NHR Graduate School Course Week 2023
Topic outline
-
The Summer School, organized by the NHR Alliance and the NHR centers at Paderborn, Aachen, and Karlsruhe, aims to convey HPC knowledge and software development skills to the PhD students of the NHR Graduate School.
Date: June 12-16, 2023
Place: Paderborn Center for Parallel Computing (PC2), Mersinweg 5, 33100 Paderborn (View on map)
This Moodle page provides all necessary information to the attendees, including (links to) lecture slides and miscellaneous material.
Travel Advice
Bus from Paderborn main station to University:
Local busses in Paderborn are operated by PaderSprinter. Leaving the train station you find the bus station to your right hand side on the same side of the road. Take line 4 towards Dahl or line 9 towards Kaukenberg and leave the bus at exit Uni/Südring. During the teaching term, there is also a special Uni Linie service connecting the main station and the University. You find the exact schedule and fare information at the pages of the PaderSprinter (German only).
After leaving the bus at the Uni/Südring stop, you can reach us in about 5 minutes by walking (see the UPB campus map). Follow the Südring street (into the direction of Südringcenter/Husener Str.) until you reach building L at the intersection of Südring and Pohlweg, then turn right into Pohlweg. Immediately after the crosswalk, take the first left into Mersinweg. Building X is easily recognizable by its antracite-colored facade.
On-site Wifi
The recommended and simplest way to log in to the university Wifi on campus is via eduroam. If you already use it at your home institution, it should work automatically on the UPB campus as well. If you don't use it yet, follow the setup instructions of your home institution (ideally prior to your travel).
Alternatively, we can hand out temporary Wifi access accounts that are valid for the course week. Reach out to Lukas (lukas.mazur@uni-paderborn.de) or Carsten (carsten.bauer@uni-paderborn.de) if you need one.
Important links
PC2 JupyterHub: https://jh.pc2.uni-paderborn.de/
-
14:00-17:00 Welcome Event
Location: Paderborn Center for Parallel Computing (PC2), Room X0.101
Agenda
- 14:00 - Welcome address by Prof. Dr. Christian Plessl (director of PC2 and chairman of the NHR association board)
- 14:20 - Getting to know each other
- The fellows introduce themselves and their research plans (5 min. each)
- 2nd cohort (started 2023)
- Short break
- 1st cohort (started 2022)
- The fellows introduce themselves and their research plans (5 min. each)
- 16:00 - Guided tour through the compute centre
From 17:00 Informal Get-Together
There will be beverages, pretzels and other small snacks.
-
09:00-16:00 Julia for High-Performance Computing
Location: Paderborn Center for Parallel Computing (PC2), Room E5.333
Lecturers: Dr. Carsten Bauer (PC2), Dr. Xin Wu (PC2)
Material: https://git.uni-paderborn.de/pc2-public/julia-nhr-courseweek-2023
Agenda
- 09:00-09:30 Introduction, Why Julia?
- 09:30-10:30 Julia fundamentals
- 10:30-10:45 Break
- 10:45-11:15 Generic programming
- 11:15-12:00 Guided Exercise
- 12:00-13:00 Lunch break
- 13:00-14:00 Parallelisation
- Multithreading
- GPU acceleration
- 14:00-16:00 Guided Exercises
- 16:00 Walk into the City to the Tourist Information
This day provides a condensed excerpt of our "Julia for High-Performance Computing" course, last given at HLRS in Stuttgart.
Prerequisites / Preparation: You should be familiar with the basics of Jupyter and the command line. Ideally, you've already downloaded and played around with Julia before the course.
From 16:30 Guided Paderborn City Tour
Start and End: Tourist Information, Marienplatz, Paderborn (view on map)
Language: English
-
09:00-16:40 Continuous Correctness Checking your HPC Application
Location: Paderborn Center for Parallel Computing (PC2), Room E5.333
Lecturers: Joachim Jenke (RWTH), Alexander Hück (TUDa), Simon Schwitanski (RWTH), Adrian Schmitz (RWTH)
Slides: https://git.uni-paderborn.de/nhr-summerschool-correctness-and-ci/material
Agenda
- 09:00 - 09:15 - Introduction + Hands-On Setup
- 09:15 - 09:45 - MPI Correctness Checking with MUST
- 09:45 - 10:30 - Hands-on 1: Running MUST on a Simple Example Code (incl. Coffee Break)
- 10:30 - 11:00 - MUST: Advanced Options and Workflows
- 11:00 - 12:00 - Compiler-Supported Analysis: Archer and TypeART
- 12:00 - 13:00 - Lunch break
- 13:00 - 14:00 - Hands-on 2: Applying Archer and TypeART to LULESH
- 14:00 - 14:30 - From Intuition to Systematic Debugging
- 14:30 - 15:00 - Continuous Integration of HPC Applications in GitLab (incl. Coffee Break)
- 15:00 - 15:30 - Hands-on 3: Hello World in CI
- 15:30 - 16:00 - Continuous Integration with MUST
- 16:00 - 16:30 - Hands-on 4: CI with MUST on LULESH
- 16:30 - 16:40 - Discussion and Wrap-up
Prerequisites / Preparation: You should be familiar with the basics of the command line and GitLab.From 16:40 Free Time
- 09:00 - 09:15 - Introduction + Hands-On Setup
-
09:00-16:00 Data Parallel Neural Networks
Location: Paderborn Center for Parallel Computing (PC2), Room E5.333
Lecturers: Dr. Charlotte Debus (KIT), Dr. Marie Weiel (KIT), David Li (KIT)
Materials https://github.com/mcw92/nhr-summerschool
Agenda
- 09:00 - 10:15 - Introduction to Neural Networks
- Backpropagation and Stochastic Gradient Descent
- Layer Architectures
- Training a Neural Network
- 10:30 - 12:00 - Hands-on Session
- Neural Networks with PyTorch
- 12:00 - 13:00 - Lunch break
- 13:00 - 14:15 - Data-parallel Neural Networks
- Parallelisation Strategies for Neural Networks
- Distributed SGD
- IID and Large Minibatch Effects
- 14:30 - 16:00 - Hands-on Session
- PyTorch Distributed Dataparallel
- 16.00 Walk into the City to the LWL Museum
Prerequisites / Preparation:
You should be familiar with basic Python and Jupyter. A basic understanding of machine-learning techniques and gradient-based optimisation is beneficial. Furthermore you should have basic knowledge on distributed computing and communication via MPI.
From 17:00 Archäologische Spurensuche
Start and End: LWL Museum Kaiserpfalz, Paderborn (View on map)
Language: English
After that dinner together at Kitzgams, Paderborn (View on map)
- 09:00 - 10:15 - Introduction to Neural Networks
-
09:00-16:00 Final Day
Location: Paderborn Center for Parallel Computing (PC2), Room X0.101 and X0.207
Lecturers: Konstanze Bittmann - Communication, Nora Benariba + Lawrence Oduro-Sarpong (glokal e.V.) - Antibias
Agenda
- 09:00-14:00 - 1st. cohort: Communication Workshop
- 12:00 - 12:30 Lunch
- 09:00-14:00 - 2st. cohort: Antibias Workshop
- 12:00 - 12:30 Lunch
15:00 Departure