Centre for High Performance Computing 2025 National Conference

30 Nov 2025, 08:00 → 3 Dec 2025, 22:20 Africa/Johannesburg

The 19th CHPC National Conference.

The aim of the conference: to bring together our users so that their work can be communicated, to include world renowned experts, and to offer a rich programme for students, in the fields of high performance computing, big data, and high speed networking.  The CHPC National Conference is co-organised by the CHPC, DIRISA and SANReN.

Cape Town

The CHPC 2024 Conference will be an in-person event with a physical programme hosted at the Century City Conference Centre, Cape Town.

For more information please see the main conference site.

Programme

• 30 November: Workshops
• 1–3 December: Conference

Evening Social Events

• Monday 1 December: Conference Dinner 
• Tuesday 2 December: Poster Showcase Cocktails [Exhibition Hall]
• Wednesday 3 December: Awards Dinner 

Registration

Online registration will close on Friday 27 November 2025. Thereafter only onsite registration (at full fees) will be available at the venue.

Co-Organisers

 

 

Sponsors

Diamond

 

Platinum

 

 

Gold

 

     

application-1-tutorial-workshop.odt csir_client_registration_information.docx Privacy-Notice-for-events.pdf

Sunday, 30 November

08:00 → 09:00 Registration

09:00 → 10:30 SADC Cyber-Infrastructure Meeting

SADC Cyber-Infrastructure Meeting

09:00 → 10:30 Workshop: CSC students

09:00 → 10:30 Workshop: Climate (part 1)

09:00 MODELLING CLIMATE CHANCE IMPACTS ON FUTURE OFFSHORE WIND ENERGY USING GEOSPATIAL INTELLINECE

Offshore wind energy is the most commercially and technologically developed marine renewable energy. Conversely, the potential impacts of climate change and variability on future wind energy remains poorly understood, including shifts and variations in the general wind pattern. Therefore, this project seeks to model the impacts of climate change scenarios on the potential offshore wind energy using geospatial intelligence (Satellite data and deep learning). The set objectives of the project include: (i) to understand historical climate change and wind trend for the past thirty (30) years using meteorological, satellite datasets and deep learning approaches, (ii) build predictive model for future offshore wind energy potential under different climate change scenarios. This project will use comprehensive geospatial data to develop and predict the spatial and temporal variability of climate impacts (i.e., wind speed, wave height, sea level rise, and ocean currents) on future offshore wind energy production under different climate change scenarios. Wind speed and wind density retrieved using SAR data were used to predict future offshore wind energy. The results will contribute to a better understanding of how climate change may affect offshore wind production and ensure long term sustainability. By integrating geospatial intelligence and advanced modeling techniques, this project will provide critical insights into the resilience and adaptability of offshore wind energy systems to the challenges posed by climate change.
Key words: Climate change, Offshore wind energy, Geospatial intelligence

Speaker: Dr Kgabo Humphrey T hamaga

09:00 → 10:30 Workshop: DL_POLY (part 1)

09:00 Different Techniques of Force-field Derivation and Setting Up Molecular Dynamics (MD) Calculations at CHPC Using DL_POLY Code

Molecular dynamics (MD) is a computer simulation method for studying the physical movements of atoms and molecules. The MD method can assist one in obtaining the static quantities and dynamic quantities. This method gives a route to dynamical properties of the system: transport coefficients, time-dependent responses to perturbations, rheological properties and spectra. The atoms and molecules are allowed to interact for a fixed period of time, giving a view of the dynamic evolution of the system. The DL_POLY Code parallel molecular dynamics simulation package will be utilised for exploration of such properties of molecular systems.

Speaker: Cliffton Masedi (University of Limpopo)

Cliffton_Masedi-Proposal_for_a_Workshop_at_2024_CHPC_National_Meeting.pdf

09:00 → 10:30 Workshop: TBA

10:30 → 11:00 Break

11:00 → 12:30 SADC Cyber-Infrastructure Meeting

SADC Cyber-Infrastructure Meeting

11:00 → 12:30 Workshop: Climate (part 2)

11:00 → 12:30 Workshop: DL_POLY (part 2)

11:00 → 12:30 Workshop: SDC students

11:00 → 12:30 Workshop: TBA

12:30 → 13:30 Lunch

13:30 → 15:00 SADC Cyber-Infrastructure Meeting

SADC Cyber-Infrastructure Meeting

13:30 → 15:00 Workshop: Materials Modelling (part 1)

13:30 Exploiting exascale HPC for materials modelling; UK's Excalibur Project

GULP is one of the leading materials software for modelling materials using the method of interatomic potentials, likewise CASTEP and CP2K are two of the leading electronic structure codes for modelling materials. In this workshop the theory and a practical guide to how these can be employed to exploit HPC will be taught by developers of these codes.

The ChemShell computational chemistry environment will be introduced with a focus on multiscale quantum mechanical/molecular mechanical (QM/MM) modelling of materials systems. The course will cover how to set up QM/MM models for a range of materials chemistry problems and running calculations through interfaces to QM and MM codes including CASTEP and GULP. Recent developments in ChemShell for materials modelling carried out under the PAX-HPC project will be explored.

Speakers: Prof. Scott Woodley (UCL), Prof. Thomas Keal (STFC), Dr Marcello Puligheddu (STFC), Dr Matt Smith (University of York), Dr Alexey Sokol (UCL), Dr Rajany KV (STFC)

Scott_Woodley-tutorial-workshop.pdf

13:30 → 15:00 Workshop: Molecular Dynamics (part1)

13:30 Windows Subsystem for Linux with GPU support for molecular dynamics simulations

Windows Subsystem for Linux (WSL) is a feature of the Windows operating system that enables you to run a Linux file system, along with Linux command-line tools and graphical user interface (GUI) applications, directly on Windows. Unlike conventional virtual machines such as those run with Oracle Virtualbox or VMWare, WSL requires fewer resources (CPU, memory and storage) and can access all the hardware components of your machine (including the graphical processing unit (GPU)) that Windows has access to.

There are several Centre for High Performance Computing (CHPC) users that still make use of the compute resources for both testing and production, especially when it comes to molecular dynamics codes such as AMBER. This is not an ideal situation for testing purposes as it would mean that individuals need to queue, sometimes for long periods of time, before discovering something might be wrong with their setup. Recently, AMBER has become opensource for non-commercial use, which means that users no longer need to do their testing on the CHPC as they can test on their local laptop/desktop prior to submitting simulations to the queuing system at the CHPC.

In this workshop we will:
• Setup WSL on a Windows machine.
• Install a version of Ubuntu using WSL.
• Install essential libraries needed to compile AMBER in Ubuntu.
• Install CUDA Toolkit in Ubuntu for GPU support.
• Compile the serial, parallel and GPU versions of AMBER.
• Export the above WSL instance so that it can be deployed onto other laptops/desktops.

This workshop will be ideal for researchers, scientists and students that make use of the CHPC resources for their computational chemistry research.

Prerequisite:
Laptop with Windows 10 or 11 (If you have Linux, you will still be able to learn how to compile the software package/s).
Preferably 8GB RAM or more (not a must, but things can be slow with less RAM).
Nvidia GPU (not a must as you can still get the CPU version of the code compiled and running).

Note: We can consider other applications should there be time as it is possible to install opensource electronic structure codes within WSL.

Speaker: Krishna Govender (University of Johannesburg)

13:30 → 15:00 Workshop: Practical QC (part 1)

13:30 Hands-On Practical Introduction to Quantum Computing

This practical introduction to quantum computing aims to offer a foundational understanding of key quantum computing concepts, algorithms, and practical applications. The lectures will cover the basics of quantum computing including qubits, entanglement, and quantum gates, as well as an introduction to quantum circuits. As an example, we will explore in the tutorial the quantum dynamics of a spin system on quantum computer.

Target Audience: Students, academics and industry representatives interested in a practical introduction to quantum computing.

Prerequisites: Basic Python knowledge

Speaker: Francesco Petruccione (UKZN)

13:30 → 15:00 Workshop: Wind energy (part 1)

13:30 A near real-time spatial monitoring and forecasting of wind energy using geospatial big data

This workshop will focus on the integration of geospatial datasets, and deep
learning algorithms for real-time monitoring and forecasting of offshore wind
energy. The session will cover the framework for data retrieval, pre-processing,
integration of remotely sensed datasets and the development of predictive models
to optimize wind turbine performance. After developing the predictive models, we
will integrate climate scenarios to forecast the state of wind in real-time
monitoring and near-future prediction. Participants will learn how to integrate
cutting-edge geospatial, and meteorological datasets with deep learning
algorithms to predict energy production. The workshop will provide valuable
insights for renewable energy-related professionals and stakeholders on how
geospatia

The global demand for renewable energy, particularly wind energy, is escalating due to the urgent need to combat climate change and decrease reliance on fossil fuels. The study aims to develop a system using geospatial data and deep learning techniques for monitoring and forecasting wind energy. The study seeks to answer three key questions: (i) to develop a framework for integrating and processing geospatial big data for wind energy monitoring. (ii) implement a near-real-time data acquisition pipeline for continuous monitoring, and (iii) develop a predictive model using deep learning algorithms and statistical methods. The use of geospatial and meteorological datasets, turbine performance data (wind speed and direction, theoretical power and active power) and Recurring Neural Network -Long Short-Term Memory will enable near-real-time monitoring and prediction of wind energy. The model performance will be evaluated using statistical indicators like stability tests and forecast accuracy metrics like MAE and RMSE, to measure its stability under different conditions. The proposed model will be used to provide accurate wind patterns and energy potential insights, thereby optimizing wind turbine performance and energy production through the integration of various datasets. The study’s results will enhance wind energy predictions, aid in better grid planning, decrease fossil fuel reliance, and enhance grid stability.

Keywords: Deep Learning, geospatial Big data, remote Sensing, wind energy

Speakers: Prof. Phila Sibandze,, Prof. Kgabo Humphrey Thamaga

A_near_real-time_spatial_monitoring_and_forecasting_of_wind_energy_using_geospatial_big_data_-workshop.pdf

15:00 → 15:30 Break

15:30 → 17:00 SADC Cyber-Infrastructure Meeting

SADC Cyber-Infrastructure Meeting

15:30 → 17:00 Workshop: Materials Modelling (part 2)

15:30 → 17:00 Workshop: Molecular Dynamics (part 2)

15:30 → 17:00 Workshop: Practical QC (part 2)

15:30 → 17:00 Workshop: Wind energy (part 2)

17:00 → 18:00 Intermission

18:00 → 19:30 Workshop: Materials Modelling (part 3 & 4)

18:00 → 19:30 Workshop: Molecular Dynamics (part 3 & 4)

18:00 → 19:30 Workshop: TBA

18:00 → 19:30 Workshop: Wind energy (part 3)

18:00 A deep learning-based wind energy turbines suitability location analysis and transferability

Despite significant advancements in the energy industry over the past decade, most global regions still face challenges in ensuring the security and supply of fossil fuels. The challenge in wind power usage is identifying the optimal location for turbine installation to maximize energy generation while minimizing environmental and socioeconomic impacts. This study aims to explore a data-driven deep learning-based modeling framework that predicts land suitability for large-scale wind energy development by inventorying current wind farms and using spatial decision criteria. The proposed framework will use recurrent neural and convolutional neural networks to simulate intricate interactions between meteorological, environmental, and infrastructure-related spatial variables influencing wind energy potential. The model will use various spatial datasets, including wind speed data, topography, and environmental constraints, to assess its transferability to different wind regimes, environmental conditions, and infrastructure challenges across various geographic regions, Furthermore, the offshore and inland regions will be utilized to identify wind potential locations using LiDAR and SAR data from Sentinel-1 satellites for suitable evaluation and detection. The results of this study will be used to translate renewable energy sources and reduce climate change by improving wind energy potential evaluation accuracy.

Keywords: Deep learning, Inland, Transferability, Turbines suitability, wind energy

Speaker: Mr Phila Sibandze

Monday, 1 December

08:00 → 09:00 Registration

09:00 → 10:30 Keynote: Opening

Convener: Chair: Mr Mervyn Christoffels (CHPC)

09:00 Welcome

Speaker: Chair: Mr Mervyn Christoffels (CHPC)

09:05 DSI Welcome

Speaker: Mr Charles Mokonoto (Department of Science and Technology)

09:15 CSIR Welcome

Speaker: Dr Lulama Wakaba

09:25 NICIS Welcome

Speaker: Dr Happy Sithole (CHPC)

09:45 Cyberinfrastructure Collaboration: Towards Accelerated Impact - The Role of Being a Good Ancestor

In the rapidly evolving landscape of High-Performance Computing (HPC), the theme of "Cyberinfrastructure Collaboration: Towards Accelerated Impact" underscores the critical need for synergistic efforts to drive innovation and societal progress. This keynote presentation will explore the profound concept of being a "good ancestor" and its pivotal role in shaping impactful cyberinfrastructure collaborations.

As we delve into the intricacies of HPC, we must recognize that our technological advancements are not just for the present but are legacies for future generations. Being a good ancestor involves making conscientious decisions that prioritize sustainability, ethical considerations, and long-term benefits over short-term gains. This perspective encourages us to build resilient and adaptable cyberinfrastructures that can withstand the test of time and evolving challenges.

The presentation will highlight key strategies for fostering effective collaborations that embody the principles of good ancestry. These include:

1. Sustainable Design and Development: Emphasizing eco-friendly practices and energy-efficient technologies to minimize environmental impact.
2. Ethical Data Management: Ensuring data privacy, security, and equitable access to information.
3. Inclusive Innovation: Promoting diversity and inclusivity in research and development to harness a wide range of perspectives and talents.
4. Long-term Vision: Planning and implementing projects with a foresight that considers future societal needs and potential technological advancements.

By integrating these principles, we can create a cyberinfrastructure that accelerates current impact and lays a robust foundation for future generations. This approach enhances the immediate benefits of our collaborations and ensures that we leave a positive and enduring legacy.

Speaker: Mr Warren Hero*

11:00 → 11:15 Refreshment Break

11:15 → 12:45 HPC Applications: 1

Convener: Chair: Prof Lyudmila Moskaleva (University of the Free State)

11:15 Simulating Potential Impacts of Solar Radiation Modification on African Climate

Increased greenhouse gas emissions continue to warm the planet, fuelling climate extremes such as heatwaves, floods, and droughts. These events devastate socio-economic activities globally, but their impacts are even more severe in Africa, where people are more vulnerable. International negotiations on reducing greenhouse gas emissions are slow, while emissions themselves continue to rise. Solar Radiation Modification (SRM), which involves reflecting a small portion of incoming sunlight back into space, has been proposed as the cheapest and fastest way to cool the planet. Interest in SRM research has grown in recent years, but its deployment remains contentious due to the associated risks. This study explores the potential impacts of SRM on the African climate through advanced climate simulations. By simulating various SRM scenarios, we aim to understand how SRM intervention could alter temperature, precipitation patterns, freshwater availability, and heat stress across the continent. Our findings indicate that while SRM could potentially reduce average temperatures and heat stress, it may also lead to unintended consequences such as changes in rainfall distribution and increased frequency of droughts in certain regions. These results underscore the complexity of SRM as a climate intervention and highlight the need for comprehensive risk assessments before considering its implementation. This research contributes to the broader discourse on climate geoengineering by providing region-specific insights into the potential benefits and risks of SRM for Africa.

Speaker: Prof. Babatunde Abiodun* (UCT)

Babatunde_Abiodun_Talk_Title_and_Abstract.pdf

11:35 Drug Design, Resistance and Metabolisms through CHPC

One major health challenge in the Continent is the underrepresentation of African populations in the various aspects of drug development. Firstly, drug development is a long and expensive process. Hence, the focus of drug companies is often towards more profitable markets, leaving the specific needs of Africa populations out of their scope. This results in limited development of treatments for diseases that disproportionately affect African populations. Even when there are drugs against pathogenic diseases, pathogens ultimately gain resistance against any drugs, leaving drugs, which take years of work and millions of dollars to develop, ineffective. Thus, scientific priority should be not only on the design of new efficacious drugs, but also on drugs capable of bypassing pathogen’s resistance. Secondly, the African Continent hosts the most genetic diversity in the world. This diversity is not well studied, and the incorporation of the unique genetic make-up of African populations into clinical trials is limited. As a result, medications are not optimized for African populations, which may lead to variability in drug metabolism, i.e. lack of efficacy due to insufficient drug exposure or adverse reactions due to increased exposure. Furthermore, there is also a growing hypothesis on the connection between drug resistance and drug metabolism (human pharmacogenomics). This becomes particularly important in the African context, as the Continent has the highest incident rates in many diseases, and equally very high unique population specific variation profiles in the human genomes.

