Virtual Tutorials and Webinars
This section contains details on the virtual tutorials and webinars offered by the ARCHER service.
These virtual sessions usually take place at 15:00 UK time on Wednesdays.
The virtual tutorials and webinars are live online interactive sessions conducted using Blackboard Collaborate. As well as presenting on selected topics in high-performance computing relevant to ARCHER users and reporting on scientific software development work done as part of ARCHER eCSE projects, these sessions provide a forum for users to ask questions on any topic related to the ARCHER service. They also provide an opportunity for attendees of recent ARCHER training courses to ask questions that may have arisen since attending the course.
Webinars usually have a presentation followed by the opportunity for attendees to ask questions.
Virtual tutorials normally start with a presentation and live demonstration on a particular topic followed by time for discussion with participants, both about the presented materials and in response to general questions about ARCHER and HPC/parallel computing in general. We hope that the discussions will be driven by what users wish to know and we expect the format to be that of a Question and Answer (Q&A) session.
Upcoming Sessions
Chrome or Firefox are recommendedfor connecting to BlackBoard Collaborate but most other modern browsers will work to attend a session - please see our
BlackBoard Collaborate page for more information and advice on testing your connection
prior to the tutorial.
There are also apps available for BlackBoard Collaborate but these have reduced functionality and we recommend using the browser where this is an option.
BlackBoard provide lots of help and documentation if you have any questions or issues.
There are also apps available for BlackBoard Collaborate but these have reduced functionality and we recommend using the browser where this is an option.
BlackBoard provide lots of help and documentation if you have any questions or issues.
| Title | Description | Type | Associated files | Date |
|---|---|---|---|---|
| None Scheduled |
Previous Sessions
Slides and video/audio recordings are available for some of the virtual tutorial and
webinar sessions. Most videos are hosted through the
ARCHER Youtube channel. Any .jar recordings should play automatically using Blackboard Collaborate. Please leave us feedback by clicking the 
icon to help us improve future courses.
| Title | Description | Type | Associated files | Date |
|---|---|---|---|---|
| EIS-2 - A general purpose, high performance input deck and maths parser | Christopher Brady, University of Warwick The EPOCH Input System Version 2 (EIS-2), available from https://github.com/csbrady-warwick/EIS-2 is a publicly available Fortran "input deck" reader and maths parser well suited to high performance computing. It was developed under an ARCHER eCSE grant as a more general, higher performance replacement for the existing deck reader and maths parser in the EPOCH particle in cell code developed by CCP-Plasma. As development was advancing it became clear that the code for EIS-2 was general enough that it would be useful as a stand-alone library so we added some user documentation and some additional features to expand the code beyond what was needed for EPOCH and released it. EIS-2 is a static library that allows codes to give a user extensive control over how the code runs through standard hierarchical block based input files and also allows a user to specify expressions through normal infix mathematical expressions. Released under the BSD three clause license, being written in standards conforming Fortran 2003 and having no library dependencies it can be incorporated into both open and closed source code bases. This webinar will cover the basic design of EIS-2, covering the lexer, the parser and the AST based simplifier for the maths parser and a quick overview of the deck parser. It will cover simple examples of making use of EIS-2 and how to incorporate it into existing code. It will give a brief overview of possible future work on EIS-2, including C interoperability allowing C and C++ codes to use EIS-2, extending the AST simplifier to optionally work with LLVM to provide near native code performance and using the information contained in the deck file structure to provide automatic interfaces to a code from supported scripting languages. |
Webinar |
Video Slides |
12th February 2020 15:00 |
| Cray Compilation Environment Update and modern Fortran |
Harvey Richardson, CRAY This webinar will outline some new developments in the Cray Programming Environment and will then focus on presenting examples of the latest features from Fortran 2018 that are supported by the Cray Fortran compiler. Note that these developments are not available on ARCHER today but this webinar should prove useful to people who have access to other Cray systems or who are planning to use the next national supercomputing service that will follow on from ARCHER. |
Virtual tutorial | Video | 5th February 2020 |
| Transferring Data from ARCHER at End of Service |
Andy Turner and Caoimhin Laoide-Kemp, EPCC, The University of Edinburgh The end of the ARCHER service is approaching and many users have large numbers of files and/or volumes of data on the service which may need to be moved to alternative locations before the end of service. Recently, we have updated the ARCHER Data Management Guide with useful advice on transferring data between different sites and systems. In this webinar, we will discuss the issues around data transfer, present practical advice to users about data transfer and answer questions and concerns users may have on issues around data management and transfer. |
Virtual Tutorial | Video Slides |
29th January 2020 |
| Enabling multi-node MPI parallelisation of the LISFLOOD flood inundation model |
Arno Proeme, EPCC The availability of higher-resolution topographic datasets covering greater spatial domains has soared in recent years, pushing the limits of computational resources beyond those typically found in regional HPC services. In addition, many countries that are subject to large-scale flooding each year do not have access to real-time flood forecasting software. This webinar describes how HAIL-CAESAR, a geosciences code that implements the LISFLOOD flood inundation model, was ported to make use of LibGeoDecomp - a C++ stencil code HPC library - to enable multi-node parallelism. Whilst currently single inundation scenarios can take multiple days to run using standard hydrological modelling software, this project paves the way for ensemble runs that can be initiated on the basis of a 24 or 48 hour rainfall forecast and complete within shorter timescales, which should ultimately have major implications for flood warnings in developing countries. |
Webinar | Video Slides |
4th December 2019 |
| A Fully Lagrangian Dynamical Core for the Met Office/NERC Cloud Model | Steven Boeing, University of Leeds and Gordon Gibb, EPCC We discuss the development of an essentially Lagrangian model of atmospheric cloud dynamics, carried out in the eCSE project �A fully Lagrangian Dynamical Core for the Met Office/NERC Cloud Model�. We describe the work, which aimed to incorporate the �Moist Parcel in Cell� (MPIC) code into the Met Office/NERC Cloud Model (MONC), thereby granting it MPI parallelism and the ability to scale beyond a single node. In the Moist-Parcel-In-Cell (MPIC) model, parcels represent both the thermodynamic and the dynamical properties of the flow. The representation of parcel properties is fully Lagrangian, but an efficient grid-based solver calculates parcel advection velocities. The Lagrangian approach of MPIC has a number of advantages: in particular, parcel properties are naturally conserved, and the amount of mixing between parcels can be explicitly controlled. MPIC also lends itself well to parallelisation, because most of the communication required between processors is local, and an efficient solver is available where global communication is required. First, we describe the modifications to MONC in order for it to be parcel-aware. Subsequently, we outline the implementation of MPIC�s physics into MONC. We then provide an in-depth comparison of MPIC with the newly developed code, and investigate how it scales to many thousands of MPI processes. Finally we discuss the limitations of the code, and future work to be carried out to mitigate these limitations. We found that the code can scale up to many thousands of cores for large problem sizes, although the main limiter of performance at scale are the Fourier Transform routines. Despite good MPI performance from the code, OpenMP performance is poor, achieving a speedup of only 3 on 12 threads. Overall, on a single node the new code performs better than MPIC does, carrying out more parcel operations per core per second. We also describe how users can install the code and develop new components. |
Webinar | Video Slides |
Wednesday 13th November 2019 |
