June 2022 / Intro to Scientific Computing / HPC Summer Kickstart
Recordings are at this YouTube playlist.
All HackMD Q&A is archived here: https://hackmd.io/@AaltoSciComp/ArchiveIntroSummer2022
Thank you for attending the course! Do you wonder what to do next?
Check the “preparation” and “follow-up” sections below, browse whatever content you have missed that might be useful to you.
Know that we at Aalto (and other places) are here for you, and you can drop by and ask further questions.
The Hands-on Scientific Computing course (if you are in Finland) can give you credits for topics including what we learned here.
Day 1 summary: We covered a bunch of high-level topics for new researchers, see the playlist. For day 2 (Thursday), make sure you can connect to the cluster - see the schedule for details.
Day 2 summary: We did what was on the schedule. There was a lot of discussion in the sessions, recordings show the discussion + solutions to the exercises. No extra preparation needed for tomorrow, but the hands-on work gets more involved and interesting.
Day 3 summary: we looked at different ways to make computation faster by parallelizing it, not at the level of writing new parallel code but using parallel code. We heard from CSC.
Date: 7, 9-10 June 2022, 11:50-16:00 EEST (Helsinki time).
Join via Twitch: https://twitch.tv/coderefinery
Kickstart is a three × half day course for researchers to get started with high-performance computing (HPC) clusters. The first day serves as a guide to your career: a map to the types of resources that are available and skills you may need in your career, so that you can be prepared when you need more in the future. This part is especially suitable to new researchers or students trying to understand computational/data analysis options available to them. It won’t go into anything too deep, but will provide you with a good background for your next steps: you will know what resources are available and know the next steps to use them.
The second and third days take you from being a new user to being competent to run your code at a larger scale than you could before. This part is good for any researcher who thinks they may need to scale up to larger resources in the next six months, in any field. Even if you don’t use computing clusters, you will be better prepared to understand how computing works on other systems. If you are a student, this is an investment in your skills. By the end of the course you get the hints, ready solutions and copy/paste examples on how to find, run and monitor your applications, and manage your data.
If you are at Aalto University: the course is obligatory for all new Triton users and recommended to all interested in the field.
If you are not at Aalto University, you can follow along with the course and will learn many things anyway. The course is designed to be as useful to people outside of Aalto, but some of the examples won’t directly work on your cluster (most will, anyway we will give hints about adapting). Known sites partnering with this course:
Regardless of where you are from, you may register using the registration form below to get emails about the course.
Participants from University of Helsinki can follow how to connect to their Kale/Turso cluster by following their own instructions.
Participants from University of Oulu: please follow instructions on how to access the Carpo2 computing cluster.
Tampere: this course is recommended for all new Narvi users and also all interested in HPC. Most things should work with simply replacing triton -> narvi. Some differences in configuration are listed in Narvi differences
This is an online hybrid of MOOC and interactive: everyone may attend the livestream at https://twitch.tv/coderefinery, no registration needed! This is the primary way to watch all sessions. Zoom is used for exercise sessions of partner audiences, and HackMD is used for a continuous Q&A session.
Time, date: 7 and 9-10 June (Tue, Thu-Fri). 11:50-16:00 EEST
Place: Online via public livestream, Zoom exercise sessions for partners.
Registration: Please register at this link https://link.webropol.com/s/scicompintrosummer2022. Attending individual sessions is fine.
Cost: Free of charge for FGCI consortium members including Aalto employees and students. Livestream is free to everyone.
Additional course info at: firstname.lastname@example.org
All times are EEST (Europe/Helsinki time)!
The daily schedule will be adjusted based on the audience’s questions. There will be frequent breaks and continuous questions time going on, this is the mass equivalent of an informal help session to get you started with the computing resources.
Subject to change
Schedule may still have minor updates, please check back for the latest.
Day #1 (Tue 7.jun): Basics and background
11:50–12:00: Connecting time, basics, and icebreaker
12:00–12:10 Introduction, about the course Richard Darst and other staff Materials: Summer Kickstart intro
12:10–12:25: From data storage to your science Enrico Glerean and Simo Tuomisto
Data is how most computational work starts, whether it is externally collected, simulation code, or generated. And these days, you can work on data even remotely, and these workflows aren’t obvious. We discuss how data storage choices lead to computational workflows. Materials: SciComp Intro
12:25–12:50: What is parallel computing? An analogy with cooking Enrico Glerean and Thomas Pfau
In workshops such as this, you will hear lots about parallel computing and how you need it, but rarely get a understandable introduction to how they relate and which are right for you. Here, we give a understandable metaphor with preparing large meals. Slides
13:00–13:25: Behind the scenes: the humans of scientific computing Richard Darst and ???