The integration of different aspects of bioinformatics, computational chemistry, population genomics, and intelligent systems for drug development and pharmacogenomics holds enormous potential to address these gaps. Our research aims to revolutionize drug development for diseases related to Africa, preempt drug resistance issues while developing drugs, and advance pharmacogenomics by bringing together cutting-edge computational technologies that would not be feasible without CHPC. This talk will present some case studies from our recent research and demonstrate the need for HPC resources.

Speaker: Prof. Ozlem Tastan Bishop* (Rhodes University)

11:55 Molecular Modelling in Research: From Basic Research Methods to Advanced Research

Our research group utilizes High-Performance Computing (HPC) infrastructure to implement software programs facilitating calculations including, but not limited to, molecular docking, Quantum Mechanics/Molecular Mechanics (QM/MM), Molecular Dynamics Simulations (MDS), Molecular Mechanics Generalized Born Solvent Accessible Surface Area (MM-GB/SA), Quantitative Structure-Activity Relationships (QSAR), Pharmacophore modeling, Reaction-based Enumeration, Ligand Designer, and Free Energy Perturbation plus (FEP+) techniques. These molecular modeling methodologies are employed to engineer New Molecular Entities (NMEs), predict conformations, binding modes, pharmacophore features essential for binding, conduct library screening, and develop both training and testing models with an assessment of their relative and/or absolute binding affinities. This presentation will examine our application of these diverse molecular modeling techniques within research encompassing prostate and breast cancer, malaria, diabetes, SARS-CoV-2, tuberculosis, and Alzheimer's disease. Further investigation will address our utilization of Transition State Modeling to facilitate the study of functionalization of sp2-sp3 bonds of aryl halide coupling with α-ketones, exemplifying the significant utility of unliganded transition metal catalysts. This approach has enabled the synthesis of compounds conceived with molecular modeling techniques, achieving high yields. Consequently, this presentation will also demonstrate how contemporary medicinal chemistry methodologies assist synthetic organic chemists in designing synthesizable compounds. Additionally, the presentation will discuss the synthesis of various compounds, subsequently evaluated through cell-based and biophysical assays, providing insights into the synergistic effects of molecular modeling, synthesis, and bioassays within a university-affiliated drug discovery framework. Lastly, patents granted and publications by postgraduate students mentored within this research group will be presented.

Keywords: Molecular Modeling, High Performance Computing, NMEs, university-affiliated drug discovery

Speaker: Dr Njabulo Gumede* (Walter Sisulu Unibversity)

12:15 Elevating Material Simulations with Open Source – From Consumer to Contributor

Computational Material Science is a well established field in South Africa. A great many presentations during the annual conference of the South African Institute of Physics are dedicated to materials simulations with Density Functional Theory. There are however few contributions which go beyond the application of existing codes and algorithms.
This can be attributed in part to the perceived high obstacles from closed source programmes and programmes which are badly documented and difficult to understand. In this contribution I would like to introduce the open source DFT code GPAW, which is written largely in Python. It is not only comparable in features to expensive closed source software like VASP, it also has some unique features. More importantly for this contribution, GPAW is very easy to read and modify.
I will showcase this using the electron-phonon coupling and Raman spectroscopy codes I contributed to the GPAW package, which is documented in our lasted review paper “GPAW: An open Python package for electronic structure calculations” published in The Journal of Chemical Physics ( https://pubs.aip.org/aip/jcp/article/160/9/092503/3269902 ) and has been used to predict the Raman spectrum if a large number of 2D materials.
It is my hope to motivate more South African researchers to participate in software development to raise our global profile and give back to the community.

Speaker: Dr Robert Warmbier (University of the Witwatersrand)

chpc_warmbier.pdf

12:35 Q&A

Q&A

Speaker: Q&A

11:15 → 12:45 HPC Technology: 1

Convener: Chair: Mr Faheem Essack

11:15 Digital Twins for Energy Efficient and Sustainable Datacenters (HPE)

Digital Twins are the basis of a rapidly growing field of study that pairs physical systems with their digital representation to enhance understanding of the physical counterpart. Digital twins of data-centres are becoming a reality, using bi-directional feedback loops to link operational telemetry with models of associated sub-systems, which are combined with visualisation and analytics components. We will discuss on-going collaborative research into such datacenter digital twins, which promise improved system behaviour prediction, optimised system operation, and ultimately better-informed decision-making for operating large-scale compute centre installations with associated power, cooling, and management infrastructure in a sustainable and energy efficient manner.

Tim Dykes | Senior Research Engineer | HPE HPC/AI EMEA Research Lab

Speaker: Mr Tim Dykes

11:35 AI next trends: the critical HW and SW marriage (Intel)

In this session we'll walk through the rapid evolution of the AI workloads, and learn key insights on how next gen Hardware and Software will look like.

Speaker: Mr Walter Riviera (Intel)

11:55 Dell AI Factory (Dell)

TBC

Speaker: Mr Jacques Visagie

12:15 Pinnacle Presentation (TBC)

TBC

Speaker: Mr Roger Crighton (Supermicro)

12:35 Q&A

Speaker: Q&A

11:15 → 12:55 ISSA: Cybersecurity 1

Convener: Chair: Jan Eloff (University of Pretoria)

11:15 Digital Forensic Investigation and Linguistic Analysis of Cyber-crimes: A Systematic Review

Digital forensic investigation (DFI) and linguistic analysis presents unique challenges and opportunities. The study examines how DFI has evolved, noting key theories and models. These are then evaluated based on compliance with established standards and their level of comprehensiveness in terms of addressing important aspects of an investigation. Linguistic analysis of cybercrimes is then explored to establish how it has been incorporated in existing DFI models. The systematic review method is employed to gather literature sources from journals, conference proceedings, and electronic databases. The aim is to identify gaps and propose future research direction for DFI involving linguistic analysis of cybercrime, whileguiding practitioners in the field on best practices for conducting DFI of cybercrime. Findings reveal that linguistic analysis in DFI models has historically been limited, as well as research on the incorporation of artificial intelligence and machine learning. However, with the emergence of semantic analysis in digital forensics (DF), linguistic analysis is now receiving more attention and recognition of its significance.

Speaker: Mopati Bernerdict Kekgathetse (UP)

11:35 Privacy and Legal Implications Regarding the Processing of Honeypot Data

Cyberattacks have been an ever-increasing threat against the cyber infrastructure of organisations. The act of exploiting known network vulnerabilities appears to be highly appealing to hackers where their potential payout is to find and collect valuable data housed by an organisation. To compensate for this matter, security teams can design and deploy highly advanced security tools to thwart cyberattacks, and one such tool is a honeypot. Honeypots possess the functionality of baiting intruders to interact with them whilst preventing said intruders from affecting real production and service systems. Ultimately, honeypots collect data associated with an intruder and the attack, which reveals valuable information that can be analysed and used to combat similar incidences. However, with the introduction of modern privacy laws, a number of consequences exist with the data honeypots collect. The paper will explore the limitations on processing honeypot data with the aid of related works published regarding honeypots, the POPI Act and the GDPR through literature revisions. Thus, this paper will discuss the privacy and legal implications that arise with processing data collected by a honeypot from the perspective of privacy laws established by both the European Union and the South African government.

Speaker: Bahle Ngoma

11:55 Conceptual Model for taxation and regulatory governance among South African crypto asset holders

The rapid adoption of digital assets has revolutionised the global financial landscape, bringing new opportunities and challenges. In South Africa, digital asset adoption has surged,
driven by economic factors and a tech-savvy population. However, this growth has outpaced regulatory development, particularly around tax compliance. This paper proposes a
conceptual model aimed at addressing the non-compliance issues among crypto asset holders. The model incorporates advanced mechanisms for visualising crypto address interactions and generating crypto tax Non-Fungible Tokens as a verification tool. By mapping and monitoring crypto transactions, the model provides regulatory bodies with enhanced tools to track, verify and enforce tax obligations transparently and efficiently.

Speaker: Pardon Ramazhamba (CSIR NGEI)

12:15 Countering mobile application threats with a structured threat modelling approach

The development of secure mobile applications is a crucial and complex task. This research focuses on threat modelling techniques to enhance mobile application security. A technique is proposed to analyse mobile application vulnerabilities, categorised by mobile application architectural layers, and classify vulnerabilities using STRIDE and DREAD. By identifying and scrutinising vulnerabilities, the research proposes a practical and comprehensive four-step threat modelling approach to mitigating mobile ap-plication security risks and ensuring the robustness of mobile applications. The approach contributes to clarifying the steps to be taken to secure mobile applications.

Speaker: Christoff Jacobs (University of Johannesburg)

12:35 Social Network Graphs in Email Forensics: Analysis of a Personal Email Dataset

Visualisation techniques to aid in email forensic investigations was proposed in the literature, often social
network graphs. Current literature does not deal with the interpretation and insights that can be gained from
the graphs. When many nodes are depicted in such a graph, it becomes difficult to extract useful insights
from social network graphs. The research that will be presented at the conference, attempts to address this
shortcoming by interpreting a social network graph constructed from a personal email box, containing more
than 60 000 emails, with 4 380 email addresses (nodes), and 8 132 edges in the resulting graph.
The main contributions of this research are, to demonstrate how to interpret social email graphs, simplify the
graphs to improve interpretation, and identify structures which provides insights into the possible emails that
flowed, e.g. mailing lists. The main results are summarised in the research paper in the form of 11 deductions
taken from the exploratory analysis of the graph from the personal email dataset. The presentation at the
conference will focus on sharing insights of how this research can potentially be used in investigations.

Speaker: Riaal Domingues (University of Pretoria)

11:15 → 12:45 Special: Quantum Computing 1

Convener: Chair: Mr Nyameko Lisa

11:15 Building a Quantum-Ready Africa: Opportunities and Challenges

Quantum computing represents a transformative leap in technology with the potential to address complex challenges across sectors such as health care, energy, and finance. For Africa, embracing quantum technologies offers unique opportunities to drive innovation, enhance research capacity, and contribute to global advancements. This talk explores the steps needed to build a quantum-ready Africa, including capacity building, fostering collaborations, and aligning quantum innovation with sustainable development goals. By addressing key challenges such as infrastructure, education, and access, we can position Africa as a leader in this emerging frontier.

Speaker: Prof. Franchesco Petruccione* (Stellenbosch University)

11:55 A new approach on a D-wave for obtaining thermodynamic properties of a solid solution.

Alloys, solid solutions, and heavily doped systems, where compositional variation of a multi-component system produces enhanced physical and chemical properties, are critical for key technologies including energy generation and conversion, catalysis, optoelectronics, and smart buildings. However, the simulation of such disordered materials to predict their properties remains an outstanding challenge, even when crystallinity is assumed to be maintained. As x increases in the binary solid solution X_(1-x)Y_x, there is a “combinatorial explosion” resulting from the fractional occupancy of the sites in the crystallographic unit cell, so that the computation of the energy of all possible combinations becomes intractable. In this talk I will present our latest efforts to exploit the D-wave hardware to extract thermodynamical properties for a solid solution, demonstrated using three different types of materials. The approach developed avoids the poor scaling sometimes experienced when using a QUBO model.
If I have time, I will also present work where we have exploited more traditional computer resources to compute the free energy for charging the cathode of a lithium ion battery.

Speaker: Prof. Scott Woodley (UCL)

12:15 The UK National AI Research Resource “AIRR”

The government department of science innovation and technology (DSIT) and the UK science funding council (UKRI )have kick started a large scale AI national service called the AI Research Resource AIRR. This sovereign AI capability will launch in early access mode in early 2025, and consists of two AI systems an HPE NVIDEA GPU system at Bristol University and a Dell Intel GPU system at Cambridge. These two systems are to be run as a single service and accessed via a single federated access mechanism. Todays talk will outline these systems and describe some early pioneer projects that have been run on the Cambridge system as the system part of the system bring up process.

Speaker: Dr Paul Calleja (Cambridge University)

12:35 Questions

Questions

Speaker: Questions

12:45 → 13:30 Lunch

12:45 → 13:30 Lunch

12:45 → 13:30 Lunch

12:55 → 13:30 Lunch

13:30 → 15:00 HPC Applications: 2

Convener: Chair: Prof Ozlem Tastan Bishop (Rhodes University)

13:30 The Impact and Benefits of Partnering for Operational Weather Forecast Continuity

The South African Weather Service (SAWS) operates under the SAWS Act (Act No. 8 of 2001), which mandates it to provide essential weather forecasts and warnings to safeguard lives and property. To meet this mandate, SAWS runs numerical weather prediction (NWP) models on its high-performance computing (HPC) system to simulate atmospheric processes.

The current NWP configurations generate high-resolution weather forecasts for the Southern African Development Community (SADC) region (with a 4.4 km grid spacing) and South Africa (1.5 km grid spacing). Both models are run concurrently four times daily, providing forecasts up to three days ahead. These outputs support weather-sensitive sectors and stakeholders, including aviation, marine services, and disaster management.

In November 2023, SAWS relocated its head office, necessitating the transfer of critical operational systems, including the HPC infrastructure. Central to the success was the collaboration with the Centre for High-Performance Computing (CHPC). The two entities have a long-standing Memorandum of Agreement (MoA) that grants SAWS near real-time access to computational resources and backup services in case of HPC failures. This redundancy allows SAWS to maintain a significant component of weather forecasting services with minimal disruptions. Once the HPC was disabled for transfer, a mirror of the SADC (4.4 km) configuration was activated immediately on CHPC’s Lengau system. Months of meticulous planning, rigorous backup and failover testing, and continuous coordination between the teams ensured this smooth transition of NWP operations.

The effectiveness of this failover approach underscored the critical importance of resilient and redundant HPC resources in sustaining continuous forecasting operations. The seamless collaboration between SAWS and CHPC ensured uninterrupted service, emphasising the critical roles of high-performance computing and partnerships in delivering essential meteorological services across South Africa.

Speaker: Ms Stephanie Landman* (South African Weather Service)

CHPCconf2024-Abstract.0-SAWS_StephanieLandman.pdf

13:50 Computational modelling of functional materials and chemical reactivity using molecular dynamics, DFT and wave-function-based approaches

Quantum chemical methods, in particular, those based on density functional theory (DFT), can be successfully used to achieve a mechanistic understanding of reactivity at the microscopic level, which is required to optimise functional materials (such as catalysts, electrocatalysts, semiconductors, optoelectronic materials) with respect to their target properties.

The presentation will highlight the various research topics and achievements of the research programme led by L. Moskaleva at UFS. Our research applies the tools of computational chemistry to study phenomena such as gas-phase reactions, reactions in solution and at solid surfaces.

We apply DFT methods, wave-function based methods, ab initio molecular dynamics (AIMD) simulations, and microkinetic modelling to study material properties and complex reaction mechanisms, either in the gas phase, or on various solid transition-metal catalysts, including rare earth oxides and coinage metals, in particular on Au-Ag alloys. Our research efforts have been largely focused on nanoporous gold (np-Au), a novel catalyst proposed for environmentally friendly applications but also an interesting material from the point of view of fundamental research on gold. Since the discovery of the catalytic activity of gold at the nanoscale, as opposed to the very inert bulk Au, there has been an ongoing debate in the scientific community, as to whether gold on its own (without any support material) can be catalytically active. Our research has contributed to this scientific debate and related questions about the interplay between the topology, composition, and catalytic properties of alloyed nanostructures.

We also work on the topics related to chemical kinetics of hydrocarbon combustion and on functional materials, such as luminescent organometallic complexes. In collaboration with experimentalists, we have computationally characterised a series of luminescent binuclear Au-Au complexes with N-substituted bis(diphenylphosphino)amine ligands. Our time-dependent density functional theory (TDDFT) calculations provided valuable insights into the interpretation of the photophysical properties of the complexes, highlighting the phosphorescent nature of the emission and explaining the differences in emission wavelengths observed between complexes with different counterions.

We are grateful to the CHPC for providing state-of-the-art computing facilities that allow us to use computational chemistry software and perform sophisticated calculations of molecular and crystalline systems.

Speaker: Prof. Lyudmila Moskaleva* (University of the Free State)

14:10 Unravelling plant-biotic/abiotic stress using computational biology and bioinformatics tools.

Major discoveries in molecular biology, along with advances in bioinformatics technologies, have resulted in an exponential increase in the biological data supplied by the scientific community over the past few decades. Genomics, transcriptomics, proteomics, and metabolomics are four of the techniques that have impacted studies in plant-biotic/abiotic stress. When used singly, each of these procedures can generate a massive amount of data, however, when utilized together, they have the potential to fully dissect a system at the transcriptional and translational levels.

Our research focuses on understanding interactions between plants and biotic/abiotic factors for resistance/tolerance breeding purposes in crops such as cassava, wheat, sweetpotato and common bean, among others. The use computational biology and bioinformatics can enhance an understanding of plant-stress responses and aid in the development of stress-resistant/tolerant plants. These advancements can boost yields, increase agricultural productivity, and enhance global food security. Access to computing resources such as the CHPC is needed to achieve these results, hence, the presentation will showcase studies that were achieved using these tools, including the ongoing and future studies.