| Enabling distributed kinetic Monte Carlo simulations for catalysis and materials science | Michail Stamatakis, UCL Heterogeneously catalysed reactions constitute more than 70% of all chemical synthesis processes, supporting the UK chemical industry with a turnover of ~£50 billion. Developing such processes, however, is non-trivial and currently relies largely upon trial-and-error experimentation as well as empirical models of catalytic reaction kinetics. On the other hand, a molecular-level understanding of the physical and chemical events giving rise to catalysis, such as adsorption/desorption, surface diffusion and reaction, can yield truly predictive models, making it possible to design catalytic materials and reactors from first-principles. Such fundamental understanding can be achieved by means of kinetic Monte Carlo (KMC) simulations, which can predict performance metrics, such as activity, selectivity, yield and stability of the catalyst. However, KMC algorithms are inherently serial and therefore difficult to parallelise. Thus, it is challenging to simulate systems which are subject to or exhibit spatial variations at long scales. To tackle this challenge, in the context of a recent eCSE project we have implemented the "Time Warp" algorithm in our software Zacros (http://zacros.org), a general-purpose Fortran-based KMC code for catalysis and materials science. In this webinar we will discuss the key idea behind the code and provide some details on the implementation, both at the algorithmic level as well as the technicalities of programming it in Fortran. We will further present benchmarks and discuss ideas for enhancing the performance of the code. |
Webinar | Slides
Video |
25th September 2019
|
| Good Practice for transferring data |
Andy Turner and Caoimhin Laoide-Kemp, EPCC, The University of Edinburgh The end of the ARCHER service is approaching and many users have large numbers of files and/or volumes of data on the service which may need to be moved to alternative locations before the end of service. Recently, we have updated the ARCHER Data Management Guide with useful advice on transferring data between different sites and systems. In this webinar, we will discuss the issues around data transfer, present practical advice to users about data transfer and answer questions and concerns users may have on issues around data management and transfer. |
Virtual Tutorial | Slides Video |
4th September 2019 |
| OpenMP on GPUs | Mark Bull, EPCC This VT is an introduction to programming GPUs using OpenMP's vendor-neutral target offload directives. It will cover the underlying offloading model, how to exploit parallelism on the device, how to manage data transfers between the host and device, and some basic performance issues. |
Virtual tutorial | Slides Video |
28th August 2019
|
| Modern Fortran | Adrian Jackson, EPCC, discusses the features of "modern" Fortran (Fortran90 and beyond), and the steps that
need to be considered to port older Fortran (primarily Fortran 77) to new Fortran standards. We also discuss
some of the newest Fortran features (in Fortran 2003 and beyond) such as Object-Oriented Programming (OOP). Whilst Fortran is generally viewed as a procedural programming language there are features in the most recent versions of the standards that enable development in OOP or OOP-like ways. We will introduce these language features and discuss how they can be used in large-scale scientific applications. |
Webinar | Video Slides |
26th June 2019 |
| Open Source HPC Benchmarking | Andy Turner, EPCC There is a large and continuing investment in HPC services around the UK, Europe and beyond and this, along with new processor technologies appearing on the HPC market, has led to a wider range of advanced computing architectures available to researchers. We have undertaken a comparative benchmarking exercise across a range of architectures to help improve our understanding of the performance characteristics of these platforms and help researchers choose the best services for different stages of their research workflows. We will present results comparing the performance of different architectures for traditional HPC applications (e.g. CFD, periodic electronic structure) and synthetic benchmarks (for assessing I/O and interconnect performance limits). We will also describe how we have used an open research model where all the results and analysis methodologies are publicly available at all times. We will comment on differences between architectures and demonstrate the benefits of working in an open way. |
Webinar | Video Slides |
22nd May 2019 |
| HPC-Europa3 research visits | HPC-Europa3 research visits: programme overview and visitor presentation Catherine Inglis, EPCC and Livio Carenza, Department of Physics, University of Bari HPC-Europa3 is an EC-funded programme which funds collaborative visits for European researchers in any discipline which can use High Performance Computing. UK-based researchers can benefit in two ways: by visiting a research group in another country, or by hosting a collaborative visit here in the UK. This webinar will begin with an overview of the programme and how to apply, presented by Catherine Inglis, the HPC-Europa3 Transnational Access Co-ordinator at EPCC. Following this overview, one of the current visitors will give a presentation on the work he has been doing during his visit. Lattice Boltzmann Methods for Active Fluids - Livio Carenza, Department of Physics, University of Bari (Italy) Abstract: Active Matter is a peculiar kind of Soft Matter whose constituents are able to inject energy in the system at the level of the individual constituents, so that active systems evolve inherently far from thermodynamic equilibrium. In this presentation I will provide an overview of continuum approaches suitable to model active matter in fluid environments, starting from a coarse-grained description of the active constituents and their active effect on the surrounding fluid. I will provide an overview of the basic features of Lattice Boltzmann, the numerical method implemented to integrate dynamical equations of active matter and present some details regarding the parallelization of the code used for simulations. Furthermore I will provide an overview of some interesting phenomena arising in active matter, ranging from Spontaneous Flow to Active Turbulence. |
Webinar |
Video Catherine Inglis' Presentation Slides Livio Carenza's Presentation Slides |
24th April 2019 |
| Implementation of multi-level contact detection in granular LAMMPS to enable efficient polydisperse DEM simulations | Tom Shire, University of Glasgow and Kevin Stratford, EPCC Discrete element modelling (DEM) is a computational tool for predicting how granular materials will respond during loading, flowing or other processes found in nature or in industry. Each grain in a system is modelled as an individual element, and the elements interact to model to overall system response. The most time consuming parts of a DEM simulation are calculating which grains are in contact with each other and communicating this information between processors when running simulations in parallel. In most DEM codes these stages become prohibitively time consuming when a wide range of particles are employed, as the number of potential inter-particle contacts dramatically increases. This eCSE project has implemented an improved contact detection scheme for DEM simulations of materials with a wide range of particle sizes into LAMMPS, a widely used open-source software which is especially efficient for massively parallel simulations which can be run using supercomputers such as ARCHER. The new scheme limits the search range over which particle-contacts must be checked by defining different ranges depending on the size of particles - e.g. potential contacts between two large particles use a larger search area than between a small and large or two small particles. This webinar will introduce the basics of DEM and the new method to improve detection for polydisperse materials. Following this preliminary results and future plans will be discussed. |
Webinar |
Video Kevin's Slides Tom's Slides |
Wednesday 20th March 2019 |
| Developing Dynamic Load Balancing library for wsiFoam |
