Accessing the visualization system
The visualization computer
viz.hpc.taltech.ee can be accessed by SSH from within the university network, from FortiVPN or using a jump-host:
ssh viz.hpc.taltech.ee ssh viz -J base.hpc.taltech.ee
The access to viz is limited to using ssh-keys, no password login. Therefore the ssh-key must be copied to base. More can be found here.
The base home directories are shared with viz.
Using GUI software remotely
At least one of these methods should work for any user, which one depends on the configuration of the client computer (your desktop/laptop).
SSH is essential on all platforms, an X-server and VirtualGL are highly recommended and Xpra and VNC are recommended to have on the client computer.
Window manager or Desktop envirionment for remote use
The default window manager for VNC and X2GO is fvwm2, a configurable, lightweight and fast window manager.
You can get a context menu by clicking on the background (left, middle and right give different menues). By the way, a nice X11 feature is easy copy-and-paste, marking with the left mouse button automatically puts the selection into a copy buffer and clicking the middle mouse button inters at the current mouse cursor position. No annoying ctrl+c, ctrl+v necessary.
Within VNC or X2GO you are running a complete desktop session. Typical modern desktop environments require a lot of memory just for the desktop environment! For this reason only resource-efficient window managers like
tritium are installed.
Software to be automatically started can be configured in
Available Visualization software
ParaView (native install, just run
paraview, to use GPU run
MayaVi (native install, just run
OpenDX (native install, just run
RasMol (native install, just run
In-situ visualization (in preparation)
The following are installed on our cluster
(Ascent)[https://github.com/Alpine-DAV/ascent] on the
ampnodes. On the GPU nodes, Ascent is available in a CUDA-accelerated version.
Ascent on green nodes
module load green-spack/0.19.0 module load ascent
Ascent on amp (CUDA accelerated)
module load amp-spack/0.19.0 module load ascent/0.8.0-gcc-9.3.0-txjp
Catalyst on amp
module load amp-spack/0.19.0 module load
Catalyst can be used within OpenFOAM and (NEK5000)[https://github.com/KTH-Nek5000/InSituPackage] simulations.