Speaker: Dr Molemi Rauwane* (Nelson Mandela University)

14:30 Fitting Classical Forcefields Programmatically for Application in Molecular Dynamics Simulations

Experiments rely on models and theory to study the properties of physical systems. Consequently, computer experiments or simulations complement experiments and have enabled theoretical studies of properties of physical systems through the implementation of models and theory. Interatomic potentials have been successfully applied to computational studies involving up to millions of atoms. Obtaining interatomic potentials for a particular material typically involves fitting parameters of an analytical function such as the Buckingham to reproduce experimental properties of that material. Finding good parameters out of billions of possible parameter combinations makes obtaining good interatomic potentials very time-consuming and could take up to a year or more. We have recently developed a program which greatly simplifies the process of fitting Buckingham interatomic potentials and greatly reduces fitting time to a few weeks. The program implements the OpenMPI framework enabling it to run on high-performance computing systems on any number of processors. The core of the fitting algorithm is an error variable akin to a cost function in machine learning which allowed to program to intuitively optimize the interatomic potential parameters. When compared to experimental properties, preliminary fitting performed on cubic Li2O with space group FM3-M yielded percentage differences of -0.59 % for lattice parameters a, b, and c, -0.04 % for the bulk modulus, 0.00 %, 0.00 %, -63.39 % for the elastic constants C12, C12 and C44, and -29.35 % the static dielectric constant. These results were obtained by running the program on 24 processors for 4 days on a high-performance computing system.

Speaker: Dr Kenneth Kgatwane (University of Limpopo)

14:50 Q&A

Q&A

Speaker: Q&A

13:30 → 15:00 HPC Technology: 2

Convener: Chair: Mr Hannes Steyn

13:30 H3C Presentation (TBC)

TBC

Speaker: Mr Christo Cilliers

13:50 Huawei Presentation (TBC)

TBC

Speaker: Mr Collen Maphosa

14:10 The Data Platform to Accelerate HPC Deployments (VAST Data)

The VAST Data Platform accelerates HPC by delivering the performance and scale of parallel file systems with the simplicity of NAS, all at archive economics. Powered by VAST’s DASE architecture, our customers build scale-out systems delivering TB/s and millions of IOPS of performance without the rigidity and overhead of legacy architectures.

Speaker: Mr Kim Ormiston (Sponsor)

14:30 NVIDIA Accelerated Computing (NVIDIA)

NVIDIA’s Accelerated Computing Platform is designed to accelerate various workloads, including AI, HPC, Professional Visualisation & Gaming. The platform provides faster performance, increased productivity, improved accuracy, and cost savings for organizations across industries.

NVIDIA’s Accelerated Computing Platform is designed to accelerate various workloads, including:

• Artificial Intelligence (AI): NVIDIA’s GPUs and Tensor Cores accelerate AI workloads, such as deep learning, natural language processing and computer vision.
• High Performance Computing (HPC): NVIDIA’s GPUs and networking technologies accelerate HPC workloads, such as scientific simulations, data analytics and visualization
• Professional Visualization: NVIDIA’s GPUs and Quadro professional visualisation workloads, such as 3D modelling, engineering, and video production
• Gaming: NVIDIA’S GeForce GPUs accelerate gaming workloads, providing fast performance, low latency, and high-quality graphics

NVIDIA’s Accelerated Computing Platform:

• NVIDIA GPU Architecture: NVIDIA’s GPU Architecture is designed to accelerate compute -intensive workloads
• CUDA: CUDA is NVIDIA’s parallel computing platform and programming model and allows developers to harness the power of NVIDIA GPU’S
• NVIDIA Networking: NVIDIA’s networking technologies such as NVIDIA Mellanox, provide high-speed networking solutions for accelerated computing workloads
• NVIDIA Software: NVIDIA provides a range of software development kits (SDKs) and tools to support accelerated computing workloads.

Numerous Industries benefit from NVIDIA Accelerated Computing, just to name a few: Healthcare (acceleration of medical imaging, genomics, and drug discovery workloads) Finance (acceleration of financial modelling, rick analysis and trading workloads) Scientific Research: (acceleration of scientific simulations, data analytics and visualization workloads) Gaming and Professional Visualization: (faster performance and high-quality graphics)

Speaker: Ms Belinda Leigh Duke (Partner)

14:50 Q&A

Q&A

Speaker: Q&A

13:30 → 15:00 ISSA: Cybersecurity 2

Convener: Chair: Kerry-Lynn Thomson (Nelson Mandela University)

13:30 Guidance for the Creation of Management and Cost Accounting Computerized Information Systems: Assurance of Information Integrity and Flexibility

Current corporate governance literature emphasizes providing information that maintains integrity to demonstrate accountability through assurance. The Institute of Directors in Southern Africa (IDSA) suggests the implementation of a Combined Assurance Model (CAM) that includes the usage of various assurance controls to support the integrity of information through a strong control environment. IDSA further asserts that "Technology is now part of the corporate DNA" and that the perspective on financial information reporting is evolving, emphasizing that governing bodies cannot claim ignorance of this transformed business/organizational environment. One of the practices that can be used for the directing and controlling of an organization is called Management and Cost Accounting (MCA). The information produced by cost accounting methods (systems) is the information used for financial accounting and management accounting directing and controlling purposes. In recent years, the practice of MCA has moved to MCA Computerized Information Systems (CIS), also known as Enterprise Resource Planning (ERP), Supply Chain Management (SCM) and Customer Relationship Management (CRM) systems. However, the implementation of prominent existing MCA CISs has proven challenging. These systems often lack the flexibility needed for MCA, leading to information integrity issues. This research paper reviews literature regarding governance and technology to propose guidance on creating a strong technological MCA CIS control environment (an MCA CIS CAM) that maintains flexibility and information integrity for the practice of MCA and ultimately aids the demonstration of accountability through assurance.

Speaker: Emile Westraadt

13:50 Using artificial intelligence techniques to standardize intelligent and automated generation of digital forensic reports.

Abstract. The present study investigates if the digital forensics report can be generated automatically by using some of the artificial intelligence techniques, specifically the natural language processing. A model has been developed to assess if it is feasible to automate the generation of a digital forensic report using artificial intelligent techniques. One of the main purposes for this study is coming from a point where human errors, structure of the reports, critical evidence that should take part of the digital forensic report are omitted during the generation of digital forensic report as well as the interpretation of the evidence drafted by an investigator during investigation. In addition, the standardization of this report happens to be imminent, especially when it is being presented in a court of law. Given the rise of cybercrime, more research is needed to better improve the process of automating the generating digital forensic report using some intelligent techniques.

Speaker: Idani Mulaudzi

14:10 A Decentralized E-voting System Using Blockchain

This paper presents the design, implementation, and comprehensive evaluation of a decentralized blockchain-based voting system aimed at revolutionizing electronic voting (e-voting). Leveraging blockchain technology, the system offers a transparent, secure, and publicly verifiable voting platform, addressing key limitations found in traditional e-voting approaches. The systememploys Proof of Work (PoW) as its consensus mechanism, ensuring strong security through computational challenges. To protect vote integrity, the Digital Signature Algorithm (DSA) with SHA-256 is utilized for signature authentication, while the Advanced Encryption Standard (AES) ensures data confidentiality. Furthermore, Elliptic Curve Cryptography (ECC) enables secure and efficient public vote auditing. Extensive experimental evaluations, including tests against attacks such as vote sniffing, signature spoofing, and denial of service (DoS), demonstrate the system’s robustness and resilience. The results confirm that the proposed blockchain architecture significantly enhances security and transparency, contributing to the evolving landscape of e-voting. This work underscores the potential of decentralized platforms to transform electoral processes by bolstering trust, accessibility, and overall democratic integrity.

Speaker: Daniel Ramotsoela (University of Cape Town)

14:30 Changes in Observed Internet Background Radiation Traffic in the 155/8 netblock

This research investigates changes in Internet Background Radiation (IBR) by analysing data captured from network telescopes. Network telescopes provide a unique insight into unsolicited network traffic and can be indicative of widespread malicious activity. The primary focus of the study is on a comparative analysis between network data from 2017 and 2023, captured from the same IP netblock. The methodology is grounded in descriptive statistical analysis. Among our findings were changes in protocol distribution, with an increase in TCP traffic, a decrease in UDP traffic, and a substantial increase in ICMP traffic, primarily from Russia, while observing a notable decrease in the Russian overall attributed traffic. A sharp decrease in specific destination port targeting for both TCP and UDP traffic suggests broader scanning activity than before.

Speaker: Barry Irwin (Noroff University)

14:50 Q&A

13:30 → 15:00 Special: Quantum Computing 2

Convener: Chair: Mr Nyameko Lisa

13:30 Solving a quadratic optimization problem by means of quantum computing using amplitude encoding

Quantum computers have the potential to be faster at solving certain problems, such as optimization problems, than their conventional equivalents [1]. These speedups are made possible by the fact that quantum computers are based on quantum bits (qubits), which may use superposition or entanglement, two peculiar properties of quantum physics. In this work, we explore the quantum gradient descent method [2] and describe a modified version of it to find the minimum of a quadratic cost function. We simulate the algorithm using the so-called amplitude encoding technique, to approach the minimum of a quadratic cost function of 2 variables and of 8 variables and verify the result. We present the quantum circuit for one step and iterate it to find the optimal solution after several steps.

Speaker: Helarie Rose Medie Fah

13:50 Quantum Computing: a tool for simulating physical systems in photonics

Quantum computing offers capabilities for simulating complex physical systems. In photonics, it plays a crucial role in modelling the behaviour of both bright laser beams and single-photons that are propagated through the atmosphere, turbid, and other complex media. This can be crucial for applications in fields such as biological imaging, LiDAR, laser light communication and surveillance systems. In this talk, I will demonstrate how quantum computers can assist in modelling laser beam propagation through complex media, address inference problems in optics, and enable for the classification and characterization of optical fields.

Speaker: Isaac Nape (University of the Witwatersrand)

14:10 Data base search through quantum imaging

Searching algorithms play a crucial role in quantum computing, enabling the efficient identification of specific elements within large datasets. Grover’s algorithm, a key example, significantly speeds up searches for unstructured data compared to classical methods. Recent advancements have shown the potential of using optical fields as a computational resource to implement Grover’s algorithm, highlighting the advantages of quantum techniques for solving unstructured search problems. We present a novel approach to performing search algorithms through quantum imaging, establishing a link between search algorithms and quantum imaging techniques through complex light. By employing spatially entangled photon states, we demonstrate the equivalence between quantum ghost imaging process and Grover’s algorithm. Our results indicate that entangled structured light offers computational advantages, requiring only a single iteration in our optical adaptation - unlike the typical √N iterations due to the higher-dimensional nature of photon wave-functions

Speaker: Paola Andrea CONCHA OBANDO (Wits University)

14:30 Training phase masks for diffractive networks / Emulating quantum computing with optical matrix multiplication

Training phase masks for diffractive networks

Diffractive optical networks have been shown to be useful in a wide variety of application in the realms of optical computing and information processing, such as modal sorting and multiplexing. These systems utilise repeated phase modulations to transform a set of input states contained within a particular basis into a set of target states contained in some new arbitrary basis. Traditionally, these phase masks are trained iteratively through randomly generated matrices or by optimising for individual pixels using various parameter search methods. In this work, we construct a diffractive network capable of sorting various structured light modes into pre-defined channels using optical aberrations. As such, we leverage phase masks constructed from a superposition of Zernike polynomials, whose component weightings are trained analogously to the weightings within a neural network. In this way, we can solve for the desired transformation by optimising for the weightings of fewer aberrations rather than many individual pixels, considerably reducing the number of variables needing to be solved.

Emulating quantum computing with optical matrix multiplication

Optical computing offers a powerful platform for efficient vector-matrix operations, leveraging the inherent properties of light such as interference and superposition, which also play a crucial role in quantum computation. In this work, we bridge classical structured light with quantum computing principles by reformulating photonic matrix multiplication using quantum mechanical concepts like state superposition. We highlight the tensor product structure embedded in the cartesian transverse degrees of freedom of light, which serves as the foundation for optical vector-matrix multiplication. To demonstrate the versatility of this approach, we implement the Deutsch-Jozsa algorithm, utilising a discrete basis of localized Gaussian modes arranged in a lattice formation. Additionally, we illustrate the operation of a Hadamard gate by harnessing the programmability of spatial light modulators (SLMs) and Fourier transforms through lenses. Our results underscore the adaptability of structured light for quantum information processing, showcasing its potential in emulating quantum algorithms.

Speakers: Hadrian Bezuidenhout (University of the Witwatersrand), Mwezi Koni (University of the Witwatersrand)

14:50 Q&A

15:00 → 15:30 Break

15:30 → 17:00 Keynote: Industry Inquisition

Convener: Facilitator: Mr Dan Olds (OrionX)

15:30 A Brief Tour of Your Data Center Future

In this informative and fast-moving presentation, long-time industry analyst Dan Olds conducts a whirlwind trip discussing why the HPC/AI market is increasingly difficult to measure and highlights new technologies that will help data centers deal with rapidly increasing compute demand. Spoiler alert: he declares air cooling dead and explains why. As his big finish, Olds boldly declares that the only way forward for data centers is to become radically more efficient. Not just talk about it, but actually do it.

Speaker: Mr Dan Olds* (Olds Research)

15:50 Industry Inquisition

This session has become a signature event at CHPC conferences. The rules are brutally simple. Vendors have five minutes and only three slides to put their best foot forward to the audience and the inquisitors. The panel includes industry analyst Dan Olds along with two standout students from the cluster competition who have been briefed on the vendors and their slides.

After their five-minute presentations, the presenters will be asked three questions, two of which they know are coming followed by a final, secret, question. Frank and tough questions will be asked. Answers will be given. Punches will not be pulled. The audience will be the ultimate judge of which vendor did the best job. It’s fun, brisk, and informative.

Speaker: Mr Dan Olds (Olds Research)

17:00 → 17:45 Keynote

Convener: Chair: Dr Happy Sithole (CHPC)

17:00 Challenges and Opportunities of HPC in the age of AI

Large scale AI applications are drivers for the design and deployment of the next generation of supercomputers. While large language model training and generative AI applications take the headlines, scientific workloads are starting to utilize AI as algorithmic extensions to their existing implementations. We will discuss how the needs between these communities differ, how system software and system middleware need to develop to support these use cases, and hopefully demonstrate how once again supercomputing turns compute-bound problems into I/O-bound problems.

Speaker: Dr Utz-Uwe Haus* (HPE HPC/AI EMEA Research Lab (ERL))

17:45 → 18:00 Intermission

18:00 → 19:30 Conference Dinner

Tuesday, 2 December

06:00 → 07:00 Morning Walk Wellness

Wellness and Social activity: meet at 6am in front of the City Lodge hotel. A walk for 1 hour. Hosted my Mervyn.

08:00 → 09:00 Registration

09:00 → 10:30 Keynote: Tuesday

Convener: Chair: Mr Sabelo Dlamini

09:00 Building global federated infrastructure in the R&E community

Trust is the basis for any human and technical interaction. Research collaborations have been sharing resources and data for decades; however the open science wave is pushing to make scientific research and its underlying data more accessible to all levels of society,

This talk explores how trust is established in research collaborations and infrastructures and how Authentication and Authorisation Infrastructures (AAIs) allow authorised users to access shared resources. Examples of these AAIs, along with lessons learned are provided for the European Open Science Cloud (EOSC) and EuroHPC.

Speaker: Ms Licia Florio* (NORDUnet)

09:45 Cyber Intelligence at Scale: Risk Evaluation of DNS as Critical infrastructure

The domain name system, or DNS, is a critical component of the Internet ecosystem we use. Almost every single transaction and connection from email to online commerce makes use of DNS as an initial a fundamental step. While the primary purpose in the eyes of the public is to mask the complexities of host addressing, and location, it’s use has evolved to be critical for a whole lot more, One of the oldest and arguably the second most important use being its foundation for email delivery though the use of MX records. In recent years we have seen the introduction and gradual adoption of several security measures rooted within and implemented by extending the DNS protocol. These include DNSSEC, SPF, and more recently CAA for improving e the security of SSL certificate issuance. In essence the global domain Domain Name Systems should be regarded as critical infrastructure. However, for many organisations, especially those reliant on hosting providers, ISPs or MSPs, despite the requirement for functional DNS, the deployment and operation of the servers (as outlined in RFC 2182) and associated domain zones, are often neglected. This may be due to the ‘care and feeding’ been seen as 'too complex', mundane or unexciting in comparison to more exciting areas with ‘Cyber operations’ such as Threat Intelligence, Malware Analysis and ML/AI based security solutions. The irony is these all have a strong dependence on DNS!

