply 5.9.1<br />
*:alias panoply='/vol/vb_apps/software/PanoplyJ/5.9.1-Java-25.36/panoply.sh'<br />
*:For MacBook users it may be necessary tor run this XQuarts config (while XQuarts completely off):<br />
*:defaults write org.xquartz.X11 enable_render_extension 0<br />
* NCL module is currently not working, rather use the alias below for NCL 6.6.2:<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code> <br />
*:alias ncl-activate='conda activate /Data/gfi/users/local/share/envs/ncl_stable'<br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2036Using cyclone2026-04-09T19:25:48Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* PanoplyJ module is currently not working, rather use this alias for starting Panoply 5.9.1<br />
*:alias panoply='/vol/vb_apps/software/PanoplyJ/5.9.1-Java-25.36/panoply.sh'<br />
* NCL module is currently not working, rather use the alias below for NCL 6.6.2:<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code> <br />
*:alias ncl-activate='conda activate /Data/gfi/users/local/share/envs/ncl_stable'<br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2035Using cyclone2026-04-09T19:09:01Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL module is currently not working, rather use the alias below for NCL 6.6.2:<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code> <br />
*:alias ncl-activate='conda activate /Data/gfi/users/local/share/envs/ncl_stable'<br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2034Using cyclone2026-04-09T18:50:17Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code> NCL module is currently not working, rather use the following alias:alias ncl-activate='conda activate /Data/gfi/users/local/share/envs/ncl_stable'<br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=NCL&diff=2033NCL2026-04-09T18:42:42Z<p>Ngfih: </p>
<hr />
<div>Usage of ncl and ncl scripts at GFI.<br />
<br />
Until we have a working NCL module, you can put this alias into your login file (~/.bash_profile) for startup of ncl 6.6.2, startup via calling the alias:<br />
<br />
alias ncl-activate="conda activate /Data/gfi/users/local/share/envs/ncl_stable"<br />
<br />
Below you find a list of existing NCL scripts developed at GFI.<br />
<br />
[[NCL_geostrophic|Calculate geostrophic wind]]<br />
<br />
[[Category:NCL]]<br />
[[Category:WRF]]<br />
[[Category:Models]]<br />
[[Category:Libraries]]</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=2031The GFI computing system2026-03-31T08:13:08Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute uses a collection of department compute server nodes named "cyclone". Cyclone consists of three separate compute nodes which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) applications. The new cyclones replace the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The servers are maintained by [https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. And please be accurate to mark your support request with some ref. to "Group of scientific computing" - and with all necessary details for UiB support to decode your issue description.<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is no queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
Users are expected to read and relate to the "MOTD" message at ssh-login. This monitors computer load and memory consumption. Please follow server load during login period - top/htop and free ('free -h -t') utilities. Do not initiate another heavy compute process on multiple cores if the compute node already runs heavy on load or memory consumption, then rather check the alternate server.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is a resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environment is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs so that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, MacOS clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2030Using cyclone2026-03-26T12:53:13Z<p>Ngfih: /* Displaying GUIs on your screen */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux and macos systems) by adding the following lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2029Using cyclone2026-03-26T12:43:54Z<p>Ngfih: /* Displaying GUIs on your screen */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding the following two lines: <br />
<br />
<pre>ForwardX11 yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2028Using cyclone2026-03-26T12:27:13Z<p>Ngfih: /* Displaying GUIs on your screen */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. Inclusion of <code>-C</code> option as well is an advantage for speeding up the X11 response. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes<br />
Compression yes</pre><br />
<br />
Windows will require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.<br />
<br />
For Mac the solution should be pretty similar when using XQuartz - for Tahoe and later macos versions it may come necessary with the following initial setting for Xquartz, entered from your Mac command shell:<pre>defaults write org.xquartz.X11 enable_render_extension 0</pre><br />
<br />
And it may come fortunate with later follow up toggling "Full-screen mode" in XQuarts settings -> Output.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2024Using cyclone2026-03-20T12:57:40Z<p>Ngfih: /* Accessing the data from your own machines */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,ServerAliveCountMax=3,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding or uncommenting the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes</pre><br />
<br />
For Mac the solution should be pretty similar when using XQuartz, windows may require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2023Using cyclone2026-03-20T12:49:40Z<p>Ngfih: /* Accessing the data from your own machines */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
For access to Sigma2 storage resources via sshfs access from cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
