Master’s in HPC & GPU for Simulation and AI
Online master's programme of 1,900 hours in HPC and GPU computing for simulation and AI: CUDA/OpenMP, scaling, performance and cost. 12 modules covering HPC architectures, parallel CPU programming, CUDA, portability models, accelerated numerical methods, CFD and FEM in HPC, large-scale AI training, AI for scientific simulation, performance optimisation, HPC operations and infrastructure procurement, with an assigned technical tutor.
Master’s in HPC & GPU for Simulation and AI — studying from the United Kingdom
Data science, artificial intelligence and simulation are central to engineering practice across the United Kingdom, with strong hubs in London, Cambridge, Oxford, Bristol and Edinburgh. Universities, research institutes such as the Alan Turing Institute and industrial users in aerospace, automotive, energy and finance demand skills in modelling, machine learning and computational analysis. Engineering simulation tools are widely used for design validation and digital twins. Organisations seek professionals who can turn large datasets into engineering decisions while meeting data protection and governance requirements.
Fees for the United Kingdom: £8,280 (Pound sterling), with instalments and employer sponsorship available.
Relevant bodies in the United Kingdom
Typical roles
Questions from the United Kingdom
Which UK bodies recognise data and AI qualifications?
In the UK, the British Computer Society (BCS), the Royal Statistical Society (RSS) and the Institute of Analytics are relevant professional bodies, and the IET covers many engineering-related AI and data areas. SEIUM's qualifications are private and not accredited by these bodies. Check directly with BCS or RSS whether your study would be considered towards their membership or certification.
Do UK employers in data and AI value private online qualifications?
UK employers in finance, retail, health, technology and public sector analytics typically assess candidates on technical skills, portfolio work and experience with real data. A private qualification may help demonstrate specialist learning, but its value depends on each employer's judgement. Publishing projects and showing practical results will usually matter most.
Can I study this online from the UK alongside work, and what English level is required?
Yes, the programmes are fully online and can be studied from anywhere in the UK without a student visa. Teaching is in English, so you need a solid working level of technical English. An assigned technical tutor will support you. Check the programme hours against your working commitments and ask about instalment payments or employer sponsorship.
Take the controls.
Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.
What you will be able to do
Understand CPU, GPU and high-performance cluster architectures.
Program parallel applications with MPI, OpenMP and CUDA.
Accelerate numerical methods and engineering simulations on GPUs.
Train AI models in a distributed, efficient way.
Develop surrogate models that combine simulation and AI.
Analyse and optimise the performance of scientific codes.
Operate and size HPC infrastructure sustainably.
Who it's for
- HPC engineers.
- CUDA developers.
- Computational simulation engineers.
- AI infrastructure engineers.
- HPC system administrators.
- Scientific computing researchers.
- Engineering or science graduates looking to work in supercomputing centres.
What you'll work with
- MPI 4.0 Standard (MPI Forum) (message passing interface)
- OpenMP 5.2 Specification (shared-memory parallel programming)
- SYCL 2020 Specification (Khronos) (heterogeneous programming in C++)
- IEEE 754 (floating-point arithmetic)
- ISO/IEC 30134-2 (PUE) (data centre power usage effectiveness)
- EN 50600 (data centre facilities and infrastructure)
- ISO/IEC 14882 (C++) (C++ programming language)
- CUDA Toolkit
- Slurm
- OpenFOAM
- PyTorch
- NVIDIA Nsight
12 modules
1.1Moore's law and the end of Dennard scaling1.2Multicore CPU architecture1.3Memory hierarchy and caches1.4GPU architecture and SIMT1.5Clusters, nodes and InfiniBand networks1.6Lustre and GPFS parallel storage1.7FLOPS, bandwidth and roofline metrics1.8The TOP500 and Green500 lists1.9Energy cost of computing1.10Case study: roofline analysis of a code
2.1Data and task parallelism2.2OpenMP for shared memory2.3SIMD vectorisation and AVX-5122.4MPI for distributed memory2.5Collective communications2.6Domain decomposition2.7Load balancing2.8Strong and weak scalability2.9Amdahl's and Gustafson's laws2.10Lab: hybrid MPI+OpenMP parallelisation
3.1The CUDA programming model3.2Threads, blocks and warps3.3Global memory, shared memory and registers3.4Coalesced memory access3.5Occupancy and latency3.6Streams and concurrency3.7cuBLAS, cuFFT and cuSPARSE libraries3.8Debugging and profiling with Nsight3.9Unified Memory3.10Practical exercise: optimised convolution kernel
4.1HIP and AMD GPUs4.2SYCL and oneAPI4.3Directive-based OpenACC4.4Kokkos and RAJA4.5Accelerated Python with Numba and CuPy4.6JAX and XLA compilation4.7Multi-GPU programming4.8NCCL communication4.9Comparison of portable performance4.10Workshop: migrating a CUDA code to SYCL
5.1Dense linear algebra on GPUs5.2Sparse and iterative solvers5.3Scalable preconditioners5.4Multigrid methods5.5Distributed FFT5.6Mixed precision and its stability5.7Parallel random number generation5.8Monte Carlo methods on GPUs5.9Verification of numerical results5.10Lab: conjugate gradient on GPU