This talk has a dual focus initially presents an overview of the state of DNS operations for several ccTLD’s and top domains globally. A concern worth raising particularly considering the increased global geopolitical tensions is where is ones DNS hosted physically and logically, and who has control? An evaluation of risk, particularly the dependency on key providers (for example about a third of the .no domains surveyed are hosted by a single provider), as well as adherence to good practice is presented. The secondary part of the talk presents several short case studies of the adoption rate of security functionality (primarily the adoption of DNSSEC and CAA records) within and offered by DNS for ccTLDs investigated.

The final element is a discussion about undertaking research such as this at ‘internet scale’, including data collection, processing storage and validation.

Speaker: Prof. Barry Irwin* (Noroff University)

10:30 → 11:00 Break

11:00 → 12:30 HPC Applications: 3

Convener: Chair: Prof Regina Maphanga (CSIR)

11:00 Turbomachinery Modelling on the CHPC

The turbomachinery research group at Stellenbosch University consists of five to seven postgraduate (master's and PhD) students that are all doing CFD simulations of rotating machinery. The work stretches from modelling the noise and performance of large diameter cooling fans (used by Eskom), to the performance of small centrifugal compressors (used in a solar Brayton cycle) to the development of rocket engine turbines (used by the UKZN ASRI group).

The focus of the work is on supporting local technology and in all the cases, the simulations have been coupled to actual experimental evaluations. The group has been making use of the CHPC for its CFD simulation work during the past five years. The use of the CHPC enables the accurate and detailed modelling of aspects like gas turbine combustion, fan noise and blade tip leakage flow that have not previously been possible within our group. Mesh sizes of 20 million + are now common and the information gained from these simulations give researchers the ability to present their work next to researchers from much more famous entities.

The results achieved to date have been world class and the group is continuously trying to increase the complexity of its outputs and it is hoped that they will shortly be able to model 100 million + size meshes.

Speaker: Prof. Johan van der Spuy* (Stellenbosch University)

11:20 High Resolution Modelling of Tropical Cyclone Characteristics using the Lengau Cluster

Parts of southern Africa and Madagascar are majorly impacted by landfalling tropical cyclones (TCs). Despite the devastating impacts associated with these systems, little is known about the changing attributes of TCs in a warmer world, especially in this region. Therefore, we are undertaking novel, bespoke high resolution climate simulations to generate the first convective-scale (km-scale) ensemble projections of future climate change over southern Africa. These simulations were generated using the Conformal Cubic Atmospheric Model (CCAM) run on the CHPC supercomputer. To optimise modelling of TCs, a set of experiments were conducted using ERA5 reanalysis data, to test different model setups on specific, historical events. Namely, TC Idai (the deadliest event in region), TC Freddy (the longest-lived event), TC Eline (the second longest lived event) and TC Kenneth (the strongest event to make landfall in Mozambique) were considered. This then informed the setup for the downscaling of five CMIP6 models, which enables for the examination of projected changing attributes of tropical cyclones in this region. The outputs from this study will be used in compound flood risk assessments and enable further research into climate change adaptation options and improved early warning in the region.

Speaker: Dr Jessica Steinkopf (Global Change Institute, Wits University)

11:40 Density functional theory of thermoelectric properties of graphene

Thermoelectricity is the phenomenon direct and thermodynamically reversible conversion of heat to electricity and vice versa. Graphene is one of the carbon-based materials with a low dimensions and mechanically robust, and it tend to have better performance in thermoelectric. Electronic and thermoelectric properties were computed and analysed using density functional theory with a full-potential all-electrons linearised augmented plane waves. The electronic band structure shows a zero band gap with conduction and valence band meeting above the fermi level between the symmetry points H and K. The thermoelectric transport coefficients, Seebeck coefficient, electrical conductivity, thermal conductivity, power factor and figure of merit were calculated using first principle of density of states. Furthermore, a high Seebeck coefficient, electrical conductivity and high power factor were recorded, and these make graphene to be one of the good thermoelectric material and might be applicable for solar cells applications.

Speaker: Dr Moshibudi Ramoshaba (University of Limpopo)

12:00 Multiscale modelling of chemical systems on HPC and quantum computers

Multiscale quantum mechanical/molecular mechanical (QM/MM) techniques are a well established approach for simulating chemical reactivity including a realistic description of the surrounding environment. The ChemShell project is a leading software package for performing QM/MM simulations, developed at the UK’s Daresbury Laboratory and collaborating research groups around the world. The latest Python-based, open source version of ChemShell maintains a strong emphasis on performance on high performance computers and is widely applicable across a range of research areas from enzyme modelling to materials chemistry. In this talk, recent developments in ChemShell are explored focussing on their computing aspects, including complex workflows targeting exascale calculations through the UK’s ExCALIBUR programme, and efforts towards integrating HPC with emerging quantum computing technology in multilayer embedding schemes.

Speaker: Prof. Thomas Keal (STFC Scientific Computing)

12:20 Q&A

Q&A

Speaker: Q&A

11:00 → 12:30 HPC Technology: CHPC Cloud

Convener: Chair: Ms Dorah Thobye (Staff)

11:00 Bridging the Digital Divide with Cloud-Hosted Geo-Location Spectrum Databases

Affordable and widespread broadband access is essential for achieving the United Nations Sustainable Development Goals (UN SDGs) cantered on connectivity. This presentation highlights efforts to close the digital gap in South Africa’s underserved rural and township communities through innovative use of a CSIR-developed, Geo-Location Spectrum Database technology, hosted by the Centre for High Performance Computing. Utilizing Dynamic Spectrum Access principles, this technology taps into underused radio spectrum to enable wireless network operators to deliver cost-effective wireless broadband to remote areas. This CSIR developed technology has empowered women- and youth-led ICT SMMEs to deploy and manage low-cost network infrastructure, providing affordable internet to schools, businesses, and households in previously disadvantaged regions. By creating job opportunities and fostering digital inclusion, this project supports national goals to accelerate progress towards the UN SDGs.

Speaker: Luzango Mfupe* (CSIR)

11:20 Image based analysis and crop prediction using smartphone.

The 2019 White Paper on Science, Technology, and Innovation in South Africa
emphasises the core themes of inclusivity, transformation, and partnerships. Following
the introductory discussion with the farmers, it became clear that the majority of them have
an urgent need for up-to-date soil-health data and assistance with managing their crops
in order to maximize their profits. Addressing this, our consortium aims to create a
smartphone app that will empower small-scale farmers with immediate soil-health
predictions derived from soil digital images. Additionally, this app will extend its
functionality to provide crop recommendations and assist the farmer in smart decision-
making by incorporating regular weather updates and nowcasting of local weather
parameters. An integrated energy and sustainability management module will be
designed to monitor and control energy consumption and carbon emission. This project
proposal unites partners from South Africa, Scotland, and Turkey, leveraging their
collective expertise in science, agriculture and digital technologies.

This innovation integrates advanced technologies such as digital image processing (DIP),
machine learning (ML), and artificial intelligence (AI), with deep-learning algorithms and
large language models (LLM) used for ML training, and AI chatbot services for enhanced
functionality. The app architecture includes a mobile front-end API connected to a
dynamic back-end system, featuring a database, knowledge base, and external APIs, with Python serving as the core programming language. Therefore, this project is using
Sebowa Cloud Services – Commercial (4x vCPU, 16GB RAM, 200 GB Disk).

Speaker: Lindokuhle Mdletshe* (INNOBIZ Group (Pty) Ltd)

ABSTRACT_CHPC_National_Conference_2024.pdf

11:40 OpenStack Adoption at CHPC: Advancing AFIS and Emerging DevOps Projects Through Scalable Cloud Solutions

OpenStack is a cloud computing platform utilized by the Centre for High Performance Computing (CHPC) to manage a wide range of software services. Its user-friendly interface enables the creation, scaling, and management of virtual machines with minimal effort, offering a flexible and scalable Infrastructure as a Service (IaaS) solution. The Advanced Fire Information System (AFIS), is a satellite-based tool that delivers near real-time fire information to users globally. AFIS is an internationally acclaimed Software as a Service (SaaS) which is traditionally hosted on self-managed infrastructure. AFIS’s backend and frontend systems are being transitioned to CHPC's OpenStack environment. This migration has been successful, with several AFIS subsystems deployed without performance issues. The shift to OpenStack provides significant advantages, such as mitigating the risks associated with infrastructure failure and reducing the need for manual intervention in server management. Costs associated with server maintenance and infrastructure are significantly saved on. Minimal modifications were required for the AFIS subsystems software deployment. Looking ahead, OpenStack presents a promising solution for migrating other current projects, including community safety project backends and geocoding engines and emerging DevOps applications such as the TB, E-Coli and HPV projects. Lastly, the integration of OpenStack has facilitated enhanced collaboration with other CSIR groups, CHPC, SANReN, and DIRISA.

Team members: Kirodh Boodhraj (Project Lead), Tau Makgaile, Karen Steenkamp, Gert Wessels, Chris Mahlathi, Phelisa Ntayiya, Roger Daniels and Lufuno Vhengani (previous team member).

Speaker: Kirodh Boodhraj* (CSIR)

Kirodh_Boodhraj_Abstract.pdf

12:00 Infrastructure consideration for Science Engagement Information Management Systems on CHPC

The Science Engagement Information Management Systems (SEIMS) is a digitalisation initiative of the Department of Science and Innovation (DSI)’s Science Engagement (SE) Programme in collaboration with the CSIR-NGEI and other stakeholders. The SE programme has the main objective of creating a society that is scientifically literate towards socio-economic emancipation. SEIMS consists of a number of initiatives run by different stakeholders with the aim of expediting the monitoring and evaluation processes in SE programmes by eradicating manual data management. The infrastructure of SEIMS is split between production and testing environments. The production environment is publicly available and requires high availability. As a result, a topology that supports these requirements was adopted. On the other hand, the testing environment mainly required availability and a topology supporting this requirement had to be adopted. For this purpose, the CHPC infrastructure, a DSI supported cyber-infrastructure was utilised. The CHPC
collaborated with the CSIR-NGEI’s Software Architecture and Solutions (SAS) group for the sourcing and configuration of the infrastructure necessary for the SEIMS initiative. This presentation with outline experiences and lessons learned through utilization of the infrastructure and how it helped fulfill the requirements and demands of the SEIMS product.

Speaker: Muzi Matyila* (CSIR)

12:20 Q&A

11:00 → 12:30 ISSA: 3

Convener: Chair: Barry Irwin (Noroff University)

11:00 A chatbot LLM based access control mechanism

Recent Artificial Intelligence(AI) advancements, notably in Large Language Models(LLMs), have enhanced Natural Language Processing(NLP) capabilities like Text-to-SQL. Businesses are increasingly using LLMs for domain-specific applications such as chatbots, but this raises security concerns including data access control. This research addresses these concerns by developing a secure access control mechanism for Text-to-SQL applications. While there exists literature that aims to improve the technical aspects of Text-to-SQL systems, it lacks solutions for access control. This paper proposes a prototype integrating an access control layer within the Text-to-SQL process to ensure secure and authorized data access while maintaining usability and performance. The research is validated through the development of a domain-specific chatbot prototype that demonstrates its effectiveness in mitigating security related access control risks.

Speaker: Christian Still (University of Pretoria)

11:20 Enhancing Forensic Readiness through an Integrated Approach to Fraud Risk Management throughout the Digital Transformation Lifecycle

In the contemporary digital landscape, organisations are increasingly undertaking complex Digital Transformation initiatives to enhance, among other aspects, operational efficiency and drive innovation. However, these transformations expose organisations to heightened risks related to digital crimes. Traditional Digital Forensic Readiness frameworks fail to effectively integrate within the Digital Transformation lifecycle. This shortcoming leaves organisations vulnerable to sophisticated fraudulent activities and hampers
their ability to effectively respond to and investigate digital crimes. This study employs a mixed-methods approach, beginning with a qualitative phase grounded in secondary research. The study concludes with a discussion on the Forensic Readiness Architecture (FORAC) Continuum, a new novel approach designed to embed Digital Forensic Readiness throughout the Digital Transformation lifecycle. Finally, the study presents a guideline on the application of the FORAC Continuum, to facilitate self-assessment and for use by organisations undertaking Digital Transformation initiatives.

Speaker: Antonio Pooe (University of Johannesburg)

11:40 Solving the Privacy and Security Challenge using ZKP: Its Positive Impact on the Economy

The 4th and 5th industrial revolutions are improving the functioning of the working environment for different industries. However, it also introduces security and privacy challenges that lead to cybercrime which negatively impacts the economy. This article presents a comprehensive analysis of the usage of Zero Knowledge Proofs (ZKP) to protect data in the major technologies for the 4th and 5th industrial revolutions. These technologies include cloud computing, big data, Internet of Things, blockchain, 5G, artificial intelligence, and supply chain. Security and privacy challenges and solutions in these technologies were investigated. ZKP, a cryptographic method that enables verification of a party without revealing confidential details, is one of the promising solutions to fight the problem of data security and privacy. In this study, it was determined that Blockchain is the leading technology in terms of using the ZKP to improve security and enhance privacy. The paper provides the future direction to secure these technologies using cryptographic methods such as ZKP.

Keywords: Security, privacy, ZKP, 4th Industrial Revolution, 5th Industrial Revolution.

Speaker: Mrs Siphelele Myaka

12:00 Biometric authentication: Stress as a factor in keystroke dynamics

Biometrical authentication systems are gaining prominence and have become increasingly
important to ensure compliance with privacy and safety regulations. In this paper, keystroke dynamics as a
behavioral biometric approach to user authentication is evaluated in terms of the impact that stress may have
on the typing pattern of a user. To achieve this, several experiments were conducted with a group of users
that comprised working users from the industry as well as students. The experiments included stress factors
such as a physical limitation (use of the non-dominant hand to type), a time constraint, and a knowledge
constraint (typing in a foreign language). The results were compared to a baseline (normal circumstances)
typing pattern. Typing data were recorded and analyzed by a keystroke software package called
GenoGraphiX-Log 2.0. The study revealed that stress is indeed a factor in keystroke dynamics and that
typing patterns in some cases significantly differ from the normal typing patterns. This in turn may influence
the efficiency of the use of keystroke dynamics as a biometric authentication system.

Speaker: Monique Mac Donald

12:20 Q&A

11:00 → 12:30 SA NREN: SADC NRENs

Convener: Chair: Siyabonga Peter (CSIR)

11:00 Collaboration and Resource Sharing Among SADC Institutions: Are NRENs Supporting These Initiatives?

Collaboration and resource sharing are essential components for enhancing research,
education, and technological development across the Southern African Development
Community (SADC) region. National Research and Education Networks (NRENs) play a

critical role in facilitating these initiatives by providing high-speed, reliable, and cost-
effective networking infrastructure for academic and research institutions.

This paper examines the extent to which NRENs in the SADC region are supporting
collaborative efforts and resource-sharing initiatives among institutions. It explores how
these networks enable access to shared resources such as digital repositories,
educational content, collaborative platforms, and high-performance computing
facilities. The study further assesses the challenges and opportunities associated with
leveraging NRENs for fostering regional cooperation in research and development, as
well as the alignment of NREN services with the broader objectives of SADC in terms of
integration, sustainable development, and innovation.
Finally, the paper discusses the role of international partnerships and funding
mechanisms in enhancing the capabilities of NRENs to support cross-border
collaboration and resource sharing in a rapidly evolving digital landscape.

Speaker: Dr Frank Seth (TERNET)

11:30 Ubuntunet Alliance – For Research & Education Networking

UBUNTUNET ALLIANCE FOR RESEARCH AND EDUCATION NETWORKING is the alliance of Eastern and South African National Research and Education Networks (NRENs) supporting a vibrant connected research and education community

Speaker: Eliakim Okundi (UbuntuNet Alliance for Research and Education Networking)

12:00 MaREN – Malawi Research and Education Network

MaREN – Malawi Research and Education Network

Speaker: Solomon Dindi (MAREN)

11:00 → 12:30 Special: Digital Leadership

Convener: Chair: Dr Ricardo Harry (CSIR)

11:00 A multifaceted approach to the IS ecosystems implementation exploring multiple Conceptual and Theoretical Frameworks

This study investigates the implementation of Information Systems (IS) ecosystems through a multifaceted approach, integrating is ecosystems such as Artificial Intelligence (AI), Cloud Technologies, and Quantum Computing. It examines multiple conceptual and theoretical frameworks to understand how these technologies synergize in modern IS ecosystems. AI drives intelligent automation and decision-making, cloud platforms ensure scalable infrastructure, and quantum computing addresses computational challenges with groundbreaking efficiency. By exploring these intersections, the paper provides insights into designing robust, adaptable IS ecosystems that advance technological frontiers while addressing integration, scalability, and security challenges.

Speaker: Dr Ricardo Harry (CSIR)

11:20 Building NICIS’ Cloud: Localising the Cloud Platform

The introduction of cloud services in NICIS created new opportunities, but for these opportunities to be realised changes in mindset are required for the organisation. This presentation discusses the change in mindset and what this means for how NICIS manages its cloud service.