Dr. Xiaohu Guo, STFC and Dr. Scott Brown, University of Plymouth present their recent eCSE work. Offshore and coastal engineering fields are using increasingly larger and more complex numerical simulations to model wave-structure interaction (WSI) problems, in order to improve understanding of safety and cost implications. Therefore, an efficient multi-region WSI toolbox, wsiFoam, is being developed within an open-source community- serving numerical wave tank facility based on the computational fluid dynamics (CFD) code OpenFOAM?R , as part of the Collaborative Computational Project in Wave Structure Interaction (CCP-WSI). However, even using the efficiency of a multi-region approach, the computational expense of CFD is high, and hence there is a constant need to improve the efficiency of these high-fidelity codes. One way that this can be achieved is through the parallelisation of code to make use of high performance computing facilities. Although OpenFOAM?R is parallel ready, historically the MPI performance has been considered sub-optimal, but recent developments have led to a number of performance improvements being implemented in OpenFOAM?R v5.x along with new parallel I/O functionality. The developments have led to a significant performance benefit when employing a large number of processors, and it is vital that existing code is updated to be compatible with OpenFOAM?R v5.x in order to utilise this functionality. However, OpenFOAM?R v5.x still only offers static domain decomposition, which limits the choice of parallel load balancing methods to algorithms which can only take basic user defined arguments to aid in load balancing. These methods typically use blocking MPI communications that only consider the number of mesh cells and their spatial distribution (i.e. they assume homogeneity of load per mesh cell). As typical WSI simulations are often inhomogeneous with respect to the mesh, due to properties such as mesh motion and wave �relaxation�, these decomposition methods based purely on mesh resolution are likely to be sub-optimal in terms of computing resource usage. Therefore, in this work the wsiFoam toolbox is updated to utilise the new parallel I/O functionality in OpenFOAM?R v5.x. Furthermore, it is coupled with a newly implemented dynamic load balancing method, based on a new ParMETIS decomposition class, that considers all of the relevant parameters is required for efficient simulation and optimal par- allel performance of WSI problems. Benchmarking of the new functionality shows that the load imbalance is a major performance bottleneck for dynamic mesh applications but substantial improvements in computational efficiency (up to 5 times speed up) and parallel scaling have been observed through use of the new library for WSI applications. |
Webinar |
Video |
Wednesday 6th February 2019 |
| Isambard: tales from the world's first Arm-based supercomputer | James Price, University of Bristol The Isambard Project, a GW4 Alliance initiative, has evaluated the performance of its Arm-based processors for HPC, the first such results available for dual socket Cavium ThunderX2 nodes. James Price will describe the results of a detailed performance comparison between Cavium ThunderX2 Arm-based CPUs with the latest state-of-the-art Skylake x86 processors. Results focus on the HPC codes that are most heavily used on the UK�s national supercomputer, ARCHER, and show that for these kinds of workloads, ThunderX2 is competitive with the best x86 CPUs available today but with a significant cost advantage. |
Webinar |
Video Slides |
Wednesday 12th December 2018 |
| TPLS and the Use of PETSc | David Scott, EPCC The capabilities of a program called TPLS (Two Phase Level Set) will be described with particular emphasis on how PETSc (Portable, Extensible Toolkit for Scientific Computation) is used in the code. |
Webinar | Video Slides |
Wednesday 31st October 2018 |
| Benefits Realisation | Lorna Smith, EPCC The eCSE programme has allocated funding to the RSE community through a series of regular funding calls over the past four to five years. The programme has funded RSE's from across the UK to develop and enhance software for the UK's National HPC Service, ARCHER and for the wider computational science community. The programme is funded by EPSRC and NERC and understanding and quantifying the benefits of the eCSE programme is key to justifying the expenditure on the programme and to securing future funding. We have identified a set of benefits associated with the programme and developed procedures to measure and showcase these benefits. One component of this has been to develop a process to measure and quantify the financial return on investment, based on performance improvements to the code and on post project utilisation figures. This form of financial benefit is an important one, forming part of a portfolio of benefits, benefits that include scientific improvements and achievement. |
Webinar | Slides
Video
|
Wednesday 12th September 2018 |
| An Introduction to using Version Control with Git | Arno Proeme EPCC. Git is now the most commonly used version control tool for managing the development of scientific software. This webinar will give a practical introduction to using Git, be it to start keeping track of your own code or to contribute to an existing piece of software. We will cover basic concepts and terms, some common workflows, and will look briefly at the functionality of GitHub and GitLab. |
Virtual Tutorial |
Slides
Video
|
Wednesday 25th July 2018 |
| Modern C++ for Computational Scientists | Rupert Nash, EPCC, will present a four-part tutorial, over two afternoons, on the use of Modern C++ for Computational Scientist. Since the 2011 revision to the C++ language and standard library, the ways it is now being used are quite different. Used well, these features enable the programmer to write elegant, reusable and portable code that runs efficiently on a variety of architectures. However it is still a very large and complex tool. This set of online lectures, delivered over two Wednesday afternoons and including practical exercises, will cover a minimal set of features to allow an experienced non-C++ programmer to get to grips with language. These include: overloading, templates, containers, iterators, lambdas and standard algorithms. It concludes with a brief discussion of modern frameworks for portable parallel performance which are commonly implemented in C++. The course would appeal to existing C++ programmers wanting to learn techniques that are applicable to numerical computing, or C programmers who want to know what parts of the C++ standard they should prioritise when learning new features. |
Virtual tutorial |
Practical Materials Slides1 Video1 Slides2 Video2 Slides3 Video3 Slides4 Video4
|
Wednesday 13th June 2018 and Wednesday 27th June 2018 |
| Cray Programming Environment Update | Harvey Richardson and Karthee Sivalingam, CRAY, will give an overview of recent and forthcoming changes and updates in the Cray programming environment available on ARCHER. They will cover the Cray compilers, MPI and performance tools. |
Webinar | Slides available to ARCHER users - please contact the ARCHER helpdesk to request these. Video
|
Wednesday 16th May 2018 |
| HPC-Europa: EC-funded collaborative research visits using High Performance Computing | Catherine Inglis, EPCC HPC-Europa is an interdisciplinary EC-funded programme which allows researchers to carry out short "transnational access" collaborative visits to a research group working in a similar field. Visits can last between 3 weeks and 3 months, and the possible destinations include HPC centres in Italy, Spain, Germany, Netherlands, Finland, Greece, Sweden and Ireland (in addition to EPCC). The programme is open to researchers of all levels, from postgraduate students to the most senior professors. Applicants can be working in any discipline, but MUST require the use of High Performance Computing facilities (for large-scale computer modelling, etc). Closing dates are held 4 times per year. Attend this webinar to find out more about the programme in time to prepare an application for the next closing date on 17th May. |
Webinar | HPC Europa Website Video
|
Wednesday 18th April 2018 |
| Meltdown for Dummies: The road to hell is paved with good intentions | David Henty, EPCC The history of IT security seems littered with tales of apocalyptic vulnerabilities that would end the world - does your Heart Bleed at the mention of the Y2K bug? However, the recent scares over Meltdown were based on a really big problem: ANY user could read ANY memory on almost ANY system, bypassing ALL security measures. In this webinar I will try and explain the origin of Meltdown from a dummy's point of view (i.e. my own). I will use an everyday analogy of accessing documents from a lawyer's office to illustrate the origins of the vulnerability and how it can be exploited. I will also present results of recent tests exploring how much the anti-Meltdown patches affect HPC system performance. Meltdown is an extremely interesting example as, to understand it, you need to understand a wide range of issues covering both CPU hardware and system software design. Even if you're not that concerned with IT security, I hope that this presentation will help explain how modern multi-user computers actually work. Most importantly, I hope to show that the design decisions that led to Meltdown all seem perfectly sensible in themselves. However, as Saint Bernard of Clairvaux wrote in the early 12th century, "the road to hell is paved with good intentions". |
Webinar | Slides Note: the original presentation included several animations which cannot be included here. They are, however, included as part of the video recording. Video
|
Wednesday 28 March 2018 |
| Getting Access to UK Advanced Computational Facilities for your Research |
Andy Turner, EPCC, University of Edinburgh Over the past five years there has been a large investment in computational facilities from the UK research councils leading to an unprecedented level of choice for researchers wanting to use computational tools in their research. These include the UK National Supercomputing Service, ARCHER; the UK Research Data Facility; the EPSRC national Tier-2 HPC resources; and the DiRAC STFC national HPC service (as well as several other facilities). These resources provide exciting opportunities for new research but the wide range on offer can lead to confusion on what is available, how to get access to it and how to use it effectively. EPCC has been leading initiatives (including HPC-UK) to help researchers discover what is available, how to get access, and providing support and training to researchers. In this webinar I will give an overview of the facilities available and how to get access to them - particularly through the current open calls for access via the Resource Allocation Panel (RAP). There will be plenty of time for questions at the end of the webinar. |
