In Situ Analysis and Visualization with SENSEI and Ascent


Description:

The widening gap between computational and I/O rates is a key challenge for many computational science efforts. In situ processing, where visualization and analysis process data while it is being generated, is an important solution for this challenge.

The HPC visualization community provides several production tools for in situ processing and is actively researching infrastructures and approaches to help with in situ use cases on HPC platforms. This tutorial blends lectures and hands-on examples to introduce attendees to in situ processing concepts and provide practical experience using state-of-the-art and -practice in situ tools, including SENSEI, Ascent, ParaView Catalyst, VisIt Libsim, Cinema and ADIOS.

Attendees will learn how to use the SENSEI and Ascent in situ interfaces to couple simulation codes to multiple endpoints. This tutorial is ideal for those who would like to learn how to use in situ processing to analyze, visualize, or process data on HPC platforms.

Presenters:

Hank Childs, Cyrus Harrison, Matthew Larsen, Burlen Loring, Silvio Rizzi, Brad Whitlock, E. Wes Bethel, Terece Turton

Time:

Monday, 9 November 2020 2:30pm - 6:30pm EST


Pre-tutorial Setup

We will do live demonstrations during the tutorial, and we have designed the tutorial so that you can run the demonstrations yourself alongside the presenters. If you would like to run the demonstrations yourself, then we encourage you to follow the setup instructions below. That said, while we encourage you to do the demonstrations yourself, we believe those that prefer to watch us perform the demonstrations will still have a fulfilling experience.


SENSEI Virtual Box VM

Description:

Prior to the tutorial participants are invited to download and install the VM. This will enable one to follow along with the hands on exercises.

The VM contains installs and source code of the latest release of SENSEI w/ support for: Libsim, Catalyst, Ascent, ADIOS, Python; and the tutorial demos: LAMMPS OSPRay demo, Libsim Clover demo, ADIOS + X in transit demo

In a nut shell:

Download the VM image, open VirtualBox, import appliance. Go with the defaults except set the number of CPU's to the number of physical cores you have, and set the RAM to the amount of RAM you have. Start the VM. Update the "guest additions" if needed. More detailed instructions are provided in the video below

Download: SC20 VirtualBox VM Image

Setup Video: Setup Video Instructions


Ascent Jupyter Cloud Servers

A portion of tutorial will use cloud hosted Jupyter notebook servers. These servers allow you to connect to web page and run demos without any extra setup. Details on how to access will be posted at https://www.ascent-dav.org/sc/ during the tutorial.

Ascent Tutorial Links:


Cinema Tutorial Setup

There will also be a short tutorial on Cinema, a novel database approach to heterogeneous data analysis and visualization for in situ workflows.

Participants are welcome to follow along with the tutorial or simply watch it as a demo and download the tutorial to work through it later.

The tutorial will demonstrate using ParaView to export a Cinema database and will run a Jupyter notebook workflow to take in an existing Cinema database, updating and exploring it a Cinema viewer. To run the full tutorial, a participant will need:

  • Latest version of ParaView from paraview.org
  • Python3 and various packages
  • Browser security set to allow local file access.

The tutorial is available as a github repo with full details in the readme file:

Cinema SC20 Tutorial


Copyright 2020 LBNL. Comments and Questions? bloring@lbl.gov