In response to the new mindset NICIS is building a localised cloud platform. The localised cloud platform establishes a team of cloud experts in NICIS to design and support a reference cloud architecture using open-source software. Work to date on the localised cloud platform will be presented along with the announcement of a new zone of Sebowa that will be available for public use early next year.

The presentation will end with a discussion on the work underway in the development cloud and with input being solicited from the audience as to what they would like to see added next.

Speaker: Mr David Macleod (CHPC)

11:40 Guiding the adoption of Generative AI and Next Generation NLP

Generative AI and NLP solutions are quickly becoming line items in technology roadmaps across private and public sectors. This presentation provides a brief overview of the technology advancements we seek to explore to broaden and uplift sectors as well as some of the challenges generally faced when attempting to adopt these technologies in the real world.

Speakers: Avi Moodley, Privolen Naidoo

12:00 Leveraging Hybrid Cloud, HPC and Quantum Computing Resources

Quantum Computing is a disruptive emerging technology, promising the capability to solve complex problems that were previously considered intractable even on large scale HPC systems. This talk will briefly introduce quantum computing, outlining some of the opportunities and challenges it presents specifically in a South African context. Through the utilization of a demonstrative use-case, i.e. computational quantum chemistry, the talk will showcase how NICIS is uniquely poised as the hub of hybrid cloud, HPC and quantum computing resources and infrastructure. Finally the talk will focus on the transformative impact of strategically democratizing access to such and other technologies, culminating in quantum researchers leveraging cloud and HPC infrastructure, as well as classical HPC users encouraged to adopt quantum computing resources.

Speaker: Mr Nyameko Lisa

12:20 An Emerging Leadership Paradigm for Navigating a Rapidly Changing World

The rapid evolution of digital technologies has transformed leadership dynamics, particularly in fields requiring advanced technological deployment. This is especially critical in South Africa, where advanced technology plays a pivotal role in enabling scientific research, innovation, and socio-economic development. Despite the growing importance of digital leadership, existing research still needs to be expanded. It often focuses on individual leadership traits rather than collective, relational, and adaptive practices.

This research investigates how leadership manifests as a practice through interactions between leaders and stakeholders, emphasizing its collective, relational, and adaptive dimensions. Adopting the Leadership-as-Practice (L-A-P) framework, the study shifts the focus from individual-centric approaches to leadership as a shared, social, and contextual phenomenon. Using a phenomenological methodology, the research draws on the lived experiences of participants engaged in deploying HPC infrastructure at the Centre for High-Performance Computing (CHPC) in Cape Town.

Speaker: Mr Mervyn Christoffels (CHPC)

12:30 → 13:30 Lunch

13:30 → 15:00 HPC Applications: 4

Convener: Chair: Dr Njabulo Gumede (Walter Sisulu Unibversity)

13:30 Engineering Strategies for Perovskites Materials

The depletion of fossil fuels and rapid growth in world population are the main drivers of research interests to find alternative renewable energy sources that could alleviate the global energy crisis. Hence, perovskite solar cells have been largely explored as a prospective source of clean and renewable energy. They have shown remarkable progress with rapid increases in power conversion efficiency, from early reports of approximately 3% in 2009 to over 25% today. Despite their excellent optoelectronic characteristics such as tuneable band gap, high absorption coefficients, high carrier mobility, and long diffusion lengths for electrons and holes, small effective masses and facile fabrication; they still have a number of drawbacks that hinder their practical application and commercialisation. Perovskite solar cell devices must retain high efficiencies while exhibiting decent stability and acceptable degradation for practical applications. Herein, using first-principle approach we explore different engineering strategies for various perovskites materials, namely, all-inorganic halide perovskites, organic-inorganic perovskites and double perovskites crystal structures and their respective optoelectronic characteristics.

Speaker: Prof. Regina Maphanga* (CSIR)

13:50 Cosmology with radio telescopes: finding a needle in the datastack

High performance computing (HPC) has been a crucial factor in the development of cosmology in the past few years. The tension between increasing precision and computing limitations drives a lot of our research work in the development of new algorithms. I will review the research done by my group in the context of HPC focusing on 3 main topics: 1) large scale simulations of the early Universe; 2) model fitting and 3) data processing, in particular with the radio telescopes installed in South Africa. I will conclude with an outlook of the future HPC needs for observational cosmology in South Africa.

Speaker: Prof. Mario Santos* (University of the Western Cape)

14:10 1KSA - Decoding South Africa’s Biodiversity

South Africa is one of the most biodiverse countries in the world with many institutions researching and documenting local biodiversity. However, South African scientists often conduct biodiversity-related genetic sequencing overseas, to take advantage of competitive prices internationally, and the data follow. This contributes to a drain of skills, knowledge and opportunity out of South Africa.

1KSA (www.1kSA.org.za), a DIPLOMICS initiative, was launched in 2023 to demonstrate that large-scale genetic sequencing and the associated data analysis (bioinformatics) can be done in country, thereby mitigating the brain drain and ultimately benefiting people of South Africa.

Using data generated using Oxford Nanopore Technology (long-read) to conduct whole genome sequencing the 1KSA draft genome assembly pipeline has been successfully implemented on the Centre for High Performance Computing (CHPC) within nextflow and makes use of seriallong and bigmem resources. To date, 1KSA has assembled the genomes (to draft level) of nearly 50 South African species on the CHPC - 29 plants; 4 mammals; 13 fish and 3 anthropods. The expected genome size of these draft genome assemblies ranges from 162.3 Mb to 2.6 Gb. However, there are still some computational challenges that need to be addressed to tackle the sequencing and assembly of larger genomes.

Meanwhile, the successful assembled draft genomes and the raw data are stored using the Data Intensive Research Initiative of South Africa and are made known via the generation of species information cards on the 1KSA website. These sequenced genomes of biodiversity and economically important species are to become tools to enable researchers to investigate species populations, conservation and the impacts of climate change, as well as enable the discovery of novel compounds.

Speaker: Dr Patricia Swart (DIPLOMICS)

14:30 HPC in all the right places – accelerating modelling workflows in metallurgical engineering applications

Our modern technological societies depend critically on metals and alloys. Producing these materials from their original sources, which can be either natural (mined ores) or man-made (recycled wastes), is an expensive and energy-intensive process. In addition, it is damaging to the environment – production of steel alone currently contributes about 8% to the world’s total carbon dioxide emissions. Due to the phenomenological complexity of many metallurgical engineering applications, computational modelling tools play an important role in the optimisation of existing processes and the design of new ones. This is especially true today when parts of the industry are under pressure to change rapidly to address their environmental impact while continuing to deliver on important economic, social, and governance targets.

Unfortunately there are no “one size fits all” tools when it comes to modelling of metallurgical processes, and a variety of different pieces of software must often be brought to bear on a particular problem. Apart from data formatting and translation issues, this can pose some interesting challenges in assembling workflows that take advantage of high performance computing (HPC) in appropriate parts of the overall solution.

In this presentation, we will discuss a simple application of this workflow approach in the assessment of electrical performance of large-scale direct-current (DC) plasma arc furnaces used for the production of ferroalloys. Magnetohydrodynamic multiphysics models of high current DC arcs give insight into the electrical behaviour of the furnace, but they require information about the thermophysical properties of the plasma fluid, and this in turn requires knowledge of the gas compositions in the furnace while it is operating. The HPC needs of each calculation step vary widely, and providing appropriate acceleration at each step is the key to obtaining an overall workflow that runs in reasonable times.

Speaker: Dr Quinn Reynolds (Mintek)

14:50 Q&A

Q&A

Speaker: Q&A

13:30 → 15:00 HPC Technology: Education BoF

Convener: Chair: Mr Bryan Johnston (CHPC)

13:30 Welcome and Introduction

Speaker: Bryan Johnston (CHPC)

13:35 Overview of Wits HPC

13:45 Overview of new local HPC course

13:55 Are You Doing Science?

Speaker: Jay Lofstead (Sandia National Laboratories)

14:35 ® Onboarding new first time HPC users

Speaker: John Poole

14:45 ® SIGHPC Committee

14:55 SIGHPC Edu

13:30 → 15:00 ISSA: 4

Convener: Mariana Gerber (Nelson Mandela University)

13:30 Smoke Signals: Analysing and Decoding Quiet Quitting and Burnout Among Cybersecurity Professionals

In recent years, the rising issue of burnout and the new phenomenon of quiet quitting have emerged as major concerns within the workforce, espe-cially among cybersecurity professionals. These phenomena not only undermine employee well-being, but also the security of information systems. Addressing the challenges of these phenomena requires an understanding of the un-underlying organisational factors that contribute to quiet quitting and burnout. This paper analyses the sentiments of cybersecurity professionals and identifies the organisational factors contributing to quiet quitting and burnout among cybersecurity professionals. A sentiment analysis, together with a thematic content analysis was conducted on comments from the subreddit “cybersecurity”, an online forum frequented by cybersecurity professionals. Through these analyses, five organisational factors were revealed that contribute to the onset of quiet quitting and burnout, namely: Work Overload, Poor Workplace Dynamics, Cybersecurity Skills Gap, High-Risk, High-Pressured Job and Continuous Upskilling. This paper therefore contributes to the ongoing discussion regarding burnout and quiet quitting amongst cybersecurity professionals and provides a foundation for future research in this area.

Speaker: Michael De Jager (Unit for Data Science and Computing, North-West University)

13:50 Profiling Prominent Ransomware Threat Actors

In the modern era, the understanding of ransomware should not be limited to its technical aspects. Still, it must also incorporate an understanding of the covert and malicious practices of the ransomware threat actors behind it. By drawing from the strategic wisdom of Sun Tsu, the necessity of understanding the motivations and strategies of one’s adversaries to better defend oneself has become a critical aspect within the field of cybersecurity. Therefore, a comprehensive behavioural profile containing aspects such as affiliations, behaviours, business tactics, and attack strategies must be formulated to know one’s enemy better. Applying a systematic approach to behavioural profiling will, in turn, enable the ability to deconstruct and identify the aspects that contribute to ransomware threat actors’ success. In turn, proactive cybersecurity strategies can be developed to mitigate the aftermath of a ransomware attack effectively. Thus, organisations can systematically counter threats by using insights from in-depth behavioural profiles to negotiate with these ransomware threat actors.

Speaker: Hendrik Tjaart Pelser (North West University)

14:10 Contextual Factors Influencing the Cybersecurity Commitment of Government Institutions in Developing Countries

The increase in cybersecurity incidents is a growing concern for governments worldwide, especially in developing countries. Government institutions are among the top targets of cyberattacks. To address cybersecurity issues, various tools and frameworks have been developed to assess the level of cybersecurity maturity and commitments. Despite the calls on governments to develop and implement cybersecurity measures, the commitment level of government institutions toward cybersecurity remains inadequate. With the increase in cybersecurity incidents in developing countries, cybersecurity commitment of government institutions is increasingly indispensable in combating cybercrime. The current study considered the contextual factors influencing the ybersecurity commitments of government institutions in developing countries using the Technology, Organisation and Environment (TOE) framework.The study employed a qualitative case approach of government institutions and agencies responsible for cybersecurity activities in Namibia. Through document reviews and semi-structured interviews with 11 participants from five overnment institutions and agencies, the study identified and explained contextual factors influencing the cybersecurity commitments of government institutions. Data was analysed using a thematic analysis technique and the NVivo software. The study found that contextual factors, such as underdeveloped information technology infrastructure, a lack of information technology resources, inadequate support from top management, financial resources, a lack of cybersecurity skills and competencies, a lack of cybersecurity legal frameworks, and perceived cyber threats and attacks, affect cybersecurity commitment of government institutions.

Speaker: Teofelus Tuyeni (University of Cape Town)

14:30 Trust Requirements and Mechanisms in Peer-to-Peer Energy Markets

Peer-to-peer (P2P) energy markets are emerging as a promising solution to address the challenges faced by traditional energy systems. However, the decentralised nature of these markets necessitates robust trust mechanisms to ensure secure and reliable energy transactions. This paper presents a comprehensive review of trust requirements and trust-building mechanisms in P2P energy markets. It explores the role of blockchain technology, zero-trust architecture, and reputation systems in establishing trust among market participants. It identifies several trust requirements, including security, privacy, transparency, fairness, and reputation. The study further highlights the limitations of existing works and proposes future research directions to enhance trust and security in P2P energy markets. By addressing these limitations, the full potential of P2P energy trading can be unlocked, contributing to a more sustainable and resilient energy future.

Speaker: Boitumelo Leotlela

14:50 Q&A

13:30 → 15:00 SA NREN: NREN Equipment

Convener: Chair: Ndzalama Ngobeni

13:30 High-Performance Packet Processing with Programmable Data Planes

As telecommunication networks grow increasingly complex and data transfer rates soar, traditional networking equipment struggles to provide the necessary flexibility, visibility, and performance. This presentation explores our recent efforts to leverage programmable data planes for high-performance packet processing in non-enterprise networks. Our approach enables the handling of custom packet headers and protocols that are not supported by off-the-shelf hardware. We detail the hardware platform and techniques employed to achieve line-rate processing at up to 12.8 Tbps, ensuring efficient, scalable, and flexible network operations. These innovations offer practical solutions to the challenges posed by modern high-speed networks, bridging the gap between software-defined networking and specialized hardware performance.

Speaker: Ruan de Clercq (Dendrite Cyber)

14:00 The way forward with Nokia

TBD

Speaker: Craig Robert Taylor

14:30 AI-Native Networking for Modern Networks

AI-Native Networking for Modern Networks

Speaker: Lunama Miankodila (Networking OEM)

15:00 → 15:30 Break

15:30 → 17:00 HPC Applications: Users BoF

Convener: Chair: Dr Werner Janse Van Rensburg (CHPC)

15:30 CHPC Users BoF

All CHPC Users, Principal Investigators and anyone interested in practical use of CHPC computational resources are invited to attend this informal Birds-of-a-Feather Session.

At the start overviews will be presented of the recent usage of the CHPC compute resources (HPC Cluster, GPU Cluster and Sebowa Cloud infrastructure), discussion of new resources to be made available to users and discussion of questions and topics from the audience.

The following presentations will be part of the session:

• Overview of Lengau Usage - Werner Janse van Rensburg
• Overview of Sebowa Cloud Infrastructure – Dorah Thobye, Nkwe Monama
• New HPC Cluster at CHPC – Dorah Thobye, Eric Mbele
• Q&A - All

The session provides an excellent opportunity to meet up in-person with CHPC employees and to meet and engage with colleagues benefiting from CHPC services.

Speaker: Dr Werner Janse Van Rensburg (CHPC)

16:30 Computational Chemistry and Material Science Special Interest Group (SIG)

The largest user community at the Centre for High Performance Computing (CHPC) are those that belong to the domains of computational chemistry and material science (CCMM). The CCMM special interest group (SIG) was established by several senior academic staff members in the computational chemistry and material science domains that make use of the CHPC resources, to have a forum where the various user issues of this community can be discussed.

The SIG session is held annually during the CHPC conference. It is important that those who are present at the conference attend the session so that we can discuss any problems which may be experienced by the users. The aim of the session is to try and come up with possible solutions for the user related problems and see if/how these solutions maybe implemented at the CHPC to assist users with a smoother high performance computing experience.

Speaker: Krishna Govender (University of Johannesburg)

Chem_Mat_SIG_2024.pdf

15:30 → 17:00 HPC Technology: Storage & IO

Conveners: Jay Lofstead (Sandia National Laboratories), Kevin Colville (CHPC)

15:30 Reducing Data Movement Through Derived Quantity Generation

Learning from scientific simulations often relies not on the raw quantities calculated, but instead from derived values. For example, air pressures in a weather simulation in isolation are not as interesting as the air pressure gradient. If the gradients are roughly parallel, you have straight-line winds. If the gradient is rotating around a point, it is a cyclone storm. Other uses, such as reducing from a complex multi-quantity 3D model into a 2D representation for the actual value required for analysis can radically reduce data movement from the large computational system to the local analysis system. We have built tools that enable automatically calculating these derived quantities as part of normal simulation output to enable extracting the derived quantities for movement to a different machine and other purposes.

Speaker: Dr Jay Lofstead* (Sandia National Laboratories)

15:50 Characterizing the I/O Performance of HPC Platforms Using the I/O Performance Evaluation Suite

The I/O Performance Evaluation Suite (IOPS), currently under development by the TADaaM team at INRIA Bordeaux, aims to streamline the performance analysis of I/O systems on HPC platforms. The primary objective of this tool is to automate the performance evaluation process, initially determining the optimal combination of parameters (such as the number of nodes, processes, file sizes, and access patterns) that enable the system to achieve peak I/O performance. Moreover, IOPS can also be used to examine how parameters of parallel file systems, such as the number of Object Storage Targets, affect system performance under distinct scenarios.