<br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding or uncommenting the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes</pre><br />
<br />
For Mac the solution should be pretty similar when using XQuartz, windows may require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2022Using cyclone2026-03-20T12:42:23Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre>For access to Sigma2 resources via sshfs accessfrom cyclones, defining the following aliases can be convenient for mount and umount towards ~/sigma2 - and similar can be done from Sigma2-resource for access to cyclones storage resource:<pre>alias mountsigma2="mkdir ~/sigma2; sshfs <userid>@<sigma2-host>:<sigma2_path> ~/sigma2 -o reconnect,ServerAliveInterval=15,follow_symlinks,uid=$(id -u),gid=$(id -g)"<br />
alias umountsigma2="cd ; chmod go+rx ~/; fusermount -u ~/sigma2; chmod go-rx ~/; rmdir ~/sigma2"<br />
</pre><br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding or uncommenting the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes</pre><br />
<br />
For Mac the solution should be pretty similar when using XQuartz, windows may require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=2021The GFI computing system2026-03-19T13:20:13Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclones replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. And please be accurate to mark your support request with some ref. to "Group of scientific computing" - and with all necessary details for UiB support to decode your issue description.<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
Users are expected to read and relate to the "MOTD" message at ssh-login. This monitor computer load and memory consumption. Please follow server load during login period - top/htop and free ('free -h -t') utilities. Do not initiate another heavy compute process on multiple cores if the compute node already runs heavy on load or memory consumption, then rather check the alternate server.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2020Using cyclone2026-03-18T12:07:13Z<p>Ngfih: </p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connection via ssh is also open from Sigma2 resources, and sshfs is available to/from Sigma2. Otherwise, connecting from outside UiB or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding or uncommenting the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes</pre><br />
<br />
For Mac the solution should be pretty similar when using XQuartz, windows may require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Using_cyclone&diff=2019Using cyclone2026-03-17T15:30:00Z<p>Ngfih: /* Jupyter & Jupyterhub */</p>
<hr />
<div>== Connection and first steps ==<br />
Requirements: To access GFI&SKD department servers via ssh, your computer will need to be attached to UiB-network. Connecting from outside or from private platform, you will need UiB-vpn or login via UiB login server. And your UiB account will need to be enabled for connection.<br />
<br />
Use ssh to connect to one of the cyclones systems, cyclone1, cyclone2 and cyclone3, respectively:<br />
<pre>ssh -Y <username>@cyclone1.gfi.uib.no<br />
ssh -Y <username>@cyclone2.gfi.uib.no<br />
ssh -Y <username>@cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)</pre><br />
or just simply:<br />
<pre>ssh -Y cyclone1.gfi.uib.no<br />
ssh -Y cyclone2.gfi.uib.no<br />
ssh -Y cyclone3.gfi.uib.no - (cyclone3 is reserved for GPU usage only)<br />
</pre><br />
if you're connecting from a computer where you are logged in under the same user name (as usually the case with your UiB machine). When you are connecting for the first time, you will be asked to verify a signature. Reply "yes" to this question.<br />
<br />
=== Home directories ===<br />
<br />
All servers <code>cyclone*.gfi.uib.no</code> share same home directories. These home directories are separate from your standard UiB home directory. The content of your home directories will not be synced, so you will manually need to transfer your scripts to the new system.<br />
<br />
Please be aware that your local home directory is not expected to exceed 30GB at any time - even in periods when sys.adms. do not practice quotas. If you need to store larger amount of data, please plan for using the GFI&SKD storage - also described in this wiki.<br />
<br />
To monitor your home directory size, you can use these commands from cyclone command shell:<br />
<br />
<code>du -sh ~/ </code><br />
<br />
Or to include more details on both dot files/directories and non-hidden home directory subfolders:<br />
<br />
<code>cd; du -sh .[^.]* *</code><br />
<br />
=== How to access software ===<br />
<br />
Most/all scientific software is only available through EESSI software modules. Modules need to be loaded manually before the respective software can be used. We use software modules, because they provide are a clean and handy way to offer conflicting software or software versions: Instead of updating software and potentially breaking user scripts and programs new software is then made available in parallel to those that require it.<br />
<br />
=== Start-up scripts ===<br />
<br />
It might be handy to automatically load modules for software that you use frequently every time you login to cyclone. To accomplish that for bash, there are two relevant files, <code>$HOME/.bashrc</code> and <code>$HOME/.bash_profile</code> . <code>.bash_profile</code> should contain things that need to be defined at login time only, like <code>PATH</code> and other environment variables, startup programs, etc. You just need things once, not in every shell you open. In most cases, you also need the things from <code>.bashrc</code> in your login shell. That's why <code>.bash_profile</code> sources <code>.bashrc</code> as well, but .bashrc doesn't usually source <code>.bash_profile</code>. In contrast, <code>.bashrc</code> should typically contain things you want to set in every shell you open, like aliases, functions, etc. These are per shell session items that are not inherited from environment ([https://unix.stackexchange.com/questions/346336/why-have-both-bash-profile-and-bashrc source and detailed explanation]).<br />