Webinar | Slides Video
|
Wednesday 7th March |
| State of the Art I/O Tools | Elsa Gonsiorowski, Lawrence Livermore National Laboratory I/O has historically been an afterthought in application development and performance analysis. As processing power increases and accelerators increase performance, Amdahl's law increasingly points to I/O as an opportunity for performance optimization. Unfortunately, the tool framework for analysis of I/O performance is not as robust as that for computational performance. Tools that exist, such as IOR and Darshan, target a very small slice of I/O performance. Additionally complicating this space is the introduction of hierarchical storage into the I/O space. With the increased prevalence of burst buffers and other forms of hierarchical storage, we can expect to find new answers to historically well-studied problems such as optimal checkpointing frequency. This talk will cover basic I/O strategies and cover state-of-the-art I/O tools that can be used to prepare for burst buffer architectures. It will include MACSio (an I/O benchmark emulating physics-simulations) and the Scalable / Checkpoint Restart (SCR) library. |
Virtual tutorial | Slides Video
|
Wednesday 28th February 2018 |
| Massively Parallel OpenMP-MPI Implementation of the SPH Code DualSPHysics | Athanasios Mokos, University of Manchester Smoothed Particle Hydrodynamics (SPH) is a meshless computational technique that is fast becoming a popular methodology to tackle industrial applications, especially violent free-surface/hydrodynamic flows. However, most practical problems are three-dimensional, requiring tens of millions of particles to be simulated. A serial code lacks the computational power to resolve these problems in a timely manner, so most SPH codes currently use the inherent parallelism of the GPU. There are however many users, in particular industrial users, who have access to conventional high-performance computing (HPC) clusters with CPU-only nodes and are reluctant to invest resources in purchasing GPU hardware and personnel resources. Hence, developing a massively-parallel-CPU SPH code is attractive for more widespread use. We have elected to use the open source SPH software package DualSPHysics, at present the most widely used SPH code. DualSPHysics, developed by the universities of Manchester, Vigo and Parma, has been developed and optimised in close collaboration with existing academic and industrial users, improving its usability and its modularity for further development. It has already been used for the simulation of violent, highly transient flows such as slam forces due to waves on offshore and coastal structures, impact loads on high speed ships, ditching of aircraft, sloshing of fuel tanks and many more. We have integrated Message Passing Interface (MPI) functionality to the DualSPHysics C++ CPU code, enabling the code to use multiple processing nodes and making it equivalent with the GPU version. However, the introduction of MPI creates new limitations as using multiple nodes (especially over 100 cores) has created software scalability problems which are addressed by the inclusion of the Zoltan package, which is scalable up to 10^6 cores. Its integration in DualSPHysics provides efficient algorithms for domain decomposition and load balancing such as the Hilbert Space Filling Curve (HSFC). This algorithm is an efficient solution for SPH as it maintains spatial locality which is critical for neighbour searching. To reduce the memory footprint, the HSFC is not directly applied to particles but to cells of 2h x 2h to create a cell-linked neighbour list as in the current version of DualSPHysics. A halo exchange algorithm using unstructured communication will be used to transfer neighbouring particle data from different nodes. This is an ongoing project, with future work focusing on application on 3D flows and optimisation with the aim of 90% efficiency for 1024 cores. |
Virtual tutorial |
Video Slides
|
Wednesday 24th January 2018 |
| Diversifying Your Workforce | Toni Collis, EPCC The under-representation of women and minorities is a challenge that the entire supercomputing industry faces. As a community we are only just beginning to measure and understand how 'leaky' our pipeline is, but attrition rates are likely as high as the general tech community: 41% of women working in tech eventually leave the field (compared to just 17% of men). This session will discuss the work being carried out by Women in HPC to diversify HPC. I will discuss the challenges we face and suggestions for making useful changes to improve inclusion in the workplace, benefitting everyone including those that are currently under-represented. |
Virtual tutorial | Slides Video
|
Wednesday 6th December 2017 |
| OpenFOAM | Fred Mendonca, OpenCFD Ltd OpenFOAM is the freely available open-source CFD toolkit developed, maintained, released and trademark owned by OpenCFD Ltd; it has been used successfully by a number of research groups on ARCHER. The latest version OpenFOAM-v1706, released on 30th June, is now available - see installation and guidance notes at http://openfoam.com/download/. This webinar presents the latest OpenFOAM features including accessibility, performance and development/upgrade/GitLab-repository toolkits, with the opportunity for Q&A. OpenCFD Ltd has been developing and releasing OpenFOAM since its debut in 2004 and owns the OpenFOAM Trademark. OpenCFD Ltd is a wholly owned subsidiary of the ESI Group. |
Virtual tutorial |
Slides Using OpenFOAM on ARCHER hints Video
|
11th October 2017 |
| OpenMP 4 | Mark Bull EPCC The latest versions of the compilers on ARCHER now all support OpenMP 4.0. This webinar will give a tour of the new features in this version of the OpenMP standard and some examples of how they can be used. |
Virtual tutorial | Slides Video
|
Wednesday 27th September 2017 |
| Test Your Knowledge! MPI Quiz (quiz-master: David Henty, EPCC) |
Using the basic MPI functions, such as point-to-point send and
receive, is relatively straightforward. In most cases, the default
behaviour of MPI means that simple programs function correctly; these
good design choices are one of the reasons why the library has proved
so popular over the years. However, even these seemingly simple functions can be a little more complicated than they first seem. For example, if you issue a send and there is no matching receive then what happens depends on exactly how you initiated the send. Understanding the defined behaviour in these slightly more complicated situations can be very helpful in ensuring your MPI programs function correctly in all possible circumstances. This webinar will take the form of an interactive online multiple choice Q&A session which you can undertake anonymously if you wish. The questions are partly designed for fun to test attendees' knowledge, but are mainly aimed at promoting discussion about MPI and its usage in real applications. All that is assumed is a working knowledge of basic MPI functionality: send, receive, collectives, derived datatypes and non-blocking communications. We will be using a Socrative online quiz - to try out a dummy quiz, go to https://b.socrative.com/login/student/ and enter "HPCQUIZ" for "Room Name". |
Virtual tutorial |
Video Quiz Questions
|
9th August 2017 |
| Version control |
Arno Proeme EPCC. Version control tools such as Git and SVN allow us to track changes made over time to files like source code and scripts, including by different people, and to merge these changes in a systematic way. They are therefore an invaluable tool for software development and related workflows. This virtual tutorial will explain the main models underlying a wide range of version control tools and will illustrate these through live demonstrations using Git and SVN. This tutorial should prove useful to researchers already using a version control tool but uncertain of the underlying model and therefore occasionally unsure of what the tool is doing or how to get it to do what they want, as well as to researchers who are trying to decide which version control tool to choose to use in the first place to fit their needs. It should also be of interest to those who are simply trying to get a better general understanding of version control systems |
Virtual tutorial |
Slides 1 Slides 2 Video
|
19th July 2017 |