Speaker: Dr Luan Teylo* (INRIA Bordeaux)

16:10 ® I/O Insights and Recommendations for All

The complexity of the HPC I/O stack combined with gaps in the state-of-the-art profiling tools creates a barrier that does not help end-users and scientific application developers solve the I/O performance problems they encounter. To bridge this gap, we introduce Dristhi, a multi-source interactive analysis framework that empowers users to visualize I/O traces, identify bottlenecks, and gain a deeper understanding of application behavior. By combining cross-layer analysis with heuristic-based automatic detection, Dristhi provides actionable insights and recommendations to overcome common I/O performance bottlenecks. This talk will delve into the design and capabilities of Dristhi, demonstrate its use in pinpointing I/O performance issues, and highlight upcoming features that will further enhance its functionality.

Speaker: Jean Luca Bez (Lawrence Berkeley National Laboratory)

16:30 ® XIO: Towards eXplainable I/O using multi-layer monitoring solution for HPC Systems.

Modern HPC workloads exchange vast amounts of data to drive scientific discoveries While HPC systems employ diverse storage devices and tiers to support efficient data access, current monitoring infrastructures, such as Darshan and Score-P, only provide enough information to show what they see, but lack the visibility needed to fully explain observed I/O performance.

In this talk, I will present our latest survey of state-of-the-art monitoring tools deployed on modern HPC system using lists such as TOP500, Green500, IO500, and the Comprehensive Data Center List (CDCL). Then, we will introduce our latest efforts to tackle the opaque monitoring infrastructure that explores the user and kernel I/O stack to uncover causality relationships for achieving eXplainable I/O (XIO).

Speaker: Hari Hariharan Devarajan (LLNL)

16:50 Q&A

15:30 → 17:00 ISSA: 5

Convener: Chair: Reinhardt Botha (Noroff University College & Nelson Mandela University)

15:30 A Rapid Review Cybersecurity Strategic Analysis Framework: The Case of Higher Education Institutions

This paper conducts a rapid review using the adapted SVOT
(Strengths, Vulnerabilities, Opportunities, Threats) framework to synthesize current literature on cybersecurity in higher education institutions (HEIs). The review addresses (1) inherent strengths that enable HEIs to adopt robust cybersecurity measures, (2) specific vulnerabilities exposing HEIs to cyber threats, (3) opportunities to enhance cybersecurity practices, and (4) threats from sophisticated cyberattacks. Findings reveal that HEIs possess strengths like advanced technical resources and collaborative research cultures that could support innovative cybersecurity solutions. However, vulnerabilities can also arise from complex IT systems, open academic settings, and limited security funding. Opportunities include leveraging AI and machine learning for threat detection, implementing comprehensive security frameworks, and expanding cybersecurity education. Key threats involve ransomware, phishing, and state-sponsored espionage. This review underscores the need for HEIs to adopt a holistic, strategic cybersecurity approach, strengthen internal capabilities, and address administrative and cultural challenges to develop more adaptive defenses.

Speaker: Rennie Naidoo (University of the Witwatersrand)

15:50 Factors influencing Sensitive Data Protection practices in South African public schools

The current digital age has resulted in a surge in the use of Information and Communication Technology (ICT) tools that collect, store, and transmit huge volumes of sensitive data. Thus, sensitive data protection is a critical issue for all organisations in South Africa, including public schools. The problem is that schools often prioritize the benefits of using these ICT tools while neglecting the importance of protecting the substantial amounts of sensitive data produced, stored, and managed via these digital tools.  The purpose of this paper is to investigate the factors that influence sensitive data protection practices in South African public schools. A qualitative research strategy with semi-structured interviews was applied. Fifteen interviews were conducted among school administrative clerks, teachers, Department of Education finance clerks, and school social media managers. A thematic data analysis approach was used in collaboration with NVIVO to analyze the collected data.  The findings revealed both hindering and facilitating factors for sensitive data protection practices in South African public schools. Technological resources, awareness, and training do not hinder sensitive data protection practices in schools. On the other hand, organisational culture and attitudes hinder the practices. The findings revealed a conflicting landscape of compliance with the POPI Act and highlight the importance of using these factors to cultivate a culture of sensitive data protection practices in South African public schools.

Speaker: Siyavuya Ntlale (University of Cape Town (UCT))

16:10 Exploring the Cybersecurity-Resilience Gap: AnAnalysis of Student Attitudes and Behaviors in Higher Education

Abstract: Cyberattacks frequently target higher educational institutions, making cybersecurity awareness and resilience critical for students. However, limited research exists on cybersecurity awareness, attitudes, and resilience among students in higher education. This study addresses this gap using the Theory of Planned Behavior as a theoretical framework. A modified Human Aspects of Information Security Questionnaire was employed to gather 266 valid responses from undergraduate and postgraduate students at a South African higher education institution. Key dimensions of cybersecurity awareness and behavior, including password management, email usage, social media practices, and mobile device security, were assessed. A significant disparity in cybersecurity awareness and practices, with postgraduate students demonstrating superior performance across several dimensions was noted. This research postulates the existence of a Cybersecurity-Education Inflection Point during the transition to postgraduate studies, coined as the Cybersecurity-Resilience Gap. These concepts provide a foundation for developing targeted cybersecurity education initiatives in higher education, particularly highlighting the need for earlier intervention at the undergraduate level.

Speaker: Dirk Snyman (University of Cape Town)

16:30 Lacking Balance? Assessing the Content and Coverage of Cyber Security Degrees

Cyber security is now commonly encountered as a focal topic for ac- ademic degrees. However, the presence and level of representation of relevant sub-topics within such degrees can vary significantly, and as a consequence the resulting student experience and graduate perception of what cyber security is and what it involves can be similarly varied. This paper examines the situation, based upon a relevant sample of cyber security degree programmes from the UK, all of which share the common characteristic of being titled simply MSc Cyber Security. The review considers the level of coverage afforded to technical and non-technical aspects of cyber security, as well as any inclusion of non-cyber coverage within the programmes. The results reveal that candidates holding what is ostensibly the same degree (based on the title) can emerge with tangibly dif- ferent knowledge and skills, with significant variation in underlying topics cov- ered. Although this is not a problem in terms of the validity of the coverage, it can pose a issue for prospective students and employers in terms of differentiat- ing between degrees and understanding what they are offered as a consequence.

Speaker: Steven Furnell (University of Nottingham)

16:50 Q&A

15:30 → 16:55 SA NREN: SABEN, SANReN, TENET

Convener: Ntuthuko Sambo (CSIR)

15:30 South African Broadband Education Nertworks (SABEN)

South African Broadband Education Nertworks (SABEN)

Speaker: Acres Michael (SABEN)

15:50 Empowering Research and Education: An Introduction to TENET's Trust & Identity Services

Empowering Research and Education: An Introduction to TENET's Trust & Identity Services

Speaker: Wesley Barry (TENET South Africa)

16:10 The South African National Research Network (SANReN)

The South African National Research Network (SANReN)

Speaker: Ajay Makan (SANReN)

17:00 → 17:15 Intermission

17:15 → 19:15 Poster: Cocktails Poster Showcase

Cocktails Poster Session

Wednesday, 3 December

06:00 → 07:00 Morning Walk Wellness

Wellness and Social activity: meet at 6am in front of the City Lodge hotel. A walk for 1 hour. Hosted my Mervyn.

08:00 → 09:00 Registration

09:00 → 10:30 Keynote: Wednesday

Convener: Chair: Dr More Manda (CSIR)

09:00 Intel AI in Action at the 2024 Olympics: Deployments and Benefits for your Business

Intel has already put a lot of “AI in action”. Come and hear about some of the use cases that were deployed at scale during the Paris 2024 Olympic and Paralympic Games. You will be blown away with the capabilities and their results ! And the session will also deliver some details about the technologies and features of the Intel ingredients that are in these use case solutions, as well as look at the new Intel AI ingredients and solutions.

Speaker: Dr Jean-Laurent Philippe* (Intel)

09:45 Tackling the Challenges of Reproducibility and Data Sharing in Computational Science

Research and discovery are increasingly computation, data-intensive, interdisciplinary, and collaborative. However, reproducing results remains a significant challenge. Scholarly publications are often disconnected from the data and software that produced the results, making reproducibility difficult. Researchers today generate vast amounts of data, code, and software tools that need to be shared, but sharing data remains challenging, especially when data is large or sensitive. Moreover, funding agencies are increasingly requiring sharing data used to generate results, yet data is only valuable if it is reproducible. A key challenge is that reproducible artifacts are typically created only after the research is complete, hindered by a lack of standards and insufficient motivation. Despite growing recognition of the importance of reproducibility, the research community still lacks comprehensive tools and platforms to support reproducible practices throughout the research cycle, as well as a culture that educates and trains researchers on the topic.

This presentation will introduce SHARED (Secure Hub for Access, Reliability, and Exchange of Data), a new initiative at the University of Chicago to develop a comprehensive platform for data-driven research and data management. We will discuss the challenges and opportunities of reproducibility in computational research and strategies for capturing reproducible artifacts throughout the research process. Additionally, we will share progress on building a community of practice to democratize reproducibility in scientific research.

Speaker: Prof. Hakizumwami Birali Runesha* (University of Chicago)

10:30 → 11:00 Break

11:00 → 12:30 DIRISA: 1

Convener: Chair: Prof Nobert Jere (University of Fort Hare)

11:00 SAWS Climate Services and Infrastructure

The South African Weather Service (SAWS) is a key part of the country's weather system. SAWS runs a complex system of tools to observe weather. This includes manual weather stations, automatic rain measuring stations, automatic weather stations, weather radars, a network that measures sunlight, and a system to detect lightning. These systems give important information right away for predicting the weather and studying the climate. SAWS uses advanced computer models to create detailed weather and ocean forecasts. These models are important for predicting extreme weather events like cyclones and coastal storms, which can seriously affect communities and businesses. The SAWS climate infrastructure is a changing system that supports the country's work to fight against and adjust to climate change. South Africa is improving its ability to watch for, predict, and react to climate-related problems by using new technologies, strong research efforts, and working with internal and international communities.

Speaker: Dawn Mahlobo* (South African Weather Service)

11:20 Resource Scheduling and Allocation in HPC Cyberinfrastructure

Abstract
High-Performance Computing (HPC) systems play a pivotal role in modern scientific research, enabling complex simulations, data analysis, and large-scale modelling across disciplines such as climate science, genomics, physics, and engineering. As these systems grow in scale and sophistication, the efficient scheduling and allocation of computational resources become crucial for ensuring optimal system performance, maximising resource utilisation, and meeting the needs of diverse user communities. In HPC environments, resource scheduling and allocation determine how tasks are assigned to hardware resources such as CPUs, GPUs, memory, storage, and network bandwidth. Effective scheduling strategies are critical for maintaining fairness among users, optimising job throughput, reducing waiting times, and enhancing energy efficiency.
Cyberinfrastructure, the integration of advanced computing platforms with large-scale data storage and high-speed networks, addlayers of complexity to resource management. The heterogeneity of hardware, dynamic workload demands, and multi-user environments require advanced resource scheduling algorithms. Traditional approaches like First-Come, First-Served (FCFS), Shortest Job First (SJF), and Backfilling have evolved to meet these challenges, while more advanced strategies like Priority Scheduling, Gang Scheduling, and Hybrid Scheduling offer increased flexibility and efficiency. Energy-aware scheduling has also gained importance, given the significant portion of operational costs that energy consumption in large HPC systems can account for.

Speaker: Dr Olalekan Samuel Ogunleye* (University of Mpumalanga)

Prof_Ntombovuyo.Wayi-Mgwebi_Abstract.docx

11:40 Data driven decision-making and policy for application in water and sanitation systems

This project looks at the use of data to develop evidence-based decision-making and policy for application in water and sanitation systems. It includes a review and understanding of local and global databases and resources, developing data mining and management standards and practices, and applying data analysis techniques to advance water and sanitation systems in South Africa.

Data mining and management plays an important role in advancing water and sanitation systems, ensuring sufficient monitoring and the sustainable delivery of essential services. The application of data mining and management encompasses the collection, processing, and storage of vast datasets derived from various sources such as local and global databases, dashboards, sensor networks, satellite imagery, and public health records.

Data analysis and modelling techniques then facilitate the identification of patterns, trends, and anomalies, which are important for informed decision-making, strategic planning and public policy. By leveraging predictive analytics and forecasting, government departments and water management authorities can anticipate demand fluctuations, optimise resource allocation, and enhance the efficiency of water distribution networks. Similarly, in water sanitation, data mining and analysis assists in monitoring system performance, detecting potential failures, and mitigating health risks by providing early warnings of contamination events.

The adoption of robust data management frameworks ensures the integration, storage, and accessibility of diverse datasets, supporting real-time monitoring and long-term strategic initiatives. Challenges such as data privacy, accuracy, and the need for interdisciplinary collaboration need to be addressed to ensure the reliability and efficacy of these systems. The convergence of data mining and management in water and sanitation sectors holds significant promise for enhancing operational efficiency, ensuring resource sustainability, and safeguarding public health.

Data integration poses a significant hurdle due to varying formats and structures across different sources. The primary challenge is data quality and accuracy including issues like missing values and statistical outliers. Water databases may contain a diverse range of data, including spatial, temporal, and multi-dimensional information. Integrating and reconciling these different types of data can be challenging, especially when they come from various sources with different formats and structures.

Speaker: Ridhwaan Suliman* (CSIR)

12:00 DIRISA: A National Infrastructure for Research Data Management and Collaboration

As part of the global drive for cyberinfrastructure providers to continue enabling, enhancing and empowering research data management through collaborative frameworks, the Data Intensive Research Initiative of South Africa (DIRISA), as a national research data initiative, provides an integrated suite of free tools and services designed to optimize research data workflows, ultimately amplifying the impact of research and scholarly endeavours.

This presentation will showcase DIRISA's offerings and tools, which include cloud storage, ensuring that researchers can safely store and manage their data, research data management planning, seamless data transfers and ways to enhance the discoverability and citation of datasets, fostering greater visibility in the academic and research landscape.

In the presentation attendees will explore how DIRISA can serve as a vital partner in their research journey, paving the way for innovative outcomes and collaborative success, which also accelerates research impact across multiple disciplines.

Speaker: Dr Katlego W. Phoshoko* (Data Intensive Research Initiative of South Africa (DIRISA), CSIR)

12:20 Q&A

11:00 → 14:00 DIRISA: Student Datathon Presentations

11:00 → 12:30 HPC Applications

Convener: Chair: Dr Ndanduleni Lethole (University of Fort Hare)

11:00 HPC Applications Enabling Catalysis Design for Ammonia Decomposition to Produce Hydrogen

The envisaged hydrogen economy cannot be realized without a proactive approach to novel materials design. Hydrogen is emerging as a game changer in the grand realm of clean energy value chain mainly because, as an energy vector, hydrogen is both an energy carrier and an energy source. However, despite this potential, several challenges limit large scale adoption of hydrogen including the fact that hydrogen is explosive and flammable, making it difficult to store and transport. One way of mitigating this bottleneck is chemical storage in molecular carriers. On this, liquid organic hydrogen carriers have gained dominance. However, they have not reached the level of technological maturity for large scale development and adoption. Molecules such as ammonia are likely to be good alternatives since their value chain is technologically mature and they can be used as feedstock in hydrogen production. Hydrogen is produced from ammonia mainly through catalytic decomposition. The rational design of novel catalysts for ammonia dehydrogenation is not feasible via —trial and error— experimental investigations and is also computationally intensive. Hence the need for high-performance computing (HPC). With HPC coupled with density functional theory models, it is feasible for one to high-throughput screen of different catalytic processes involved in hydrogen production via catalytic decomposition, compared to conventional laboratory-based catalyst synthesis and optimization which are known to be costly, time-consuming, and wasteful in terms of materials.

Speaker: Ms Nomcebo Motsa* (Sustainable Energy and Environment Research Group (SEERG), Department of Chemical Engineering, Faculty of Engineering, Built Environment and Information Technology, University of Pretoria, Hatfield 0028, Pretoria, South Africa.)

Abstract_final_CHPC__2024-10-23.pdf

11:20 Computational Insights into Wind Patterns and Albatross Mortality

As technology advances, field researchers increasingly apply sophisticated computational methods to study biological systems, enabling insights across individual, population, and ecosystem scales that were previously challenging to obtain through traditional observational approaches. The present study focused on the impact of wind patterns on sub-Antarctic Marion Island’s ecology, combining extensive field observations, wind tunnel experiments, and numerous computational fluid dynamics (CFD) simulations (using both ANSYS Fluent and Siemens Star-CCM+). The initial phase of the project involved simulating wind flow across Marion Island (ca. 330 km²) and the adjacent Prince Edward Island. This was a time-consuming task due to the scale and complexities of modelling atmospheric flow, even with access to the resources at the CHPC. The resulting wind data have informed various ecological studies on Marion Island, particularly in phase two of this research, which focused on understanding grey-headed albatross crash-landings around an inland breeding site in relation to local wind patterns.