<br />
==== An example <code>.bash_profile</code> for inspiration ====<br />
<br />
<pre>module load CDO/2.2.2-gompi-2023b<br />
module load MATLAB/2023a<br />
<br />
alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"<br />
<br />
if [ -f ~/.bashrc ]; then<br />
. ~/.bashrc<br />
fi</pre><br />
<br />
==== An example <code>.bashrc</code> for inspiration ====<br />
<br />
<pre>alias rm="rm -i"<br />
alias cp="cp -i"<br />
alias erai="cd /Data/skd/share/Reanalysis/ERA_INTERIM/6HOURLY/"</pre><br />
<br />
<br />
== Available software ==<br />
<br />
Below is a list of frequently used software and which module to load on the new cyclone. Relevant module names are printed like for <code>MATLAB/2023a</code>. Modules can be loaded by <code>module load <module name></code>, and unloaded <code>module unload <module name></code>. [[Using modules on cyclone|More details on the module system]].<br />
<br />
To list all available modules, use the command<br />
<pre>module avail</pre><br />
<br />
All packages listed with this command should work, so please report to <code>support@hpc.uib.no</code> if they do not.<br />
<br />
=== Data analysis and plotting ===<br />
<br />
* CDO<br />
*: <code>CDO</code>, <code>CDO/2.2.2-gompi-2023</code><br />
* Dynlib<br />
*: See temporary solution for python. Requires in addition <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code> and <code>ScaLAPACK/2.0.2-gompi-2018b-OpenBLAS-0.3.1</code>.<br />
* ghostview<br />
*: ''<no module required>''<br />
* grads<br />
*: ''<no module required>''<br />
* Matlab<br />
*: <code>Matlab</code>, <code>Matlab/2023a</code>, <code>Matlab/2021b</code><br />
* ncdump and other netCDF utilities<br />
*: <code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code><br />
* NCL<br />
*: <code>NCL</code>, <code>NCL/6.5.0</code><br />
* nco<br />
*: <code>NCO</code>, <code>NCO/5.1.9-foss-2023a</code><br />
* ncview<br />
*: <code>ncview</code>, <code>ncview/2.1.8-gompi-2023a</code>. Load only when needed! The package is currently known to break some GUI programs like gedit and gvim. <br />
* python, plus selection of scientific packages<br />
*: ''First'' choose and load the python module for the python version that you want to use, either <code>Python/3.11.5-GCCcore-13.2.0</code> for python 3.11.5, or <code>Python/2.7.18-GCCcore-12.2.0-bare</code> for python 2.7.18.<br />
*: ''Then'', as a temporary solution, many scientific python packages are installed in the following virtual environments:<br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie2/bin/activate</code> for python 2.7.15. <br />
*: <code>source /Data/gfi/users/local/share/virtualenv/dynpie3-2021a/bin/activate</code> for python 3.7.0.<br />
*: For the pygrib module, <code>ecCodes/2.7.3-foss-2018b</code> is required in addition.<br />
* R<br />
*: <code>R , R/4.4.1-gfbf-2023b</code><br />
<br />
=== Text editors ===<br />
<br />
* emacs, gedit, gvim, vi(m), nedit, meld, spyder (Spyder comes available when Anaconda module loaded)<br />
*: All installed globally<br />
<br />
=== Compilers and libraries ===<br />
<br />
* blas<br />
*: <code>OpenBLAS/0.3.1-GCC-7.3.0-2.30</code>. '''Note''': The library is called <code>libopenblas</code>, so you'll need to compile with <code>-lopenblas</code>. <br />
* eccodes (replacement for grib_api)<br />
*: <code>ecCodes</code>, <code>ecCodes/2.31.0-gompi-2023b</code><br />
* gcc, gfortran<br />
*: <code>GCC</code>, <code>GCC/13.2.0</code><br />
* grib_api<br />
*: <code>grib_api/1.24.0-foss-2018b</code><br />
* HDF5 libraries<br />
*: <code>HDF5/1.14.3-gompi-2023b</code><br />
* lapack<br />
*: <code>ScaLAPACK</code>, <code>ScaLAPACK/ScaLAPACK/2.2.0-gompi-2023b-fb</code>. '''Note''': The library is called <code>libscalapack</code>, so you'll need to compile with <code>-lscalapack</code>.<br />
* MPI<br />
*: <code>OpenMPI/4.1.6-GCC-13.2.0</code><br />
* netcdf<br />
*: For C-libaries:<code>netCDF</code>, <code>netCDF/4.9.2-gompi-2023b</code>, <br />
*: For Fortran-libraries <code>netCDF-Fortran/4.6.1-gompi-2023b</code>, compile with option <code>-I/vol/vb_apps/software/netCDF-Fortran/4.6.1-gompi-2023b/include</code><br />
* proj<br />
*: <code>PROJ</code>,<code>PROJ/9.3.1-GCCcore-13.2.0</code><br />
<br />
=== Jupyter & Jupyterhub ===<br />
<br />
Jupyter-lab and jupyter notebook can be initiated when Anaconda module has been activated:<pre>ml Anaconda3<br />
jupyter-lab --no-browser --port 888X (unoccupied port, interval 8880 < X < 8951)<br />
(Prepare for copy&paste of the output line http://localhost:888X/lab?token=<unique-code>)</pre>Then from your local client computer:<pre>ssh -N -L 888X:localhost:888X <userid>@cyclone1.gfi.uib.no<br />
(same portnumber, hostname= cyclone1 or cyclone2)</pre>From your preferred browser on your local client computer, paste in the http://localhost url output line from jupyter-lab:<pre>http://localhost:888X/lab?token=<unique code></pre>Your jupyter-lab is now supposed to run towards your local browser window.<br />
<br />
You can change "jupyter-lab" to "jupyter notebook" in first startup if notebook is preferred. <br />
<br />
Here is a step by step guide to setup and use Jupyter based on the <code>dynpie3</code> environment.<br />
<br />
[[File:Get started cyclone.pdf|thumb|Step by step guide for Jupyter on cyclone]]<br />
<br />
<br />
== Accessing the data from your own machines ==<br />
<br />
Note: From your private machines you will need to be connected to the UiB VPN to be able to mount the following network drives. From UiB-client computers, UiB network connection is sufficient.<br />
<br />
On MacOS you can mount folders through <code>Finder > Go > Connect to Server ...</code>. In the dialogue window that pops up, enter<br />
<pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre><br />
as the '''Server Address'''. You might want to replace <code>gfi</code> by <code>skd</code> to access the skd disks if that is your primary affiliation.<br />
On Ubuntu this is very similar: <code>Files > Go > Enter Location ...</code>. Enter the address: <pre>smb://uib.no;<user name>@klient.uib.no/FELLES/MATNAT/GFI</pre> and enter your password in the next pop-up window, while keeping "UiB" in the "Domain" field.<br />
<br />
From Windows you can mount by navigating to <code>File Explorer > Map network drive</code> and enter <br><code>\\klient.uib.no\felles\matnat\gfi</code>. Make sure to enable <code>Connect using different credentials</code> and fill in <code><user name>@uib.no</code> and your PW.<br />
<br />
From platforms with sshfs available, alternative mount of network drive with sshfs to a dedicated directory using <code>sshfs</code> (linux: install with <code>sudo apt install sshfs</code>):<br />
<pre>mkdir ~/cyclone_gfi<br />
From UiB network: sshfs -o follow_symlinks <user name>@cyclone2.uib.no:/gfi/ ~/cyclone_gfi/<br />
From ext. networks, please connect to UiB vpn first.<br />
Unmount with: fusermount -u ~/cyclone_gfi</pre><br />
== Displaying GUIs on your screen ==<br />
<br />
The ssh-login command at the top of this wiki enables trusted X11 forwarding (the <code>-Y</code> option), which allows you to display Graphical User Interfaces (GUIs) on your screen. This way you may be able to run software on a high performance system in a more userfriendly way than by just operating via the terminal. To make this work properly you may have to edit the ssh configuration file (<code>~/.ssh/config</code> on linux systems) by adding or uncommenting the following two lines: <br />
<br />
<pre>X11Forwarding yes<br />
ForwardX11Trusted yes</pre><br />
<br />
For Mac the solution should be pretty similar when using XQuartz, windows may require additional terminal software (for establishing an ssh-connection) with inclusion of an XServer (e.g. MobaXterm, WSL or VcXsrv) or [https://hjelp.uib.no/tas/public/ssp/content/detail/service?unid=a9cc906b686940498f0cf68882fc5d20 NoMachine] - and some still prefer Putty.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=2018The GFI computing system2026-03-17T14:18:26Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclones replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
Users are expected to read and relate to the "MOTD" message at ssh-login. This monitor computer load and memory consumption. Please follow server load during login period - top/htop and free ('free -h -t') utilities. Do not initiate another heavy compute process on multiple cores if the compute node already runs heavy on load or memory consumption, then rather check the alternate server.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2017System monitoring and best practice2026-03-17T14:13:12Z<p>Ngfih: </p>
<hr />
<div><br />
Active users share cyclone computing resources (CPU and memory). With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are some user recommendations to make sure that everybody can use it more efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, top ('top -e -m'): Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mysmem='smem -t -r | numfmt --invalid=ignore --header=1 --field=4,5,6,7 --from-unit=1Ki --to=iec --format="%.1f"'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Best_practices_when_using_cyclone&diff=2016Best practices when using cyclone2026-03-17T14:05:36Z<p>Ngfih: Ngfih moved page Best practices when using cyclone to System monitoring and best practice</p>
<hr />
<div>#REDIRECT [[System monitoring and best practice]]</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2015System monitoring and best practice2026-03-17T14:05:36Z<p>Ngfih: Ngfih moved page Best practices when using cyclone to System monitoring and best practice</p>
<hr />
<div><br />
Active users share cyclone computing resources (CPU and memory). With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few user recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mysmem='smem -t -r | numfmt --invalid=ignore --header=1 --field=4,5,6,7 --from-unit=1Ki --to=iec --format="%.1f"'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Main_Page&diff=2014Main Page2026-03-17T14:03:34Z<p>Ngfih: </p>
<hr />
<div>== Geophysical Institute ==<br />
<br />
[[File:GFI.jpg]]<br />
<br />
==[[Teaching]]==<br />
<br />
==Research and PhD/Postdoc Groups==<br />
===[[Meteorology]]===<br />
===[[ResClim Bergen subgroup]]===<br />
===[[Physical Oceanography group]]===<br />
===[[High-Res Ocean interest group]]===<br />
<br />
==Observations==<br />
===[[ITAS Automatic Weather Stations]]===<br />
===[[MRR Micro Rain Radar]]===<br />
===[[TPS-3100 Hotplate pluviometer]]===<br />
===[[Parsivel2 Disdrometer]]===<br />
===[[HOBO rain gauges]]===<br />
===[[EasyLog temperature logger]]===<br />
===[[iMet UAV sonde package]]===<br />
===[[Kristine Bonnevie shiplog data]]===<br />
===[[Joining several data files with CDO and IsoFuse]]===<br />
<br />
==Shared data sets==<br />
===[[ERDDAP]]===<br />
===[[ERA5 reanalysis data]]===<br />
<br />
==IT (Hardware and Software)==<br />
===[[The GFI computing system]]===<br />
===[[Using cyclone|Using cyclones]]===<br />
===[[Using modules on cyclone|Using modules on cyclones]]===<br />
===[[Best practices when using cyclone|System monitoring and best practice]]===<br />
===[[Bedymo]]===<br />
===[[Dynlib]]===<br />
===[[Forecast plotting]]===<br />
===[[Git]]===<br />
===[[NCL]]===<br />
===[[Python]]===<br />
===[[WRF]]===<br />
===[[OpenIFS]]===<br />
===[[GFI&SKD Storage]]===<br />
===(outdated) [[Using the queue system on skd-cyclone]]===<br />
<br />
==IsoMet Tools==<br />
===[[FLEXPART]]===<br />
===[[Lagranto]]===<br />
===[[WaterSip]]===<br />
===[[Flextract]]===<br />
===[[Joining several data files with CDO and IsoFuse|IsoFuse]]===<br />