| Approaches to tackling I/O issues at large scale | Adrian Jackson and Nick Brown, EPCC look at using MPI-IO and doing inline analysis. Adrian : Improving COSA at scale: Parallel I/O and load balancing This presentation will outline the work we have undertaken on the harmonic balance CFD code, COSA, to improve both performance on large core counts and improve the usability of the code. The original parallel I/O functionality in COSA, whilst able to write large amounts of data in parallel, was sub-optimal in its use of MPI-I/O. We have optimised this, reducing the I/O time by up to 70% on large core counts. Furthermore, the domain decomposition used to split work across cores in the parallelisation meant users had to work hard to ensure the simulation input was load balanced. We have added functionality to automatically load balance simulations, meaning users have to undertake much less work preparing simulations, speeding up the whole science workflow for this application, and ensuring resources are efficiently used when running simulations. Nick : In-situ data analytics for atmospheric modelling The Met Office NERC Cloud model (MONC) has been developed by EPCC and the Met Office to enable weather and climate communities to model clouds & turbulent flows at scale on thousands of cores. The computation itself generates raw prognostic 3D fields, however the scientists are far more interested in higher level, diagnostic, information such as how the average temperature at each level in the system changes over time. As such, MONC also has to support data analytics but based upon the very significant (many GBs) size of the raw fields involved we do not want to pay the cost of physically writing raw data to disk or pausing the computation whilst analytics is performed. Bit reproducibility, performance & scalability, and the ability to easily configure and extend the analytics meant that we decided to develop our own approach. In this talk I will discuss the motivation in more detail, describe our analytics architecture, some of the specific challenges we faced and solutions to these, and discuss performance and scalability results of a standard MONC test case on ARCHER running over 32768 computational and 3277 analytics cores. |
Webinar |
Slides (Adrian) Slides (Nick) Video
|
14th June 2017 |
| Performance analysis on ARCHER using CrayPAT - a case study with live demos | Gordon Gibb Profiling tools are extremely useful in understanding where a program is spending its time, which is an essential part of any performance optimisation exercise. In this webinar we will explain how profilers work, and look in detail at CrayPat which is the tool designed specifically for Cray systems such as ARCHER. How to use CrayPat in practice will be explained using live demos on a representative parallel program. |
Virtual tutorial | Slides
Source Code Video
|
10th May 2017 |
| eCSE06-04 | Implementation of generic solving capabilities in ParaFEM Mark Filipiak, EPCC and Lee Margetts, Manchester ParaFEM is a freely available, portable library of subroutines for parallel finite element analysis, which is written in Fortran 90 and uses MPI as the parallelisation method. ParaFEM can be used to solve various types of problems, including stress analysis, heat flow, eigenvalue and forced vibration. ParaFEM implements two linear solvers, conjugate gradient and BiCGStab(l), both with Jacobi preconditioning. To extend the range of solvers, the PETSc library of solvers has been interfaced to ParaFEM. In this webinar we describe the design of the interface, note some of the technical difficulties encountered, and compare the performance of the ParaFEM and PETSc implementations of the conjugate gradient and BiCGStab(l) solvers. |
virtual tutorial | Slides Video
|
26th April 2017 |
| Compilation and the mysticism of Make | Compilation and the mysticism of Make "Make" is the standard Unix tool for managing multi-file programs, where the compilation instructions are encoded in a "makefile". Most large packages will come with associated makefiles, so it is important to understand how they work so code can be successfully compiled on new platforms, e.g. when porting a code to ARCHER. Unfortunately, makefiles have a tendency to become very complicated and attain something of a mystical status. They are handed down from father to son, mother to daughter, supervisor to student, requiring magical incantations and only fully understood by a select few ... In fact, make is fundamentally a very straightforward tool and most makefiles can be understood with some basic knowledge and a bit of practical experimentation. In an initial 45-minute presentation, David Henty (EPCC and ARCHER CSE Support) will cover:
|
Virtual tutorial | Slides Source code Video
|
12th April 2017 |
| eCSE05-05 |
Open source exascale multi-scale framework for the UK solid mechanics community Anton Shterenlikht, University of Bristol, Luis Cebamanos, EPCC and Lee Margetts, University of Manchester. Fortran coarrays have been used as an extension to the standard for over 20 years, mostly on Cray systems. Their appeal to users increased substantially when they were standardised in 2010. In this work we show that coarrays offer simple and intuitive data structures for 3D cellular automata (CA) modelling of material microstructures. We show how coarrays can be used together with an MPI finite element (FE) library to create a two-way concurrent hierarchical and scalable multi-scale CAFE deformation and fracture framework. Design of a coarray cellular automata microstructure evolution library CGPACK is described. We have developed hybrid MPI/coarray multi-scale miniapps from MPI finite element library ParaFEM and Fortran 2008 coarray cellular automata library CGPACK. We show that independently CGPACK and ParaFEM programs can scale up well into tens of thousands of cores. The miniapps represent multi-scale fracture models of polycrystalline solids. The software from which these miniapps have been derived will improve predictive modelling in the automotive, aerospace, power generation, defense and manufacturing sectors. The libraries and miniapps are distributed under BSD license, so these can be used by computer scientists and hardware vendors to test various tools including compilers and performance monitoring applications. CrayPAT tools have been used for sampling and tracing analysis of the miniapps. Two routines with all-to-all communication structures have been identified a primary candidates for optimisation. New routines have been written implementing the nearest neighbour algorithm and using coarray collectives. Strong scaling limit for miniapps has been increased by a factor of 3, from about 2k to over 7k cores. Excessive synchronisation might be one contributing factor to the scaling limit. Therefore we conclude with a comparative analysis of synchronisation requirements in MPI and coarray programs. Specific challenges of synchronising a coarray library are discussed. |
Webinar | Slides Video
|
29 March 2017 |
| eCSE05-13 | Optimisation of LESsCOAL for large-scale high-fidelity simulation of coal pyrolysis and combustion Kaidi Wan, Zhejiang University, and Jun Xia, Brunel University This webinar will summarise the eCSE05-13 project on optimising LESsCOAL (Large-Eddy Simulations of COAL combustion) on ARCHER. The specific objective was to achieve 80% of the theoretical parallel efficiency when up to 3,000 compute cores are used on ARCHER. The software package can be used for a broad range of turbulent gas-solid and gas-liquid two-phase combustion problems using high-fidelity simulation techniques. The Navier-Stokes equations are used for the turbulent carrier gas-phase. For the second discrete phase, its characteristic size is much smaller than the Eulerian grid spacing and therefore a point-source Lagrangian approach can be applied. Full coupling between the two phases on total and species mass, momentum and energy has been incorporated. The major work included upgrading the pressure, radiation and particle modules, which had been identified as the 3 major hurdles degrading the scaling performance. The load balance of particles has been achieved via on-the-fly redistributing particles among all the cores. In addition, using one-sided communication MPI_put instead of collective communication MPI_allgather has greatly enhanced the parallel efficiency of redistributing particles among cores. A wavefront sweep algorithm and a priority queuing technique have been compared during optimising the radiation model. The former approach is more suitable for modelling radiation in a long channel/tube with a large aspect ratio. The priority queuing technique has been found to be more suitable for the cases under consideration in this project. For the pressure module, 14 different combinations of a solver and a pre-conditioner available in the open-source software package HYPRE have been tested. It was found that GMRES-PFMG and PCG-PFMG consume less computational time per time step for the cases under consideration. After optimisation, the parallel efficiency of the code reaches 80% of the theoretical one for weak scaling. For strong scaling, the parallel efficiency is over 80% when up to 1,200 cores are used for typical cases, but reduces to 48% on 3,072 ARCHER cores. The root cause is the low grid and particles numbers in charge by each core, which leads to a high communication/computation ratio. It is clear that future work is still needed to further optimise the parallel efficiency of the code. On the other hand, weak scaling is more relevant with laboratory-scale turbulent pulverised-coal combustion cases under consideration, i.e. in general the compute core number increases with the problem scale, which is measured by total grid and particle/droplet numbers. |