Field observations of crash sites on Marion Island showed that most grey-headed albatross crashes occurred during departure flights, where low altitude and variable wind vectors pose risks. To complete an aerodynamic investigation a wing geometry was constructed using a combination of wind tunnel testing, 3D scanning and photographs of grey-headed albatrosses in flight. The extensive aerodynamic investigation was completed using CFD analyses of the generated geometry in gliding flight, incorporating over 350 simulations with a fine 3D mesh. The simulated aerodynamic loads showed that these birds can generate lift up to nine times their weight at high airspeeds and positive angles of attack. Under prevailing westerly winds near the inland breeding site, high cross-winds combined with moderate downdrafts create downforces (negative lift) comparable to the bird’s weight, leading to high-impact crashes that are often fatal.

Given the albatross’s low natural mortality rate, the current level of wind-related deaths is concerning. This study provides rare insight into how on-land wind patterns affect seabirds and highlights the need for ongoing monitoring of changing conditions.

Speaker: Ms Janine Schoombie* (University of Pretoria)

11:40 Acceleration of Electronic Structure Codes on Heterogeneous Hardware

We describe the development of a new library, libGint, for the calculation of the four-centre, two electron integrals required by local basis set ab initio electronic structure codes such as CP2K and CRYSTAL. The focus of this new work is the acceleration of the calculation via Graphical Processing Units (GPUs), particularly for codes that employ periodic boundary conditions. The core kernels for general contracted gaussian basis set have been rewritten allowing efficient calculation on a GPU. Initial integration into CP2K has been completed and correctness demonstrated. The current challenges involve optimizing the existing code, particularly batching the integrals so they can be calculated effectively on one or more GPUs, and development of a flexible application programming interface.

Speaker: Dr Marcello Puligheddu (Science and Technology Facilities Council)

MP_IB_CHPC24.pdf

12:00 Rational design of porphyin and BODIPY analogues for biomedical applications: further progress and future perspectives

Over the last two years, considerable further progress has been made in using a rational design approach [1,2] guided by calculations with the Gaussian 09 software package on the Lengau cluster and an application of Michl’s perimeter model [3] to prepare novel main group element complexes of porphyrin and boron dipyrromethene (BODIPY) analogues that are suitable for use as photosensitizer dyes in photodynamic therapy against cancer and bacteria [4-8] and as optical limiters in applications relevant to the protection of human eyesight from intense incident laser beams [9-10]. There has been a strong focus on exploring how the lowest energy porphyrin absorption band can be substantially red-shifted into the phototherapeutic window by introducing reduced and confused pyrrole moieties [2,4-7]. Future directions on the use of the Gaussian 09 software package in the context of this research will be described.
References
[1] J. Mack, Chem. Rev. 2017, 117, 3444-3478.
[2] B. Babu, J. Mack, T. Nyokong, Dalton Trans. 2023, 52, 5000-5018.
[3] J. Michl, Tetrahedron 1984, 40, 3845-3934.
[4] R. C. Soy, B. Babu, J. Mack, T. Nyokong, Molecules 2023, 28, 4030.
[5] S. Dingiswayo, K. Burgess, B. Babu, J. Mack, T. Nyokong, Photochem 2023, 3, 313-326.
[6] R. C. Soy, B. Babu, J. Mack, T. Nyokong, Photodiagn. Photodyn. Ther. 2023, 44, 103815.
[7] A. K. May, B. P. Ngoy, J. Mack, T. Nyokong, J. Porphyrins Phthalocyanines 2024, 28, 88-96.
[8] R. C. Soy, D. Mafukidze, J. Mack, T. Nyokong, Eur. J. Inorg. Chem. 2024, 27, e202400072.
[9] A. K. May, J. Mack, T. Nyokong, J. Porphyrins Phthalocyanines 2023, 27, 591-599.
[10] G. Kubheka, J. Mack, T. Nyokong, J. Porphyrins Phthalocyanines 2024, 28, 61-71.

Speaker: Prof. John Mack (Rhodes University)

12:20 Q&A

Q&A

Speaker: Q&A

11:00 → 12:30 HPC Technology

Convener: Chair: Dr Oz Parchment (University of Cambridge)

11:00 Empowering Southern Africa's Researchers: Overcoming HPC Barriers with MATLAB Tools

As scientific data gets bigger and its analyses involves increasingly complex algorithms, access to High Performance Computing (HPC) resources in the form of multi-core CPUs and GPUs become essential for scientific research.

One main impediment to harnessing these resources remains the high barrier to entry – HPC systems often require a deep understanding of computing architecture and command line interfaces, a skill not many researchers possess.

The approach to parallel computing within MATLAB is to abstract away a lot of this complexity from the end-user, so that they can progress seamlessly from their prototype on the desktop onto the cloud or the cluster. The power of the Parallel Computing Toolbox and MATLAB Parallel Server is free and available already to more than 20+ Universities and Institutions in Southern Africa, like CSIR and many others. In this presentation, I will show how MathWorks tools can streamline research computational effort with three ways to make parallel programming more accessible

1. the ability to scale from local machine to cluster using minimal changes to the code,
2. the ability to submit jobs remotely to HPC clusters from within the users’ local MATLAB environment using MATLAB Parallel Server,
3. free training workshop being held globally in partnership with HPC centers to upskill scientists and researchers in parallel programming.

Speaker: Dr Marco Rossi (MathWorks)

11:40 Sustainable AI Infrastructure: A Mandatory Pathway for Supercomputing Centres

As artificial intelligence drives more and more innovation in industry and science with the emergence of Large-Language Models (LLMs), the demand for such computing resources is growing substantially. This development poses major challenges for supercomputing centres around the world, as they need to provide state-of-the-art AI resources while meeting their sustainability goals and mandates.
While today's world-leading supercomputer according to the Top500 list require about 23 MWatt (Frontier) to provide computing power to the scientific community, we see an exponential increase in power consumption for AI training and inference. Estimates for training and interference of current LLMs are huge, and newer models will require even more power, representing an unnoticed increase in energy consumption.
For the operational requirements of HPC and AI supercomputers, we need to evaluate the total power consumption and thus the total energy based on the Power Usage Effectiveness (PUE). A four-pillar model was developed at the Leibniz Supercomputing Centre (LRZ) to implement a holistic optimisation strategy for energy efficiency, which includes the building infrastructure, system hardware and software as well as the actual applications. By extending this to AI hardware and cooling, optimising AI models and implementing AI-driven resource management, supercomputing facilities can meet these requirements while offsetting their environmental footprint.

Speaker: Maximilian Höb (Leibniz Supercomputing Centre (LRZ))

12:30 → 13:30 Lunch

12:30 → 13:30 Lunch

12:30 → 13:30 Lunch

13:30 → 15:00 DIRISA: 2

Convener: Chair: Prof Nobert Jere (University of Fort Hare)

13:30 Data sharing, fabric and effective governance in the error of big data analytics

In the era of big data, organizations are increasingly leveraging advanced data management strategies to enhance operational efficiency and drive innovation. This paper explores three critical components of modern data management: data sharing, data fabric, and data governance.
Data sharing facilitates seamless access and exchange of data across different departments and external partners, fostering collaboration and informed decision-making. However, it also raises concerns about data security and privacy, necessitating robust mechanisms to ensure data integrity and compliance.
Data fabric represents an architectural approach that integrates various data sources, both structured and unstructured, into a unified, intelligent data management framework. This approach enhances data accessibility, quality, and real-time analytics, enabling organizations to derive actionable insights from their data assets.
Data governance encompasses the policies, procedures, and standards that ensure data is managed effectively and responsibly. It addresses issues related to data quality, privacy, and compliance, ensuring that data is accurate, secure, and used ethically. Effective data governance is crucial for maintaining trust and accountability in data-driven environments.
Together, these elements form the backbone of a resilient data strategy, empowering organizations to harness the full potential of their data while mitigating risks associated with data misuse and regulatory non-compliance

Speaker: Mawande Mongo* (Altron)

13:50 Applications of 4IR on the diagnosis and management of STIs among key populations in Sub-Saharan Africa: A Systematic Review

Introduction: The Fourth Industrial Revolution (4IR) is trending because of the major transformations it has brought to human life. Artificial intelligence including machine learning are 4IR technologies that can generate intelligent machines that can be used for the diagnosis and management of HIV and associated sexually transmitted infections. Key populations are disproportionately affected by HIV and STIs due to specific risk behaviors, marginalization, and structural factors that contribute to a lack of access to health services. The 4IR technologies are used in reporting the key populations’ STI vulnerability, transmission, and treatment.
Aim: To explore the use of 4IR technologies in the diagnosis and management of STIs for key populations in Sub-Saharan Africa.
Methods: A review of the literature published from 2015 onwards was done. Manual and electronic searches on various databases including PubMed Central, SCOPUS, and Science Direct were conducted. The Preferred Reporting Items for Systematic Reviews and meta-analysis statements for protocol guidelines were followed, and the review is registered in the International Prospective Register of Systematic Reviews database (Ludwig-Walz, Dannheim, Pfadenhauer, Fegert & Bujard, 2023). PROSPERO Registration ID is CRD42023468734.
Results: Different machine learning algorithms including random forest classifier, support vector machine, and logistic regression can be used to generate models to predict STIs. The 4IR technologies can help to track people who have accessed STI services including those who have the potential to transmit infections including prevention and care for the sake of enhancing patient outcomes.
Conclusion: Machine learning models can help identify individuals at high risk of contracting HIV and assist policymakers in developing targeted HIV prevention and screening strategies informed by socio-demographic and risk behavioural data. There remains a gap in HIV diagnosis for key populations. The 4IR technologies can use available data for building models on HIV diagnosis and care among key populations in Sub-Saharan Africa and significantly improve elements required to facilitate diagnostic and management approaches.

Speaker: Dr Claris Siyamayambo* (1South Africa Medical Research Council/ University of Johannesburg (SAMRC/UJ) Pan African Centre for Epidemics Research (PACER) Extramural Unit, Faculty of Health Sciences, South Africa)

Claris_Annual_CHPC_Conference_DIRISA_Track_2024_Abstract.docx

14:10 Crafting Open Data for Open Science: Technical Innovation and Data Management in Environmental Research — The SAEON Open Data Platform

Crafting Open Data for Open Science: Technical Innovation and Data
Management in Environmental Research — The SAEON Open Data Platform

The South African Environmental Observation Network (SAEON) is one of the National
Research Foundation (NRF)’s Research Infrastructure Platforms and serves as a sustained, coordinated, responsive and comprehensive South African earth observation network. SAEON delivers long-term, reliable data for scientific research and informs decision-making to support a knowledge society and improve quality of life.

The Open Data Platform (ODP) is one of SAEON’s research data infrastructure comprising an aggregation of databases, services and web applications that facilitate the preservation, publication, discovery, and dissemination of earth observation and environmental data in South Africa. The ODP was certified as a trustworthy data repository by CoreTrustSeal in 2023. In the evolving landscape of environmental data, which continuously changes in response to technological, societal, and scientific developments, effective curation and publication processes are crucial for ensuring dataset accessibility and usability. SAEON’s commitment to the FAIR principles — making data Findable, Accessible, Interoperable, and Reusable — guides data management and dissemination practices. SAEON ensures that metadata is comprehensive and adheres to established standards, enabling users to understand the context and quality of the data.

While the assemblage of systems constituting the ODP has grown organically over many years, an information architecture has simultaneously evolved to enable the centralised metadata management system to accept data submissions from a variety of sources. It permits quality control and value-added annotations by SAEON’s data curation team. Additionally, it allows for interoperability with and publication of metadata to a variety of data cataloguing systems both locally and globally. The development of this abstract information architecture has led to the emergence of useful high-level patterns including many-to-many connectivity between data producers (archives) and data consumers (catalogues), differentiated access control supportive of multitenancy, and an extensible ontological framework.

In this presentation, we will cover:

• SAEON’s approach to managing environmental data publication, including rigorous curation process, quality checks, and the assignment of DOIs;
• The role of comprehensive metadata in enhancing discoverability and user engagement;
• Characteristics of the ODP information architecture and its utility in real-world use cases.

Speaker: Mark Jacobson* (South African Environmental Observation Network)

Mark_Jacobson_Abstract.pdf

14:30 A Secure Data-Centric Model for SMMEs Across Africa: Bringing Stakeholders Together

Cybersecurity continues to be a threat to many sectors and individuals within Africa. As a result, Small, Medium, and Micro Enterprises (SMMEs) are also affected. SMMEs face numerous challenges related to data security, particularly as they increasingly rely on digital tools and platforms for their operations. The situation is worse for SMMEs, particularly those in rural or underserved areas. There have been differently initiatives currently in palace to alert the African community on cybersecurity. However, implementing a Secure Data-Centric Model for SMMEs in Africa comes with several challenges due to the unique socio-economic and technological landscape of the continent. Additionally, researchers seem to have noticed that cybersecurity initiatives are mainly dominated by those in the Information Technology space, which seems not to be enough. To overcome these challenges, African SMMEs may need support from governments, industry bodies, and international organizations in the form of funding, training, and access to affordable. Collaborative efforts to raise awareness, improve regulatory frameworks, and build local cybersecurity capacity are also crucial in helping SMMEs in Africa implement effective secure data-centric models. Despite growing accessibility, the cost of advanced cybersecurity solutions can be prohibitive for many SMMEs.
A systematic review on current cybersecurity data and interviews with cybersecurity experts was done. This was supported by the recently completed Pan African Information Communication Technology Association Cybersecurity Conference held in August 2024. The data from the presenters and may points raised during the conference were noted and considered for this talk. The main research question for this talk is:
How can secure data-centric model for SMMEs across Africa be designed through stakeholders engagements?
Results are clear on the specific stakeholders that are required, and the key components of the data centric model components are well documented. As an example, collaboration among these stakeholders—government, industry, academia, and civil society—is essential to create an enabling environment where secure data-centric practices can flourish among SMMEs across the continent. The Secure Data-Centric Model for SMMEs is designed to address these challenges by creating a robust framework that prioritizes data security while fostering collaboration among key stakeholders. We argue that the Secure Data-Centric Model for SMMEs is a comprehensive approach to ensuring data security while fostering collaboration among key stakeholders. By focusing on secure data practices, stakeholder engagement, and regulatory compliance, the model empowers SMMEs to thrive in the digital economy while protecting their most asset i.e. data. In summary, a Secure Data-Centric Model for SMMEs is about creating a comprehensive, scalable, and collaborative approach to data security, ensuring that small businesses can protect their valuable information.

Speakers: Mr Sonwabo Mdwaba* (Pan African Information Communication Technology Association), Prof. Nobert Jere (University of Fort Hare)

CHPC_Abstract_Mdwaba__Jere_2024.pdf

14:50 Q&A

13:30 → 15:00 HPC Applications

Convener: Chair: Ms Nomcebo Motsa (Sustainable Energy and Environment Research Group (SEERG), Department of Chemical Engineering, Faculty of Engineering, Built Environment and Information Technology, University of Pretoria, Hatfield 0028, Pretoria, South Africa.)

13:30 Exploration of future magnetic and cathode materials through computer-based simulations.

The field of materials science has seen a significant increase in the use of computational techniques to study the properties and behaviour of different materials at the atomic and molecular level. A key computational method employed in the study of solid-state materials is Density Functional Theory (DFT), which accurately forecasts the electronic structure, mechanical properties, and thermodynamic stability of materials without the necessity for intricate experimental procedures. At the Department of Physics, University of Fort Hare, our focus lies on the simulation of magnetic and cathode materials for potential applications in data storage and rechargeable batteries. Our research utilizes a first-principles approach within the Density Functional Theory (DFT) framework to explore the diverse properties of transition 3d elements and platinum (M-Pt) alloys, as well as M2O4-based (M: Mn and V) cathodes, highlighting their potential for future technological advancements. All calculations are conducted using the CASTEP simulation code integrated into the Materials Studio software and the HPC platform offered by the Centre for High Performance Computing. Our investigations have revealed that alloys like FePt, MnPt and CoPt exhibit excellent properties such as high magnetic moments and magnetocrystalline anisotropies rendering them suitable for magnetic spintronic devices and magnetic data recording applications. Furthermore, polymorphs of CaMn2O4 have been predicted to possess exceptional thermodynamic, mechanical and dynamical stability, good electronic conductivity and intercalation potentials that fall within the prescribed pulse discharge/charge range of 2.7 V to 3.9 V. These properties ensure retention of structural integrity over multiple charge/discharge cycles, good electrochemical performance and reversibility of Ca during charge/discharge process.