===[[Rayleigh model]]===<br />
===[[FLIIMP - FARLAB liquid water isotope measurements processor]]===<br />
===[[FaVaCal - FARLAB water vapour isotope calibration processor]]===<br />
<br />
==Social Activities and Groups==<br />
===[[Football]]===<br />
===[[Gale Brewing Club]]===<br />
===[[Geosupen]]===<br />
===[[GFI-social]]===<br />
===[[Kakeklubben]]===<br />
===[[Music]]===<br />
===[[PhD Forum]]===<br />
===[[Schafkopf]]===<br />
===[[Volleyball]]===<br />
<br />
[[Category:IT]]</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2013System monitoring and best practice2026-03-17T13:54:52Z<p>Ngfih: </p>
<hr />
<div><br />
Active users share cyclone computing resources (CPU and memory). With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few user recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mysmem='smem -t -r | numfmt --invalid=ignore --header=1 --field=4,5,6,7 --from-unit=1Ki --to=iec --format="%.1f"'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2012System monitoring and best practice2026-03-17T13:53:47Z<p>Ngfih: </p>
<hr />
<div><br />
Active users share cyclone computing resources (CPU and memory). With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few user recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2011System monitoring and best practice2026-03-17T13:50:35Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.<br />
<br />
'''Useful memory monitoring aliases''' for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2010System monitoring and best practice2026-03-17T13:46:35Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
Some useful memory monitoring aliases for your login files:<br />
• alias pstop-cpu='ps axcu --sort -%cpu | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias pstop-mem='ps axcu --sort -pmem | head -n30 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-cpu='ps xcu --sort -%cpu | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
• alias mypstop-mem='ps xcu --sort -pmem | head -n10 | numfmt --invalid=ignore --header=1 --field=5,6 --from-unit=1Ki --to=iec | column -t'<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2009System monitoring and best practice2026-03-17T13:35:41Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, 'top -e -m': Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
CPU usage on cyclones is currenly limited by the operating system to eight (8) CPU cores per user session. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Please limit the number of CPUs your programs use.<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
Cyclone1 and Cyclone2 have a physical limit of 512GB RAM. There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Some times, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2008System monitoring and best practice2026-03-17T13:26:17Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
Be aware: Compared to <code>top</code> , the numerical "used" memory value in <code>htop</code> more closely represents actively used memory by processes, excluding buffers and cache. From <code>top</code> "used" memory value often includes buffers and cache by default, which can be misleading as this memory can be freed if needed.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclones are currenly limited by the operating system to eight (8) CPU cores. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2007System monitoring and best practice2026-03-17T13:13:42Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are:<br />
• htop, top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• (h)top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
• free -h -t: Monitor total memory, used, free and available<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclones are currenly limited by the operating system to eight (8) CPU cores. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2006System monitoring and best practice2026-03-17T13:08:01Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Read and relate to motd message (Message Of The Day) at ssh login - also given at command: "motd"<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 CyclonesMonitor] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclones are currenly limited by the operating system to eight (8) CPU cores. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2005System monitoring and best practice2026-03-17T13:01:26Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Monitor cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 Cyclones] <br />
<br />
Monitor GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 Cyclone3 GPU]<br />
<br />
Monitor storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Other useful monitoring commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (64 CPU cores for cyclone1 & 2). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclones are currenly limited by the operating system to eight (8) CPU cores. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user session can allocate >120GB memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2004System monitoring and best practice2026-03-17T12:53:23Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Lookup cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 MonitorCyclones] <br />
<br />
Lookup GPU resources included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 MonitorCyclone3 GPU]<br />
<br />
Look up storage quotas and backup status: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2003System monitoring and best practice2026-03-17T12:49:22Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
Lookup cpuload and memory, included history (username:gfireader, pw: gfimonitor):<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100 MonitorCyclones] <br />
<br />
Lookup GPU resources:<br />
[https://monitor.gfi.uib.no/d/ads6vth/nvidia-dcgm-exporter-dashboard-cyclone3?orgId=1&from=now-5m&to=now&timezone=browser&var-instance=cyclone3.gfi.uib.no:9400 MonitorCyclone3 GPU]<br />