Webinar |
Video Slides
|
22 March 2017 |
| Parallel IO on ARCHER | Andy Turner, EPCC The ARCHER CSE team at EPCC have recently been benchmarking the performance of different parallel I/O schemes on ARCHER and this has led to the production of the "I/O on ARCHER" chapter in the ARCHER Best Practice Guide (http://www.archer.ac.uk/documentation/best-practice-guide/io.php). This webinar discusses the results from the benchmarking activity and what this means for optimising the I/O write performance for applications on ARCHER. |
Virtual Tutorial | Slides Video
|
8 February 2017 |
| eCSE Tutorial | Embedded CSE (eCSE) support provides funding to both the established ARCHER user
community and communities new to ARCHER to develop software in a sustainable manner
for running on ARCHER. This enables the employment of a researcher or code developer
to work specifically on the relevant software to enable new features or improve the
performance of the code. Calls are issued 3 times per year. This tutorial provides an overview of the eCSE submission process, including material focussing on applications to use the new XC40 Xeon Phi System. |
Webinar | Slides Video
|
January 18 2017 |
| Modern Fortran | Adrian Jackson discusses the features of "modern" Fortran (Fortran90 and beyond), and the steps that need to be considered to port older Fortran (primarily Fortran 77) to new Fortran standards. We also briefly discuss some of the new Fortran features (in Fortran 2003 and beyond). | Virtual Tutorial | Slides Video
|
11 January 2017 |
| eCSE03-9 | Adjoint ocean modelling with MITgcm and OpenAD Dan Jones and Gavin Pringle Adjoint methods can be used to gain unique insights into the ocean and cryosphere, for example via (1) the construction of observationally-constrained estimates of the ocean state and (2) sensitivity studies that quantify how a selected model output depends on the full set of model inputs. In this webinar, we discuss oceanic adjoint modelling on ARCHER using two open source tools, MITgcm and OpenAD. In particular, we discuss how available test cases may be modified for new applications. This webinar may be of interest to anyone wishing to apply adjoint modelling techniques to oceanic and/or cryospheric problems. |
Virtual tutorial |
Slides Video
|
Wednesday 30th November 2016 |
| CP2K | CP2K: Recent performance improvements and new TD-DFT functionality Iain Bethune and Matthew Watkins We will present a round-up of recent additions and improvements to CP2K, most of which are available in the recent CP2K 4.1 release that is now installed on ARCHER. The eCSE06-06 project has improved performance for a large, load-imbalanced calculations, and implemented OpenMP parallelism for the pair-potential and non-local type dispersion corrections. We will also present new functionality for Time Dependent Density Functional Theory (TD-DFT) implement in eCSE 03-11, including the Maximum Overlap Method, support for hybrid density functionals (direct and ADMM methods) in linear-response TD-DFT, calculation of transition dipole moments and oscillator strengths, and parallel computation of excited states. |
Virtual tutorial |
Slides - Iain's - Matt's Video
|
Wednesday 23rd November 2016 |
| eCSE05-14 | Large-Eddy Simulation Code for City Scale Environments Z Tong Xie and Vladimir Fuka The atmospheric large eddy simulation code ELMM, (Extended Large-eddy Microscale Model) was optimized and its capabilities are enhanced within project eCSE 05-14 in order to allow simulations to allow simulations of city scale problems (~10 km). The model is used mainly for simulations of turbulent flow and contaminant dispersion in the atmospheric boundary layer, especially in urban areas. A hybrid OpenMP/MPI version of the code was developed. The new features developed include one-way and two-way grid nesting. This enables simulation of large domains in affordable resolution and high resolution in selected areas while still maintaining a simple uniform grid with the associated code simplicity and efficiency. Synthetic turbulence generation is employed at nested boundaries to generate turbulent structures not resolved in the outer grid. |
Virtual tutorial | Slides Video
|
Wednesday 9th November 2016 15:00 |
| eCSE04-07 and eCSE05-10 | Multi-resolution modelling of biological systems in LAMMPS Iain Bethune and Oliver Henrich Two recent ARCHER eCSE projects (eCSE04-07 and eCSE05-10) have implemented extensions to LAMMPS to allow efficient simulations of biological systems using multiscale or dual-resolution coarse-grained/atomistic approaches. The first project concerns the implementation of ELBA, an ELectrostatics-BAsed force-field originally designed for modelling lipid bilayers, but which is also transferrable to applications like computation of solvation free energies of biomolecules in water. Simulations using ELBA may combine coarse-grained beads with atomistic particles described using standard force-fields, for example membrane proteins embedded in a bilayer. We will discuss the implementation of a new integration algorithm for stable long-timescale Molecular Dynamics, and extensions to the LAMMPS load balancer to improve parallel performance in dual-resolution simulations. The second project implemented the oxDNA force field, a coarse-grained model for nucleic acids. New pair interactions have been developed and permit now dynamical simulation of single and double stranded DNA on very large time and length scales. LAMMPS has emerged as a suitable computational platform for coarse-grained modelling of DNA and this eCSE project created a possibility to compare and optimise them with a view towards later multiscale modelling of DNA. We will explain how the oxDNA force field is parametrised and discuss a set of new Langevin-type rigid body integrators with improved stability. |
Webinar | Slides Video
|
Wednesday 19th October 2016 |
| Using KNL on ARCHER | It was recently announce that a 12-node Cray system containing Intel's new Knights Landing (KNL)
manycore processors is to be added to the existing ARCHER service. This webinar, which follows on from the previous KNL Overview and the eCSE programme for KNL-related projects, will focus on how to use the ARCHER KNL system in practice. As ever, the main aim is to promote discussion between attendees and members of the ARCHER CSE team. The webinar should be of interest to anyone considering porting applications to Knights Landing. |
Virtual tutorial |
Slides Video Chat
|
Wednesday 12th October 2016 |
| eCSE | Embedded CSE (eCSE) support provides funding to both the established ARCHER user community
and communities new to ARCHER to develop software in a sustainable manner for running on ARCHER.
This enables the employment of a researcher or code developer to work specifically on the relevant
software to enable new features or improve the performance of the code. Calls are issued 3 times per year. This tutorial will provide an overview of the eCSE submission process, with a particular focus on applications to use the new XC40 Xeon Phi System. Following this there will be an drop in session for potential applicants to ask the eCSE team any questions they may have about the whole eCSE programme. |
Webinar | Video
|
Wednesday 21 Sept 2016 |
| The Intel Knights Landing Processor | It was recently announced that a 12-node Cray system containing Intel's new Knights Landing (KNL)
manycore processors is to be added to the existing ARCHER service. This webinar will focus on the architecture of the KNL processor, including details of its new high-bandwidth memory and a discussion of how best to use all of the 250-plus possible threads of execution in real applications. As ever, the main aim is to promote discussion between attendees and members of the ARCHER CSE team. The webinar should be of interest to anyone considering porting applications to Knights Landing. This is the first of three webinars - subsequent events will focus on applying to the eCSE programme for KNL-related projects, and on how to use the ARCHER KNL system in practice. |
Virtual tutorial | Slides Video
|
Wednesday 14th Sept 2016 |
| eCSE03-8 Construction of a general purpose parallel supermeshing library for multimesh modelling | Models which use multiple non-matching unstructured meshes generally need to solve a computational geometry problem, and construct intersection meshes in a process known as supermeshing. The algorithm for solving this problem is known and has an existing implementation in the unstructured finite element model Fluidity, but this implementation is deeply embedded within the code and unavailable for widespread use. This limits access to this powerful numerical technique, and limits the scope of present multimesh modelling applications. In general a numerical model may need to consider not only two non-matching unstructured meshes, but also allow the two meshes to have different parallel partitionings. An implementation of a parallel supermeshing algorithm in this general case is lacking, and this has limited the scale of multimesh modelling applications. This project will address these issues by:
|
Webinar | Slides Video
|
13th July 2016 |
| TAU HPC Tool | Guest speaker, Sameer Shende from University of Oregon talking about the TAU HPC Tool (Tuning and Analysis Utilities) which is available on ARCHER. | Virtual Tutorial | Slides Video
|
7th July 2016 |
| libcfd2lcs: A general purpose library for computing Lagrangian coherent structures during CFD simulations | Justin Finn, School of Engineering, The University of Liverpool A number of diagnostics have been proposed in recent years that allow transport and mixing patterns to be objectively defined in time dependent or chaotic flows. One such family of diagnostics, coined Lagrangian Coherent Structures (LCS), correspond to the locally most attracting or repelling material surfaces in a flow. They have been instrumental in developing new understanding of transport in a number of geophysical, biological, and industrial applications. However, the resource intensive post-processing procedure typically used to extract these features from experimental or numerical simulation data has probably limited the impact of LCS to date. In this web-forum I'll describe libcfd2lcs, a new computational library that provides the ability to extract LCS on-the-fly during a computational fluid dynamics (CFD) simulation, with modest additional overhead. Users of the library create LCS diagnostics through a simple but flexible API, and the library updates the diagnostics as the user's CFD simulation evolves. By harnessing the large scale parallelism of platforms like ARCHER, this enables CFD practitioners to make LCS a standard analysis tool, even for large scale three dimensional data sets. This seminar will focus on the current capabilities of libcfd2lcs and how to develop applications that use it. Development of libcfd2lcs was funded by an eCSE grant from the Edinburgh Parallel Computing Center (EPCC). It is currently accessible to all ARCHER users or can be obtained by visiting here. | Webinar | Slides Video
|
22nd June 2016 |
| Modern Fortran | Adrian Jackson will discuss the features of "modern" Fortran (Fortran90 and beyond), and the steps that need to be considered to port older Fortran (primarily Fortran 77) to new Fortran standards. We will also briefly discuss some of the new Fortran features (in Fortran 2003 and beyond). | Virtual Tutorial | Please see the re-run of this session January 2017 | 8th June 2016 |
| Introduction to the RDF |
An introduction to the Research Data Facility, what it is and how to use it. | Virtual Tutorial | Slides Video
|
4th May 2016 |
| Using MPI - Q&A Webinar |
Learning the basic MPI syntax and writing small example programs can be relatively straightforward,
but many questions only arise when you first tackle a large-scale parallel application. Typical topics
include how best to avoid deadlock, overlapping communication and calculation, debugging strategies or parallel IO. This virtual tutorial is an opportunity for users to ask any questions they want about using MPI in real applications, which we will try to answer based on the practical experience gained by the ARCHER CSE team. |
Webinar | Questions Video
|
6th April 2016 |
| OpenMP4 |
The latest versions of the compilers on ARCHER now all support OpenMP 4.0. This webinar will give a tour of the new features in this version of the OpenMP standard and some examples of how they can be used. |
Virtual Tutorial | Slides Video
|
9th March 2016 |
| ARCHER Programming Environment Update |
Harvey Richardson and Michael Neff give an overview of recent changes and updates in the Cray programming environment available on ARCHER. | Webinar | Video
|
Wednesday 10th February 2016 |
| PBS Job Submission | How PBS job submission works on ARCHER, which is not as simple as it may first appear. | Virtual Tutorial | Slides Video Chat
|
Wednesday 13th January 2016 |
| eCSE virtual drop-in session | An opportunity for ARCHER users to ask the eCSE team any questions about the eCSE programme | Virtual Tutorial |
Video Chat
|
Wednesday 16th December 2015 at 11:00 |
| Lustre and I/O Tuning | /work on ARCHER is a Lustre file system. This virtual tutorial is a brief introduction to Lustre with tips on how to choose Lustre settings that will improve I/O performance. | Virtual Tutorial | Slides Video
|
Wednesday 25th November 2015 |
| Parallel Software usage on UK National HPC Facilities 2009-2015: How well have applications kept up with increasingly parallel hardware? | One of the largest challenges facing the HPC user community on moving from terascale, through petascale, towards
exascale HPC is the ability of parallel software to meet the scaling demands placed on it by modern HPC architectures. In this
webinar I will look at the usage of parallel software across two UK national HPC facilities: HECToR and ARCHER to understand
how well applications have kept pace with hardware advances. These systems have spanned the rise of multicore architectures: from 2 to 24 cores per compute node. We analyse and comment on: trends in usage over time; trends in the calculation size; and changes in research areas on the systems. The in-house Python tool that is used to collect and analyse the application usage statistics is also described. We conclude by using this analysis to look forward to how particular parallel applications may fare on future HPC systems. Andy Turner, EPCC. |
Webinar | Slides Video Chat
|
11th November 2015 |
| Supporting Diversity | Outlining some of the issues which students may experience when taking part in training and courses, and some stategies which we can use to make our courses more accessible. | Virtual Tutorial | Slides Video
|
4th November 2015 |
| Using OpenFOAM on ARCHER | Several versions of OpenFOAM are installed on Archer. This webinar explains how to use the installed versions and how to compile OpenFOAM yourself. It also provides tips on how to use OpenFOAM efficiently on the /work Lustre filesystem. | Virtual Tutorial | Slides Video Chat
|
23rd September 2015 |
| Data Management | This talk looks at some of the potential bottlenecks of transferring data between systems, specifically in relation to the UK Research Data Facility (RDF) and ARCHER service. Archiving is proposed as a method of reducing the volume of files and addressing the bottleneck of filesystem metadata operations. Different copying utilities are examined and the trade off between security and performance considered. | Virtual Tutorial | Video Slides Chat
|
Wednesday 9th September 2015 |
| eCSE Webinar | Embedded CSE (eCSE) support provides funding to both the established
ARCHER user community and communities new to ARCHER to develop software in
a sustainable manner for running on ARCHER. This enables the employment of
a researcher or code developer to work specifically on the relevant software
to enable new features or improve the performance of the code. Calls are issued 3 times per year. This tutorial aims to give an overview of what the eCSE is and to provide examples of the kind of work which is supported. It also gives a short description of the submission and review process. |
virtual tutorial | Slides Video Chat transcript
|
26th August 2015 |
| Bash is Awesome! | Andy Turner bash has may useful features, particularly for scripting. In this short webinar I will give an intorduction to a number of the features I find the most useful on a day-to-day basis and hopefully hear from attendees which other features I should be using! |
Virtual Tutorial | Slides Example script Video Chat transcript
|
12 August 2015 |
| Compilation and the mysticism of Make | "Make" is the standard Unix tool for managing multi-file programs, where the compilation instructions are encoded in a "makefile". Most large packages will come with associated makefiles, so it is important to understand how they work so code can be successfully compiled on new platforms, e.g. when porting a code to ARCHER. Unfortunately, makefiles have a tendency to become very complicated and attain something of a mystical status. They are handed down from father to son, mother to daughter, supervisor to student, requiring magical incantations and only fully understood by a select few ... In fact, make is fundamentally a very straightforward tool and most makefiles can be understood with some basic knowledge and a bit of practical experimentation. In an initial 30-minute presentation, David Henty (EPCC and ARCHER CSE Support) will cover:
The aim of the presentation is to initiate online discussions between users and staff on compilation and makefiles, or any other aspects of the ARCHER service. Here are PDF copies of the slides and the sample programs as a gzipped tar file. |
Virtual Tutorial | Video
|
Wednesday July 8th 2015 |
| Flexible, Scalable Mesh and Data Management using PETSc DMPlex | [Full Description] | webinar | Slides Video to come |
Wednesday June 24th 2015 |
| Introduction to Version Control | [Full Description] | virtual tutorial | Slides: Part 1 Part 2 Videos: Part 1, Demo Part 2 Practical
|
June 3rd & 17th 2015 |
| Voxel-based biomechanical engineering with VOX-FE2 | Richard Holbrey (University of Hull), Neelofer Banglawala (EPCC) [Full Description] |
webinar | Slides Video Demo videos: 1, 2, 3.
