Here you will find the preparation information, materials, and requirements to be part of the Singularity labs at the Intel HPC DevCon as part of Supercomputing 2016 (click here to register for free).
Singularity is a container platform for scientific use cases enabling reproducible science, mobility of compute, and integration on traditional high performance computing resources. Singularity is installed on some of the largest systems in the world (near half million cores on TACC's Stampede and GSI's Center for Heavy Ion Research/Greencube) as well as other centers and midrange computing resources worldwide.
These sessions and workshops will be given by the primary author and project lead of Singularity (Gregory M. Kurtzer, also known for founding Centos Linux and Warewulf).
There are two sessions dedicated to Singularity on the agenda:
-
Session 1, Saturday 12th at 2:05-3:30pm - HPC Containers & Singularity (Overview and Tutorial)
-
Session 2, Sunday 13th at 9:45-10:35am - HPC Containers and & Singularity (Advanced Tutorial)
Attendance for these sessions will be limited!
This session will consist of a presentation covering an introduction to Singularity (about 30-45 minutes depending on questions) and a lab follow up with the remainder of the time.
The lab session will take the session attendees through an installation of Singularity and then a walk through of some of the basic features of building and using Singularity containers. To do this we will provide the lab members with temporary AWS compute instances that everyone will have access to for the duration of the session(s). This lab will provide attendees with the following knowledge:
- How to achieve reproducible scientific results using Singularity
- How to install and setup Singularity
- An introduction to bootstrapping (creating/building) and customizing Singularity containers
- Basic usage of Singularity containers
- How to leverage existing containers and work pushed to DockerHub
To be part of this lab, you will have to have each of the following:
- You must have a general working knowledge of Linux and be familiar with compiling and installing GNU source build packages, installing system packages and updates, using
ssh, and be fluent with the command line and a basic understanding of computing science - You must be either a scientist, engineer, or system administrator (this is a technical workshop)
- It is necessary that you have a laptop with you that is capable of web browsing and ssh’ing into remote hosts
This is the advanced session for Singularity and will include a short presentation about some of the architecture of Singularity, more advanced features, and points on system integration before continuing onto the lab which will focus on more advanced usage of Singularity including:
- Customizing bootstrap definitions to make reproducible containers based on prescriptive recipes
- Creating workflows within Singularity containers that do specific jobs and how to execute those recipes
- Running MPI through your container
- How to integrate with resource managers and job scripts
This lab will extend upon what we did with Session 1, so it is best that you either have been at the first session or you already have a general working knowledge of Singularity. We will also continue using the AWS instances from the first Session, so be sure to bring your ssh enabled laptop! If you did not attend the first session but wish to attend this session, be sure to bring your own knowledge on the points covered in the Session 1 overview.
If there are any questions, comments, concerns or feedback, related to these labs, please contact Gregory Kurtzer at [email protected].