<br />
Look up storage quotas: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2002System monitoring and best practice2026-03-17T12:22:49Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
The following Grafana site presents the current status of all cyclone nodes with current status of cpuload and memory - and some history (username:gfireader, pw: gfimonitor):<br />
<br />
[https://monitor.gfi.uib.no MonitorGFI] <br />
<br />
- Look up quotas: [https://grafse.app.uib.no/public-dashboards/22f87936e6e649f0806c36379562ebae UiB Grafana web site]<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2001System monitoring and best practice2026-03-17T12:13:45Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
The following Grafana site presents the current status of all cyclone nodes with current status of cpuload and memory - and some history:<br />
<br />
[https://monitor.gfi.uib.no monitor.gfi.uib.no] (username:gfireader, pw: gfimonitor)<br />
<br />
<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=2000System monitoring and best practice2026-03-17T12:12:57Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
The following Grafana site presents the current status of all cyclone nodes with current status of cpuload and memory - and some history:<br />
<br />
[https://monitor.gfi.uib.no/d/rYdddlPWj/node-exporter-full?orgId=1&from=now-3h&to=now&timezone=browser&var-DS_PROMETHEUS=default&var-job=node_exporters&var-node=cyclone1.gfi.uib.no:9100&var-diskdevices=[a-z]%2B|nvme[0-9]%2Bn[0-9]%2B&refresh=1m monitor.gfi.uib.no] (username:gfireader, pw: gfimonitor)<br />
<br />
<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=System_monitoring_and_best_practice&diff=1999System monitoring and best practice2026-03-17T12:08:17Z<p>Ngfih: </p>
<hr />
<div><br />
Users share cyclone computing resources (CPU and memory) at the same time. With several nodes and limited resources, it is necessary to monitor system resources and distribute users and applications on the relevant servers. Here are a few recommendations to make sure that everybody can use it most efficiently.<br />
<br />
The following Grafana site presents the current status of all cyclone nodes with current status of cpuload and memory - and some history<br />
[https://monitor.gfi.uib.no monitor.gfi.uib.no] (username:gfireader, pw: gfimonitor)<br />
<br />
<br />
<br />
Useful commands are<br />
• htop and top: Monitor usage of CPU and memory usage for the entire system. CPU usage given here is percent of one CPU core. I.e., a usage of 800% corresponds to 8 CPU cores being fully used.<br />
• top -u <your_username>: Monitor CPU and memory usage for your programs only.<br />
<br />
<br />
'''Limit your CPU use'''<br />
<br />
Often, one program running on cyclone will use one CPU with maximum 100% CPU. However, some software such as Matlab, some Python modules, some model simulations, etc, by default try to occupy the entire machine they run on (72 virtual CPU cores for cyclone). This drastically slows down the jobs of other users, especially if several such programs run at same time.<br />
<br />
CPU usage on cyclone is limited by the operating system to 25 virtual CPU cores - approx 1/3 of total. Trying to use more than that will not only slow down execution for everybody else, but also for yourself!<br />
<br />
Therefore, please limit the number of CPUs your programs use.<br />
<br />
- For Matlab: <br />
LASTN = maxNumCompThreads(N)<br />
with N set to a maximum of 8 (preferably less) - the user allows for 8 physical cores and 16 virtual cores.<br />
<br />
https://se.mathworks.com/help/matlab/ref/maxnumcompthreads.html<br />
<br />
- For Python (especially when using Pandas) and OpenMP-parallelised programs (written in Fortran, C, or any other language): <br />
1) In the shell, before starting the program:<br />
export OMP_NUM_THREADS=N<br />
2) Alternatively, within python:<br />
import os<br />
os.environ["OMP_NUM_THREADS"] = "N"<br />
<br />
Preferably, set N to a maximum of 8 (preferably less).<br />
<br />
If these limits are too restrictive for your application, you may want to look at other options for running heavily parallelized jobs, such as Fram of the Norwegian e-infrastructure.<br />
<br />
<br />
'''Limit your memory use'''<br />
<br />
This is currently less critical, as cyclone has a large amount of memory. However, sometimes, if the job uses a lot of memory it can be the sign of something going wrong in your script. Therefore, make sure to control your memory and to clear up/delete all unused variables (to deallocate the memory). Try also to close Matlab or Python when the job is finished, in the evening or before leaving for the weekend.<br />
<br />
There is a hard limit imposed by the operating system that no user can take up more than half of cyclone’s memory. If you attempt to violate this restriction, the offending process will be killed by the operating system.<br />
<br />
Thank you for following these recommendations.</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=Main_Page&diff=1998Main Page2026-03-17T11:52:10Z<p>Ngfih: </p>
<hr />
<div>== Geophysical Institute ==<br />
<br />
[[File:GFI.jpg]]<br />
<br />
==[[Teaching]]==<br />
<br />
==Research and PhD/Postdoc Groups==<br />
===[[Meteorology]]===<br />
===[[ResClim Bergen subgroup]]===<br />
===[[Physical Oceanography group]]===<br />
===[[High-Res Ocean interest group]]===<br />
<br />
==Observations==<br />
===[[ITAS Automatic Weather Stations]]===<br />
===[[MRR Micro Rain Radar]]===<br />
===[[TPS-3100 Hotplate pluviometer]]===<br />
===[[Parsivel2 Disdrometer]]===<br />
===[[HOBO rain gauges]]===<br />
===[[EasyLog temperature logger]]===<br />
===[[iMet UAV sonde package]]===<br />
===[[Kristine Bonnevie shiplog data]]===<br />
===[[Joining several data files with CDO and IsoFuse]]===<br />
<br />
==Shared data sets==<br />
===[[ERDDAP]]===<br />