|
April 29th 2015 |
| Single Node Performance Optimisation - A Computational Scientist's Guide to Computer Architecture |
This two-part tutorial provides an introduction to the architecture of processors and memory systems, and is aimed at providing a grounding in these topics for anyone who is involved in optimising the single-node performance of codes. | virtual tutorial | Videos Part 1 Part 2
|
April 8th & 15th 2015 |
| Not-so-old Fortran | (Harvey Richardson, Cray Inc.) [Full Description] | webinar | Video |
Wednesday March 18th 2015 |
| Performance analysis on ARCHER using CrayPAT - a case study | Iain Bethune, EPCC | virtual tutorial | Slides Video |
Wednesday March 11th 2015 |
| Experiences porting the Community Earth System Model (CSEM) to ARCHER | (Dr Gavin J. Pringle, EPCC) [Full Description] | webinar | Slides Video |
Wednesday February 25th 2015 |
| PBS Job Submission | An online session open to all where you can put questions to ARCHER staff. All you need is a computer and internet access. This session starts with a talk on how PBS job submission works on ARCHER, which is not as simple as it may first appear, followed by an open Q&A. | virtual tutorial | Slides | Wednesday February 11th 2015 |
| A Pinch of salt in ONETEP's solvent model | (Lucian Anton, Scientific Computing Department, STFC, Daresbury Laboratory; Jacek Dziedzic, Chris-Kriton Skylaris, School of Chemistry, University of Southampton) [Full Description] | webinar | Slides Video |
Wednesday January 28th 2015 |
| ARCHER eCSE projects | Embedded CSE (eCSE) support provides funding to both the established ARCHER user
community and communities new to ARCHER to develop software in a sustainable manner
for running on ARCHER. This enables the employment of a researcher or code developer
to work specifically on the relevant software to enable new features or improve the
performance of the code. Calls are issued 3 times per year. This tutorial aims to give an overview of what the eCSE is and to provide examples of the kind of work which is supported. It also gives a short description of the submission and review process. |
virtual tutorial | Wednesday January 7th 2015 | |
| An Introduction to GPU Programming with CUDA | Graphics Processing Units (GPUs) offer significant performance advantages over traditional CPUs for many types of scientific computations. However, adaptations are needed to applications in order to effectively use GPUs. During this tutorial, I will present an introduction to GPU programming focusing on the CUDA language which can be used in conjunction with C, C++ or Fortran to program NVIDIA GPUs. After this presentation, there will be opportunities for questions and discussion. | virtual tutorial |
Video
|
Wednesday December 10th 2014 |
| Introduction to Version Control | [Full Description] | virtual tutorial |
Slides Terminal Logs: SVN A SVN B Git Video
|
Wednesday November 12th 2014 |
| Hybridisation: Adding OpenMP to MPI for the plasma simulation code GS2 | (Adrian Jackson, EPCC) [Full Description] | webinar | Slides Video |
Wednesday October 29th 2014 |
| Parallel IO and the ARCHER Filesystem | File input and output often become a severe bottleneck when parallel applications are scaled up to large numbers of processors. In this tutorial, I will outline the methods commonly used for parallel IO and discuss how well each of these is likely to perform on ARCHER. The work space on ARCHER uses the parallel Lustre file system, and I will explain how the design and configuration of Lustre affects IO performance. The tutorial will end with a brief overview of ways to measure the IO behaviour of your own applications, and what steps to take if the performance is not as good as expected. The main aim of the 45-minute presentation is to initiate online discussions between users and staff on parallel IO, or any other aspects of the ARCHER service. |
virtual tutorial | Slides Video |
Wednesday October 8th 2014 |
| Abstraction of Lattice Based Parallelism with Portable Performance | (Alan Gray, EPCC) [Full Description] | webinar | Video Paper |
Wednesday September 24th 2014 |
| Tools for Building and Submitting an eCSE Proposal | This virtual tutorial follows on from last week's session, which provided an overview of the embedded CSE (eCSE) programme and summarised the kind of work supported. Firstly, this tutorial will provide a practical overview of the on-line submission system. Secondly, the tutorial will look at the type of supporting performance data required for a successful eCSE submission and discuss the profiling and performance monitoring tools available to obtain this data. eCSE calls are issued 3 times per year. | virtual tutorial | Slides Video |
Wednesday September 3rd 2014 |
| ARCHER eCSE Tutorial | Embedded CSE (eCSE) support provides funding to the ARCHER user community
to develop software in a sustainable manner for running on ARCHER.
This enables the employment of a researcher or code developer to work specifically
on the relevant software to enable new features or improve the performance of the code. Calls are issued 3 times per year. This tutorial aims to give an overview of what the eCSE is and to provide examples of the kind of work which is supported. We will also give a short description of the submission and review process. [Full Description] |
virtual tutorial | Slides Video |
Wednesday August 27th 2014 |
| Python for High Performance Computing | The Python language is becoming very popular in all areas of Scientific Computing due to its flexibility and the availability of many useful libraries. It is increasingly being used for HPC, with typical use cases being:
In this short tutorial we give a brief overview of the aspects of Python most relevant to HPC, and cover the pros and cons of these typical use cases. The goal is to give guidance on how to utilise the many attractive features of Python without sacrificing too much performance. The main aim of this presentation was to initiate online discussions between users and staff on using Python for HPC, or any other aspects of the ARCHER service. |
virtual tutorial | Slides Video |
Wednesday August 13th 2014 |
| Experiences of porting real codes to Intel's Xeon Phi | (Iain Bethune, EPCC) [Full Description] | webinar | (pdf) Video |
Wednesday July 30th 2014 |
| Make and Compilation Issues | "Make" is the standard Unix tool for managing multi-file programs, where the compilation instructions are encoded in a "makefile". Most large packages will come with associated makefiles, so it is important to understand how they work so code can be successfully compiled on new platforms, e.g. when porting a code to ARCHER. Unfortunately, makefiles have a tendency to become very complicated and attain something of a mystical status. They are handed down from father to son, mother to daughter, supervisor to student, requiring magical incantations and only fully understood by a select few ... In fact, make is fundamentally a very straightforward tool and most makefiles can be understood with some basic knowledge and a bit of practical experimentation. In an initial 30-minute presentation, David Henty (EPCC and ARCHER CSE Support) will cover:
|
virtual tutorial | slides and examples Video |
Wednesday July 9th 2014 |
| Optimising MPI Rank Placement at Runtime | (Tom Edwards, Cray Inc.) [Full Description] | virtual tutorial | Video
|
Wednesday June 25th 2014 |
| A Very Brief Introduction to Parallel Programming Models | In this short presentation Andy Turner (EPCC and ARCHER CSE Support) provides a brief outline of the two different parallel programming models in common use on ARCHER: shared memory (through OpenMP) and message passing (through MPI), and an analysis of their strengths and weaknesses. This requires an understanding of how an operating system works with processes and threads, so we provide a high-level overview of these concepts. | virtual tutorial | Video
|
Wednesday June 11th 2014 |
| The OpenSHMEM Single-sided PGAS Communications Library | (David Henty, EPCC) [Full Description] | webinar | (pdf) | Wednesday May 28th 2014 |
| ARCHER Filesystems |
The ARCHER system has a number of different filesystems and data storage hardware
associated with it. Understanding the purposes and configurations of the different
filesystems, the disk quotas, and how disk quotas are managed in SAFE is useful to
enable proper use of ARCHER and to ensure your jobs don't fail or you don't lose
valuable data.
A short presentation at the start of this tutorial, given by Adrian Jackson (EPCC and ARCHER CSE Support) addresses the following:
|
virtual tutorial | Slides Recording Video |
Wednesday May 14th 2014 |
| Parallel IO in Code_Saturne | (Charles Moulinec, STFC Daresbury Laboratory) [Full Description] | webinar | (pdf) Video |
Wednesday April 30th 2014 |
| PBS Job Submission | Slides Recording Video |
virtual tutorial | Wednesday April 9th 2014 | |
| Vectorisation and optimising for processor architecture | (Peter Boyle, University of Edinburgh) | webinar | (pdf) | Wednesday March 26th 2014 |
| SAFE | virtual tutorial | Wednesday March 12th 2014 | ||
| Python for HPC: Tools and performance | (Nick Johnson, EPCC) | webinar | (pdf) Meeting Notes |
Wednesday February 26th 2014 |
| OpenMP Tips, Tricks and Gotchas: Experiences from programming and teaching OpenMP | (Mark Bull, EPCC) | webinar | (pdf) Video |
Wednesday January 29th 2014 |
| Introduction to ARCHER and CRAY MPI | virtual tutorial | Wednesday January 22nd 2014 | ||
| ARCHER Tips and Tricks: A few notes from the CSE Team | (AndyTurner, EPCC) | webinar | (pptx|pdf) Video |
November 26th 2013 |
