Young Events and Training

MMM Hub Conference & User Meeting 2023

8th and 9th November 2023

HPE Customer Innovation Centre
Techworks, 1 Aldermanbury Square, London, EC2V 7HR

GPU Training

PAX-HPC General Training on Parallel I/O and GPU Programming Models

17th-19th January 2024 – Room G.03, Bayes Centre, The University of Edinburgh, 47 Potterrow, Edinburgh, EH8 9BT 

Registration: https://bit.ly/3SMZ18e

This will be a 3-day in person training event. The main training contents will focus on Parallel I/O and GPU programming models. Detailed programme/agenda will be available soon. 

 

Please complete the following registration form to attend the training. I'll much appreciate if you could try to complete the registration from by Tuesday 5th December where possible. It will help us have a better idea of the total number of PAX-HPC attendees, and will also allow us some time to prepare for the system access for our training in January.

 

MMM Hub: HPE / NVIDIA GPU Training Day – 31 March 2022

Recent developments at the MMM Hub will see the addition of 6 nodes of 8 x A100 GPUs to the machine at the end of March 2022, after the team who lead the Hub was awarded additional funding from EPSRC in November 2021.

To familiarise MMM Hub users with the upgraded system, HPE, NVIDIA and the MMM Hub ran a training day on 31 March 2022.

Watch recordings of the sessions using this link

09:00 – 12:00 (UTC) Technology and Partner sessions (UCL, HPE & NVIDIA) chaired by Owain Kenway – UCL
Overview of the new MMM Hub system – UCL &HPE
Overview of the CRAY programming environment – HPE
Multi-GPU programming with MPI and NCCL – NVIDIA
Nighty Systems and Nsight Compute tools – NVIDIA

13:00 – 16:00 (UTC) Materials community codes, experiences & lessons learned (invited speakers) chaired by Filippo Spiga – NVIDIA
VASP – NVIDIA
Questaal – King’s
QMCpack – Argonne
CP2K – CSCS
CASTEP – York

Software Spotlights

We are organising a series of 'Software Spotlight' events, with the aim of both showing off the capabilities of a particular software package from a research perspective, as well as spending some time looking at exactly how the code can be run in practice – especially on Young.

Please see the activities tab for all upcoming events. Links to recordings of past events will be listed here:

  • CASTEP, Phil Hasnip on Tuesday 19th September 2023 [link to video]
  • Chemshell, given by You Lu on Tuesday 25th April 2023 [link to video]
  • LAMMPS, given by Nicodemo Di Pasquale on Thursday 19th January 2023 [link to video]
  • VASP, given by Bonan Zhu on Thursday 20th October 2022 [link to video]
  • CP2K, given by Matt Watkins on Thursday 17th February 2022 [link to video]
  • ONETEP, given by Chris-Kriton Skylaris, Jacek Dziedzic and Arihant Bhandari on Friday 2nd July 2021 [link to video]
  • DL_Poly MD software, given by Alin Marin Elena on Wednesday 5 May 2021 [link to video] [slides]

Future talks aim to include commonly codes used on Young, such as VASP, Quantum ESPRESSO, LAMMPS, Casino and include some emerging technologies such as machine learning with Keras, Tensorflow and Torch

The format of the events are in two parts:
1. Scientific motivation, and research highlights using results obtained from the code being presented (20 minutes inc. questions).
2. Discussion of the code functionality and capabilities, with a tutorial on technical aspects of using the code, including input/output, recommended job sizes, common problems, etc with a live demo running on 'Young'/Tier-2 HPC architectures using the module environment.

We expect this kind of format to be useful for students and researchers to get started with codes of interest for many years to come, and we will record the sessions, and place them here on the MMM Hub website. These events may be split between two speakers – perhaps with one speaker giving the research highlight, while the second speaker focusing on the technical aspects in running the code, making appropriate input files, and interpreting the output.

If you'd be happy to deliver one of these sessions, even if your code is less commonly used, please get in touch

Software Training

After discussion about how best to provide training, together with survey input from the users on Thomas and Young, we have run a number of event, for example the Software Spotlights above, together with some training videos on best practice use of Young:

  1. A basic introduction video to Young is here, which includes details of hardware on Young, how to submit jobs and an overview of types of parallelism.
  2. A quick 4 minute overview of how to choose memory is here,
  3. Longer videos on memory allocation are here for new users with a hands-on example here.

The excellent documentation provided on the UCL website here explains how to get started and explains logging in with examples of how to submit jobs. However, many users have asked for general background such as bash, compiling, MPI, etc. Much of this background is not unique to Young and has been covered elsewhere, so we provide a list of existing resources below. These are from the UK national supercomputer ARCHER and it’s successor ARCHER2 – the basic skills are transferable with slight changes for Young

https://www.archer2.ac.uk/training/materials/

http://www.archer.ac.uk/training/past_courses.php

https://www.hpc-uk.ac.uk/training/

e.g. for MPI (https://www.archer2.ac.uk/training/courses/200514-mpi/)

Software Carpentry provides lots on basic computing skills, bash, version control, Python

https://software-carpentry.org/lessons/

The Software sustainability Institute provides great advice on improving research software

https://www.software.ac.uk/

The N8 CIR machine has a detailed set of training

https://n8cir.org.uk/events/online-training/

Note that Young has recently been expanded to include 8 GPU nodes, if you are interested in using larger numbers of GPUs in your research, you can apply for time on N8 CIR and learn more on GPU programming here.

Any suggestions for this page are welcome, both for material you'd like us to generate and suggestions for sources which you've found useful. If you feel a particular skill or software which you need for Young is not covered, please get in touch and we can discuss how to address this with training.

Contact

If you have any suggestions for training, would like to help by running a course on software you use on Young, have a software package you'd like to give a spotlight talk on or have any other questions, please contact the coordinator of events and training for the Materials and Molecular Modelling Hub consortium (MMM Hub): George Booth or Edward Smith