===[[ERA5 reanalysis data]]===<br />
<br />
==IT (Hardware and Software)==<br />
===[[The GFI computing system]]===<br />
===[[Using cyclone|Using cyclones]]===<br />
===[[Using modules on cyclone|Using modules on cyclones]]===<br />
===[[Best practices when using cyclone|Cyclones - System monitoring and best practice]]===<br />
===[[Bedymo]]===<br />
===[[Dynlib]]===<br />
===[[Forecast plotting]]===<br />
===[[Git]]===<br />
===[[NCL]]===<br />
===[[Python]]===<br />
===[[WRF]]===<br />
===[[OpenIFS]]===<br />
===[[GFI&SKD Storage]]===<br />
===(outdated) [[Using the queue system on skd-cyclone]]===<br />
<br />
==IsoMet Tools==<br />
===[[FLEXPART]]===<br />
===[[Lagranto]]===<br />
===[[WaterSip]]===<br />
===[[Flextract]]===<br />
===[[Joining several data files with CDO and IsoFuse|IsoFuse]]===<br />
===[[Rayleigh model]]===<br />
===[[FLIIMP - FARLAB liquid water isotope measurements processor]]===<br />
===[[FaVaCal - FARLAB water vapour isotope calibration processor]]===<br />
<br />
==Social Activities and Groups==<br />
===[[Football]]===<br />
===[[Gale Brewing Club]]===<br />
===[[Geosupen]]===<br />
===[[GFI-social]]===<br />
===[[Kakeklubben]]===<br />
===[[Music]]===<br />
===[[PhD Forum]]===<br />
===[[Schafkopf]]===<br />
===[[Volleyball]]===<br />
<br />
[[Category:IT]]</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1997The GFI computing system2026-03-17T10:29:56Z<p>Ngfih: /* System characteristics and good user practice */</p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclones replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
Users are expected to read and relate to the "MOTD" message at ssh-login. This monitor computer load and memory consumption. Please follow server load during login period - top/htop and free ('free -h -t') utilities. Do not initiate another heavy compute process on multiple cores if the compute node already runs heavy on load or memory consumption, then rather check the alternate server.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1996The GFI computing system2026-03-17T10:16:52Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclones replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1995The GFI computing system2026-03-17T10:15:20Z<p>Ngfih: /* Storage environment */</p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1994The GFI computing system2026-03-17T10:14:29Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1993The GFI computing system2026-03-17T10:12:38Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1992The GFI computing system2026-03-17T10:10:37Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running Linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
The cyclone compute system has been set up to enable users to perform smaller-scale data analysis work in a low-threshold environment. This means that there is now queue system for resource allocation. Users can use the system interactively, running applications with a GUI, or via Linux shell etc.<br />
<br />
All users share the same resources, without having to deal with a queue system. This requires users to excert discipline when using resources to prevent system instability due to too heavy load.<br />
<br />
Common use cases include data analysis connected to the shared data files in the storage environment, model simulations lasting up to several days, and tests of larger-scale computations before taking those to the national computation infrastructures NRIS.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelization jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storage, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
<br />
<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1991The GFI computing system2026-03-17T10:06:05Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized network storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently running linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1990The GFI computing system2026-03-17T10:01:18Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by [https://https://www.uib.no/en/brita UiB ITA] and compute resources are located in [https://nrec.no NREC] as virtual servers, currently operated with linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1989The GFI computing system2026-03-17T09:51:37Z<p>Ngfih: </p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by UiB ITA and compute resources are located in NREC as virtual servers, currently operated with linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no * https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders:<br />
Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and<br />
Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1988The GFI computing system2026-03-17T09:48:25Z<p>Ngfih: /* Storage environment */</p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by UiB ITA and compute resources are located in NREC as virtual servers, currently operated with linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no * https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders: Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1987The GFI computing system2026-03-17T09:45:56Z<p>Ngfih: /* Storage environment */</p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by UiB ITA and compute resources are located in NREC as virtual servers, currently operated with linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no * https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders: Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfihhttp://gfi.wiki.uib.no/index.php?title=The_GFI_computing_system&diff=1986The GFI computing system2026-03-17T09:43:14Z<p>Ngfih: /* GPU acceleration */</p>
<hr />
<div>The Geophysical Institute use a department compute server cluster named "cyclone". Cyclone consists of three separate compute nodes organized in a cluster which access the same centralized storage. Two of the nodes are for regular computational tasks, and one is set up with a GPU to enable use for machine learning (ML) application. The new cyclone replaces the older system '"cyclone.hpc.uib.no", acquired in Summer 2018, which at the time of writing this documentation is still accessible as well (14.03.2026).<br />