Who are we that provide these services. What makes it such a fascinating career? Learn about what goes on behind the scenes and how you could join us.
13:25–13:50: How you actually install software on the cluster: an example case (Conda and Python) Simo Tuomisto and ???
Software installation is one of the questions we most often get. Usually, on clusters, this happens via environments, which allows you to install specific software per-project. We’ll give a demonstration of how these work in Python. Materials for demo: Python Environments with Conda
14:00–14:50: Secure Shell (ssh) tips and tricks Thomas Pfau and Enrico Glerean
Remembering server address… Another login? Another password prompt? Again? Wouldn’t it be nice to just have a key instead of a keycode that you need to type in? Here, we will show you how to set up your computer to easily connect to the server(s) you need. And we will explain the process from keys to config. While useful, this part is skippable if you are able to connect to Triton (next section). Materials: SSH
15:00–15:45: Connecting to a HPC cluster Thomas Pfau and Simo Tuomisto
Required if you are attending the Triton/HPC tutorials the following days, otherwise the day is done.
15:00–15:20?: Livestream introduction to connecting
15:??–??: Individual help time in Zoom (links sent to registered participants)
Material: Connecting to Triton
Preparation for day 2:
Remember to read/watch the “shell crash course” (see “Preparation” below) if you are not yet confident with the command line. This will be useful for tomorrow.
Day #2 (Thu 9.jun): Basic use of a cluster (Richard Darst, Simo Tuomisto)
11:50–12:00: Connecting time and icebreaker
12:00–12:05: Introduction to days 2-3
12:05–12:30: What can you do with a computational cluster?
Several real examples of how people use the cluster (what you can do at the end of the course).
Real example 1: Large-scale computing with array jobs
Real example 2: Large-scale parallel computing
12:30–15:00: Running your first jobs in the queue
15:00–15:30: Other things you should know about the HPC environment
Day #3 (Fri 10.jun): Advanced cluster use (Simo Tuomisto, Richard Darst)
11:50: Joining time/icebreaker
11:50–13:00: Simple parallelization with array jobs: Array jobs: embarassingly parallel execution
13:00–14:00: Using more than one CPU at the same time: Parallel computing: different methods explained
14:00–14:30: Laptops to Lumi
You now know of basics of using a computing cluster. What if you need more than what a university can provide? CSC (and other national computing centers) have even more resources, and this is a tour of them. Slides here.
14:40–15:30: Running jobs that can utilize GPU hardware: GPU computing
We strongly recommend you are familiar with the Linux command line. Browsing the following material is sufficient:
How to attend: Online workshops can be a productive format, but it takes some effort to get ready. Browse these resources:
Attending a livestream workshop, good to read in detail (ignore the CodeRefinery-specific parts).
SSH client to connect to the cluster (+ be able to connect, see next point)
Zoom (if attending breakout rooms)
Cluster account and connection verification:
Next steps / follow-up courses
Keep the Triton quick reference close (or equivalent for your cluster).
Each year the first day has varying topics presented. We don’t repeat these every year, but we strongly recommend that you watch some of these videos yourself as preparation.
Very strongly recommended:
Other useful material in previous versions of this course:
While not an official part of this course, we suggest these videos (co-produced by our staff) as a follow-up perspective:
Attend a CodeRefinery workshop, which teaches more useful tools for scientific software development.
Look at Hands-on Scientific Computing for an online course to either browse or take for credits.
Cluster Etiquette (in Research Software Hour): The Summer Kickstart teaches what you can do from this course, but what should you do to be a good user.
How to tame the cluster (in Research Software Hour). This mostly repeats the contents of this course, with a bit more discussion, and working one example from start to parallel.
We hope to make a good learning environment for everyone, and expect everyone to do their part for this. If there is anything we can do to support that, let us know.
If there is anything wrong, tell us right away - if you need to contact us privately, you can message the host on Zoom or contact us outside the course. This could be as simple as “speak louder / text on screen is unreadable / go slower” or as complex as “someone is distracting our group by discussing too advanced things”.
See the schedule