Speaker: Alex Razoumov
Abstract: Does your dataset take forever to download from an HPC system? Is it too large to fit into memory on your computer? Does it take forever to render on your computer's CPU or GPU? All these problems indicate that your dataset is too large for analysis on your computer. Then why not use the same remote system on which you produced your data for rendering too?
With remote visualization, you have many choices: serial vs. parallel visualization, CPUs vs. GPUs rendering, interactive vs. batch rendering, remote desktop vs. client-server setup, and the amount of computational resources to allocate to your problem. In this webinar I will review various visualization workflows, mostly focusing on 3D visualization dealing with large datasets. I will demonstrate interactive client-server visualizations, both serial and parallel, and will use them to create and later debug and run simple scripts on a cluster's login node, inside an interactive job and with a batch job running in the background. I will discuss different visualization tools that you can use remotely but will focus primarily on ParaView.
This webinar is a 2021 take on batch visualization presentations I gave in 2015 and 2019.