6.1Discretisation for parallel computing6.2OpenFOAM on a cluster6.3Lattice Boltzmann methods on GPUs6.4Implicit FEM and mesh partitioning6.5Parallel meshing and METIS6.6Commercial solvers with HPC licences6.7Parallel I/O with HDF56.8In situ post-processing6.9Scalability of industrial cases6.10Case study: scaling a CFD simulation
7.1Accelerators and Tensor Cores7.2Distributed data parallelism7.3Model and pipeline parallelism7.4ZeRO and FSDP7.5Mixed-precision training7.6Checkpointing and fault tolerance7.7Data loader optimisation7.8Scaling laws7.9Energy cost of training7.10Lab: multi-node training with PyTorch
8.1Surrogate models of simulation8.2Physics-informed neural networks8.3Neural operators and FNO8.4Learning on meshes with GNNs8.5AI-solver coupling8.6Generating training data with HPC8.7Surrogate validation8.8Accelerated inference with TensorRT8.9Hybrid simulation-AI workflows8.10Practical exercise: surrogate of a thermal simulation
9.1Performance analysis methodology9.2Profiling with VTune and Nsight Systems9.3Hardware counters9.4Memory and locality optimisation9.5Communications and overlap9.6I/O bottlenecks9.7Compiler tuning9.8HPL and HPCG benchmarks9.9Performance reports9.10Workshop: profiling-guided optimisation
10.1Slurm and PBS job schedulers10.2Apptainer and Singularity containers10.3Module environments and Spack10.4Cluster monitoring10.5Cloud HPC and bursting10.6Capacity planning10.7Energy efficiency and PUE10.8Security in computing centres10.9User and quota management10.10Practical exercise: complete workflow in Slurm
11.1Designing a GPU cluster11.2Liquid cooling and power density11.3High-performance networks11.4Storage for AI11.5Capital and operating costs11.6On-premise versus cloud comparison11.7Public procurement specifications for HPC11.8The Spanish Supercomputing Network and EuroHPC11.9Sustainability of computing11.10Case study: sizing a corporate cluster
12.1Selection of the code or application12.2Initial performance analysis12.3Parallelisation strategy12.4GPU implementation12.5Optimisation and scaling12.6Validation of results12.7Cost and energy analysis12.8Conclusions and future work12.9Thesis report12.10Master's Thesis: defence before a panel
How you'll learn
The programme is delivered 100% online on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. There are no fixed timetables: you progress at your own pace within each intake, so it fits around your job.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is CUDA Toolkit, Slurm, OpenFOAM, PyTorch and NVIDIA Nsight, in free, academic or evaluation versions. Exercises are based on real procedures: cluster scalability studies, GPU kernel profiling and distributed training benchmarks. Assessment is continuous by module, plus the final Master's Thesis.
Standards covered: MPI 4.0 Standard (MPI Forum), OpenMP 5.2 Specification, SYCL 2020 Specification (Khronos), IEEE 754, ISO/IEC 30134-2 (PUE), EN 50600, ISO/IEC 14882 (C++).
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Where this takes you
Requirements
- University degree in engineering, science or architecture, or equivalent demonstrable professional experience.
- Internet connection and a computer capable of running the programme software.
- English sufficient to read technical standards.
Ready to start?
Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.
Frequently asked questions
The programme totals 1,900 hours over 12–14 months. You set your own pace within each intake, and your tutor helps you plan the workload around your job.
You work with CUDA Toolkit, Slurm, OpenFOAM, PyTorch and NVIDIA Nsight, using free, academic or evaluation versions. Your tutor indicates the free alternative wherever one exists.
A SEIUM Institute diploma, awarded on passing the continuous assessment and the Master's Thesis. It is SEIUM's own qualification, not an official state degree; its recognition depends on the programme completed and the regulations of each country.
Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the higher programme. Ask about the recognition policy when you enrol.
A university degree in engineering, science or architecture (or equivalent professional experience), a suitable computer and English sufficient to read technical standards.
Start dates are set by SEIUM's faculty for each intake. Our admissions team will confirm the next available date when you apply.
Related programmes
Master’s in AI-Based Predictive Maintenance Engineering
Online master's programme of 1,900 hours in AI-based predictive maintenance: 12 modules covering maintenance data, feature engineering, fault diagnosis, asset anomaly detection, remaining useful life prognosis,…
Master’s in Technical NLP and Engineering Assistants (RAG)
Online master's programme of 1,900 hours in technical NLP and engineering assistants (RAG) for documentation, logs and expert support. 12 modules covering natural language in engineering,…
Master’s in Optimisation under Uncertainty
Online master's programme of 1,900 hours in optimisation under uncertainty: robust and stochastic optimisation, scenarios and trade-offs. 12 modules covering stochastic programming, robust optimisation, chance constraints…
More in Data, AI, Simulation & Analytics
All Data, AI, Simulation & Analytics programmes → · How to apply · Fees & financing · Scholarships