Speaker: Dr Ndanduleni Lethole* (University of Fort Hare)

13:50 Architectural Variations of Li2MnO3-Li0.69MnO2 Core-Shell Cathodes During the Cycling Process

The core-shell architecture has garnered significant interest for electrode materials owing to the ability to enhance conductivity, stability and enhanced surface functionality. Due to its high capacity and energy density, the O3-type Li2MnO3 layered cathode material is a potential candidate electrode for large-scale energy storage. However, it tends to undergo structural transformation from layered to spinel configuration during charge cycles due to irreversible oxygen loss. Recent advances in surface coating techniques have improved the electrochemical performance of these cathodes by enhancing conductivity, stabilizing structures, and preventing harmful reactions with the electrolyte. One of the setbacks with this strategy is the delamination of the core from the rigid shell attributed to the radial tensile radial stress on the interface of the core shell. This compromises the mechanical integrity of the core-shell structure and affects the electrochemical performance of LIBs. In this study, Li2MnO3 is coated with Li0.69MnO2 for the first time, a layered material known for its stability (no phase transformation) and high ionic conductivity. Molecular dynamics simulations, using the DL_POLY code, were employed to examine the cycling performance of the Li2MnO3-Li0.69MnO2 core-shell system by delithiated from Li2MnO3 to LiMnO3 and to closely monitoring the potential risk of delamination under various temperature conditions. The simulations were carried out using the Nose-Hoover thermostat under the NVT ensemble with temperatures ranging from 300 K to 1500 K. The structural snapshots obtained indicated a fluctuating pattern in the system's structural stability at different temperatures. Notably, at certain temperatures, the core-shell system lost a significant number of atoms from both the core and shell, while at other temperatures, the system regained order with minimal atom loss. Additionally, the lithium diffusion coefficients varied with lithium concentration, showing higher diffusion values for Li1.3MnO3 and Li1.7MnO3 at lower temperatures and Li1.4MnO3 demonstrated highest diffusion rates of 1.25 nm/s2 at 1500 K whilst the highest attainable diffusion was that of Li1.7MnO3, 3.02 nm/s2 at 1200 K. Overall, temperatures where greater atom loss occurred show high diffusion though this trend was not consistent across all concentrations. The simulations were conducted using 48 cores and 5 nodes.

Speaker: Prof. Raesibe Ledwaba (University of Limpopo)

14:10 Computational challenges of running kilometer-scale Earth System Models in a developing country

Extreme weather events' escalating frequency and severity underscore the urgent need for high-resolution climate modelling. Developing countries, often with limited resources, face unique challenges in implementing kilometre-scale Earth System Models (ESMs) to simulate local-scale climate projections accurately. Here, I will explore the computational challenges associated with running such models and the socio-economic implications and opportunities. I will also present a case study of the Conformal Cubic Atmospheric Model (CCAM), a component of the first African ESM, at the Southern African Centre for High-Performance Computing (CHPC). By examining CCAM's performance and identifying computational bottlenecks, we aim to inform strategies for optimizing ESMs in resource-constrained environments.

Speaker: Mr Tumelo Moalusi (Witwatersrand University)

14:30 Developing a Sustainable and Performant GPU Implementation of Plane-Wave DFT

We present the GPU port of CASTEP (www.castep.org), which is scheduled for a
full public release in early 2025 and developed as part of the PAX-HPC project under the UK’s ExCALIBUR exascale-readiness programme. CASTEP is a well-established density functional material modelling software suite, consisting of 750k lines of modern Fortran with extensive support for distributed memory parallelism via MPI. The application profile is relatively flat, meaning that the run-time is not always dominated by a specific set of computational operations, and workloads per kernel are often insufficient to achieve device saturation easily.

We describe how we meet the challenges of porting the code while adhering to
our development principles of portability, sustainabilty, and efficiency. We discuss the benefits and limitations of using OpenACC/MP for device offloading, how we integrate these approaches with MPI, and illustrate performance on UK HPC.

Speaker: Mr Matthew Smith (University of York)

14:50 Q&A

Speaker: Q&A

13:30 → 15:00 HPC Technology: HPC Ecosystems BoF

Convener: Chair: Mr Bryan Johnston (CHPC)

13:30 HPC Ecosystems BoF

This session is an opportunity for members of the HPC Ecosystems Community and those that identify as associates / affiliates to convene in-person. The session will allow for members to discuss matters relating to HPC Ecosystems Project as well as broader African HPC and emerging HPC community topics. The 90-minute session will include 60-minutes of prepared talks from members of the community, followed by a further 30-minutes of open time for discussion and meaningful community engagement. Alas, muffins are not guaranteed.

Provisional talks include:

• HPC Ecosystems Project review for 2024
• University of the Witwatersrand - Open OnDemand regional Lab
• HPC Ecosystems monitoring toolchain
• Wits SIG HPC with Ecosystems
• A curriculum for SysAdmin training
• Feedback on an African HPC discovery survey
• HPC Ecosystems Project plans for 2025+

Speaker: Coordinator: Mr Bryan Johnston (CHPC)

15:00 → 15:30 Break

15:30 → 16:50 DIRISA: 3

Convener: Chair: Prof Nobert Jere (University of Fort Hare)

15:30 AcousNomaly: Learning to Detect Anomalies in Acoustic Telemetry Data using Machine learning and Deep Learning

Acoustic telemetry data plays a vital role in understanding the be-
haviour and movement of aquatic animals. However, these datasets,
which can often consist of millions of individual data points, often
contain anomalous detections that can pose challenges in data analysis
and interpretation. Anomalies in acoustic telemetry data can occur due
to various biological and environmental factors, and technological limi-
tations. Anomalous movements are generally identified manually, which
can be extremely time-consuming in large datasets. As such, this study
focuses on automating the process of anomaly detection in telemetry
datasets using machine learning (ML) and artificial intelligence (AI)
models. Fifty dusky kob (Argyrosomus japonicus) were surgically fit-
ted with unique coded acoustic transmitters in the Breede Estuary,
South Africa, and their movements were monitored using an array of
16 acoustic receivers deployed throughout the estuary between 2016
and 2021, resulting in more than 3 million individual data points. The
research approach combined the use of Neural Network (NN) models
and autoencoders to construct an efficient anomaly detection system. The model is proficient at learning the normal movement patterns within
the data, effectively distinguishing between normal and anomalous be-
haviour, and exceeding 90% across all four evaluation metrics including
accuracy, precision, recall, and F1. However, it may encounter chal-
lenges in accurately detecting anomalies where they deviate slowly from
the expected movement patterns. Despite this limitation, the model
demonstrates promising capabilities by pinpointing the precise loca-
tions of anomalous entries within the dataset. Further investigation,
including refinement and optimization of the model’s parameters and
training process, especially with memory-based NN-AE, may enhance
its ability to detect anomalies with greater accuracy and reliability.

Speaker: Siphendulwe Zaza* (MSc Applied Mathematics student from Rhodes University)

15:50 Empowering Data-Driven Research with AI Tools

Background
The growing volume and complexity of data, particularly in health and social research, present significant challenges, particularly in terms of data security and access to secure datasets. These issues are compounded when working with vulnerable populations, exposing data to potential cybersecurity risks. AI-powered tools like Chisquares are addressing these challenges by embedding advanced security features—such as encryption, data protection compliance, and secure storage—ensuring data safety while promoting inclusivity and accessibility for researchers in high-risk environments.

Methods
This study examines how AI-powered tools, with Chisquares as a case study, enhance data security and streamline research workflows. It highlights advanced security measures, including encryption, compliance with global data protection regulations, and flexible data storage options. We also demonstrate how these tools reduce the technical burden on researchers while maintaining data integrity and privacy.

Results
AI-powered platforms like Chisquares enhance research workflows by providing secure data access and storage through features such as encryption, role-based access controls, and data protection compliance. These platforms also support offline functionality, ensuring secure data handling in areas with limited internet access. Furthermore, over 80% of tasks, from data cleaning to manuscript preparation, are automated, with built-in safeguards for accuracy and security. By integrating data management into a single platform, these tools mitigate the risks associated with transferring data across multiple systems.

Conclusion
AI-powered tools like Chisquares are transforming research by embedding security throughout the process. They provide secure data access, offline functionality, and flexible data storage options, empowering researchers to handle sensitive information with confidence. These tools not only protect data but also enable impactful, data-driven research, particularly in resource-constrained and high-risk settings, ultimately supporting better public health outcomes.

Speaker: Lungile Nkosi* (University of Pretoria)

16:10 Obtaining and Using Data for Decision Support and Innovation in Resource-Constrained Environments

We live in an increasingly data-driven world, and as academics, researchers and professionals, we need the ability to manage, analyze, and interpret data efficiently. We also need to know how to source data and get around the various red-tape systems that are even more prevalent in resource-limited contexts, such as in Africa. This presentation reflects on over a decade of research and practical experience in harnessing data for decision-making in education, healthcare, and public administration, particularly within the context of Botswana.

One key area of focus has been optimizing data management processes to enhance operational efficiency and accuracy. My work on Data Matching has tackled a number of practical problems, including a project which showcases the potential of intelligent data matching techniques to streamline administrative tasks, prevent errors, and ultimately improve the quality of educational outcomes, by matching student registration records with exam scripts.

In healthcare, data-driven decision support systems have proven invaluable. My work on expert systems for HIV and AIDS Information and development of Decision Support for Provision of HIV Treatments has demonstrated the role of expert systems and artificial intelligence in improving treatment outcomes for complex diseases such as HIV/AIDS. This was extended further with a project which analyzed drug resistance in HIV/AIDS patients using clustering, which leverages AI to address drug resistance challenges in public health.

The integration of traditional knowledge systems with modern data technologies is another emerging area of interest. The study on Patient Management and Health Outcome Monitoring by Traditional Healers in Botswana provides a unique perspective on how data from non-conventional healthcare systems can be incorporated into formal health monitoring frameworks.

The COVID-19 pandemic underscored the importance of adaptable and scalable data management systems. In my publication on Experiences, Lessons, and Challenges With Adapting REDCap for COVID-19 Laboratory Data Management in a Resource-Limited Country, we explored the adoption of a research data platform to manage critical health data, highlighting the adaptability of digital tools in crisis situations.

Building on this, I am currently leading a project that focuses on predicting COVID-19 mortality in Botswana using machine learning models. By leveraging high-dimensional clinical datasets, which include both structured clinical parameters and unstructured textual data, our models aim to provide early and accurate predictions of patient outcomes. This project illustrates the power of machine learning in enhancing healthcare systems' ability to proactively manage pandemic-related challenges. The insights gained from this work not only inform clinical interventions but also help optimize resource allocation in healthcare, a critical need in resource-constrained environments like Botswana.

Through this presentation, I will synthesize these experiences to explore how data can drive innovation, improve decision-making, and overcome challenges in resource-constrained environments. The talk will also offer insights into future opportunities for data utilization in sectors ranging from education to healthcare, with a focus on low-resource settings, as well as experiences on hurdles to obtaining data for research and recommendations for better policies related to data management, sharing, and protection.

Speaker: Audrey Masizana* (University of Botswana)

2024-chpc-dirisa-keynote-masizana.pdf

16:30 Transformer-based Sense Embeddings with Deep Learning Large Language Models for Low-Resource Language Disambiguation

Determining a word's accurate meaning in each context is known as Word Sense Disambiguation (WSD)[1], and it is one of the most significant problems in Natural Language Processing (NLP)[2]. This undertaking is particularly challenging for low-resource languages like Sesotho sa Leboa since there are few annotated corpora and linguistic resources available for them. This study explores the application of many transformer-based and deep learning models for WSD in Sesotho sa Leboa, with good results despite the language's resource constraints. This study employs a variety of deep learning architectures, including transformer-based models such as Recurrent Neural Networks with Long Short-Term Memory (RNN-LSTM), Bidirectional Gated Recurrent Units (BiGRU), and an LSTM-based Language Model (LSTMLM), as well as models like DistilBERT with Naive Bayes (DistilBERT & NB), DeBERTa, T5, and ALBERT[3][4].

The study makes use of the unique hardware characteristics of the T4 GPU to improve and optimize the runtime of deep learning language models, especially big transformers. The purpose of the NVIDIA T4 Tensor Core GPU is to speed up deep learning and machine learning operations. It works especially well for training and inferring big language models. Every phase entail making efficient use of software optimizations in addition to comprehending and utilizing hardware features. The BiGRU model outperformed other deep learning language models with an accuracy of 79%, demonstrating the effectiveness of bidirectional processing effectively capturing contextual information. With an accuracy of 70%, DeBERTa beat the other transformer-based large language models to enhance pre-training techniques that prioritize spatial and contextual embeddings.

Keywords: Word Sense Disambiguation, Sesotho sa Leboa, Low-Resourced Languages, Deep Learning, Transformer Models, RNN-LSTM, BiGRU, DeBERTa, NLP

Speaker: Hlaudi Masethe* (Tshwane University of Technology)

Transformer-based_Sense_Embeddings_with_Deep_Learning_Large_Language_Models_for_Low-Resource_Language_Disambiguation.pdf

15:30 → 16:30 HPC Applications: Women in HPC BoF

Convener: Chair: Prof Khomotso Maenetja (Materials Modelling Centre)

15:30 Women in HPC BoF

“Science and Fun” Session at 2024 CHPC- Women in HPC

Location

Boardwalk International Convention Centre
Gqeberha (Port Elizabeth)

Date & Time: 04 December 2024, 15:30 -16:30
General Information
Title: “Women in High Performance Computing South Africa (WHPC-South Africa)”
Duration: 60 minutes
Type of session: Science and Fun Session
Organiser(s):
Name Affiliation Email Address
1. Khomotso Maenetja University of Limpopo khomotso.maenetja@ul.ac.za
2. Raesibe Ledwaba University of Limpopo raesibe.ledwaba@ul.ac.za
3. Rosinah Modiba CSIR RMahlangu1@csir.co.za
4. Beauty Shibiri University of Limpopo beauty.shibiri@ul.ac.za
Description:

The WHPC BOF session for 2024 will be a mix of fun and sharing of scientific ideas in a form of flash poster presentations from women in HPC and short presentation (without a poster). The session will allow participants to give an overview of their current research findings and chance to get to know each other on a lighter note (fun games in place). This will give highlights on the extensive collaborations underway with NICIS and our national and international partners.
Consequently, we are pleased to extend an invitation to both male and female conference attendees to continue where we left off with the last session during the 2023 annual conference. By bringing them together during the meeting, the initiative's primary objective was to establish a network of women in HPC in South Africa. The CHPC management team provided significant support for the session, which was sponsored and attended by both men and women.
Anticipated Goals
• Strong professional relationship
• Improve women's underrepresentation in HPC (Contribute in increasing the number of women and girls participation in HPC through training and networking)
• Share information and resources that foster growth for women in HPC (institutionally and across community)
• Raise our professional profiles
• Encourage young girls at school level to consider HPC as a career of choice

Size: 80
Target audience: Women and Men
Prerequisites: Registered CHPC conference attendees

Chair(s):
Name, Affiliation, Email Address
Dr B Shibiri University of Limpopo beauty.shibiri@ul.ac.za

Outline of programme: — Single 60 min
1. Opening – Prof Khomotso Maenetja (3 min)
2. Team building – Ms Tebogo Morukuladi (15 min)
3. Presentations – Students and Researchers (2 min each – overall time 30 min)
4. Team building – We all participate in various hands on activities (10 min).
5. Closure – Ms Christine Mkhonto (2 min)
Registration link: https://events.chpc.ac.za/event/13

Speaker: Prof. Khomotso Maenetja (Materials Modelling Centre)

17:00 → 19:00 Intermission

19:00 → 19:45 Keynote: Closing

Convener: Chair: Dr Happy Sithole (CHPC)

19:00 The Third Pillar of HPC

The advent of Exascale computing in 2022 marks a major milestone in HPC but also demonstrates its limitations for future progress. Historically, conventional MPP (and large commodity clusters) have achieved enhanced performance by a factor of 2 every two years. In addition to CPUs, GPUs have extended this through streaming SIMD computations for certain classes of application algorithms. But the end of Moore’s Law as well as Dennard Scaling is severely constraining future progress, especially with respect to cost as Frontier approaches 8,000 square feet and only one other, Aurora, has been announced since then. The major class of supercomputer computation not adequately addressed is that of dynamic adaptive graph processing required for advanced forms of machine intelligence; hence, the third pillar of computation. Graphs exhibit neither much spatial locality nor temporal locality but suggest what may be called “logical locality” as the data structures explicitly define their own topologies. However, a new approach to computer architecture, a non-von Neumann family, that is both dynamic and adaptive can easily provide an order of magnitude performance to cost advantage over current methods. While this form of improvement is particularly advantageous for dynamic graph processing, it also can enhance more typical matrix processing. This closing Keynote address will introduce the foundational concepts of the Active Memory Architecture which is being pursued by the Texas Advanced Computing Center. Questions will be addressed from the participants throughout the presentation.

Speaker: Prof. Thomas Sterling* (TACC)

19:45 → 22:00 Awards Dinner