<br />
The server cluster is maintained by UiB ITA and compute resources are located in NREC as virtual servers, currently operated with linux, Rocky 8<br />
<br />
All users of the cyclones should subscribe to the mailing list [linux@gfi.uib.no] to receive updates about maintenance downtime etc. <br />
<br />
Please notify the UiB support [https://hjelp.uib.no * https://hjelp.uib.no] in any case of problems or questions. You may give your support request some ref. to "Group of scientific computing".<br />
<br />
In the following, you will find:<br />
<br />
==Users and intended use==<br />
<br />
The intended users at GFI for the cyclones are:<br />
<br />
* MSc students for course and thesis work.<br />
<br />
* PhD students and PostDocs for scientific work.<br />
<br />
* all other researchers and technical staff for data analysis, data storage, and routine computations.<br />
<br />
==Requirements==<br />
<br />
Users should be able to use the system in interactive use (click and point, typing) and in batch use (using shell scripts). Typical tasks are data analysis, plotting. GFI will also run routine data processing of observations and forecasts on this system.<br />
<br />
Therefore, we required the system to be or have<br />
<br />
* low threshold access (easy to use)<br />
<br />
* interactive use<br />
<br />
* safe usage (against unintentional overuse)<br />
<br />
* optimized for serial I/O (as is typical for post-processing tasks)<br />
<br />
* single-node parallelisation jobs<br />
<br />
* lifetime of 5-7 years<br />
<br />
== System characteristics and good user practice ==<br />
<br />
The system has been created as a low threshold, multi-user system. Resources are limited, and shared among all users. This implies that users need to have an overview over which resources they are using.<br />
<br />
The system characteristics, and the good user practice that follows along with these, are as follows:<br />
<br />
* There is (currently) no queue system on any of the "cyclones". Computational tasks are started immediately.<br />
<br />
* Computational tasks will be terminated if users log out. To keep jobs running while logged out, start jobs in the background using the linux commands 'tmux', 'screen' or 'nohup'.<br />
<br />
* There is resource limitation per user. Each session can use up to 8 out of the 64 CPU cores. In addition, each session can allocate up to 120 GB of memory. This is meant to prevent individual users from accidentally bringing down the system.<br />
<br />
* Software packages are activated using <nowiki>'''</nowiki>module load<nowiki>'''</nowiki> command. Use 'module spider <software>' to find available software. The available software stack is identical on all cyclones.<br />
<br />
* Users have access to a home directory, a work storace, and long-term project and shared resource storage. The same storage environments is connected to all of the cyclones. Use the work storage for all non-permanent input/output. Files on the work environment are deleted after 60 days. Use NIRD for long-term storage of larger output.<br />
<br />
* The system is maintained by the experts from UiB's HPC group. There is bi-weekly maintenance planned, typically on Wednesdays. Sometimes this will involve a reboot of the system. Plan longer jobs that they don't get interrupted or interfere with maintenance scheduling.<br />
* Subscribe to the mailing list [linux@gfi.uib.no](<nowiki>mailto:linux@gfi.uib.no</nowiki>) to receive updates about problems and maintenance<br />
<br />
== Storage environment ==<br />
<br />
GFI&SKD have a disposal of approx 450TB network storage quotas from UiB ITA(May 2024). This storage is all organized via ITA standard NetApp solution and distributed towards cyclones servers, GFI&SKD Windows clients, Macos clients and Linux clients. The storage is also available for GFI&SKD users connecting from UiB-vpn.<br />
<br />
*The GFI&SKD storage is organized into larger shared areas for model data, more limited group and project areas, areas for individual users and common areas for short time storage. The different storage areas are in general maintained via ITA backup procedures and governed by group or individual storage quotas. <br />
<br />
* GFI&SKD storage is organized in two main folders: Linux: /Data/gfi - Win11/Mac: \\klient.uib.no\felles\matnat\gfi and Linux: /Data/skd - Win11/Mac: \\klient.uib.no\felles\matnat\skd<br />
<br />
== Computational resources ==<br />
<br />
The compute system is separated into three separate nodes. Two nodes (cyclone1 and cyclone2) are dedicated to computing tasks without GPU use. A third node is dedicated for GPU applications (cyclone 3). In detail, the specifications of the 3 compute nodes are:<br />
<br />
'''Cyclone 1 - CPU applications (cyclone1.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 2 - CPU applications (cyclone2.gfi.uib.no):'''<br />
<br />
AMD EPYC Processor, 64 physical cores and 512 GB memory per node, 2.3 GHz - no GPU<br />
<br />
'''Cyclone 3 - GPU applications (cyclone3.gfi.uib.no):'''<br />
<br />
Intel CPU, 16 physical cores, 128 GB memory<br />
<br />
1 NVIDIA GPU, L40S-24Q, 24 GB GPU memory<br />
<br />
== GPU acceleration ==<br />
<br />
One of the cyclones (cyclone3) is equipped with an advanced Graphics Processing Units (GPUs) of type NVIDIA L40S-24Q, and is reserved for GPU application usage only. The GPU is shared between all users on the system. <br />
<br />
The GPU usage can be monitored using the command: "nvidia-smi".<br />
<br />
Programming on this GPU is done using CUDA.<br />
<br />
Modules are available for programming with CUDA.<br />
<br />
Python users can load module PyTorch</div>Ngfih