Original public domain image from Flickr
Today, high-performance computing (HPC) is essential for addressing complex, data-heavy problems that touch every aspect of science, industry, and society. At its core, HPC refers to computing systems capable of performing quadrillions of calculations per second, far exceeding the capabilities of standard computers. These systems allow researchers to model and analyse phenomena that would otherwise take years to explore manually.
In practical terms, HPC enables climate science and modelling, materials simulation, genomics research, and energy optimisation. By rapidly processing massive datasets, these systems provide insights that are both timely and actionable. For instance, simulating the behaviour of molecules before physically creating them in a lab can dramatically reduce trial-and-error experimentation, saving both time and resources. Similarly, analysing complex energy systems can highlight efficiency improvements in industrial processes, contributing directly to sustainability goals.
HPC Supporting Data-Intensive Research
In climate science, supercomputers are building detailed simulations of weather patterns, helping to anticipate events like hurricanes or prolonged droughts. In Morocco, models of coastal hydrodynamics are informing flood management along the Atlantic coast, turning complex data into actionable guidance.
However, the use cases and benefits are different for materials science. Researchers can now simulate molecular structures before synthesising them in the lab, reducing trial-and-error testing and saving resources. Similarly, energy systems can be analysed to identify efficiency improvements in industrial processes, showing how HPC supports both innovation and sustainability.
Toubkal: Building HPC Capacity in Africa
Historically, Africa has had limited access to high-performance computing, hosting less than 1% of global data centres. To address this gap, Toubkal was launched in 2021, currently Africa’s most powerful supercomputer.
Toubkal, led by Professor Robert Basmadjian, is already being used to solve real-world problems. Its AI models improve short-term climate predictions, and hydrodynamic simulations guide local flood risk management. These examples illustrate how HPC can support research that is both globally relevant and grounded in local priorities.
Sustainability in Practice
Running a supercomputer sustainably requires more than just renewable energy. Toubkal draws most of its power from 1 MWp of on-site solar panels, harnessing Morocco’s abundant sunlight. The facility also uses advanced cooling systems to maintain a power usage effectiveness (PUE) of 1.35, a measure of data centre efficiency that compares total energy consumption to the energy used by computing equipment. By comparison, the industry average is 1.56, demonstrating its energy-efficient design.
Beyond hardware, Toubkal employs machine learning algorithms to dynamically optimise GPU frequencies according to workload. GPUs, or Graphics Processing Units, are the engines behind many HPC tasks, particularly AI and simulations. By adjusting their speed based on computational demand, Toubkal reduces energy consumption by 13% and slightly shortens execution times, while maintaining high prediction accuracy. The result is a system that balances computational power with practical energy efficiency.
Why Location Matters
By situating Toubkal in Morocco, this enables local researchers to run complex simulations without relying on overseas resources. This strengthens local expertise, encourages collaboration across African institutions, and allows the region to contribute to international research projects.
Access to HPC locally has immediate practical effects. Researchers from Sub-Saharan Africa are using Toubkal for work ranging from genomics to materials modelling. Being able to run simulations on-site reduces delays and lets scientists focus on research questions relevant to their region.
Lessons from Toubkal
Toubkal offers several lessons for HPC development. Sustainability must be considered from the design stage, not added as an afterthought. Proximity to computing resources accelerates research and enables new kinds of experimentation. And providing access to high-performance systems supports a wider, more inclusive scientific community.
Operational challenges, such as managing diverse workloads and integrating solar power, underscore the need for ongoing optimisation, training, and maintenance. These are practical considerations that any emerging HPC centre must plan for.
Looking Ahead
High-performance computing is a powerful tool, but its value is maximised when performance, sustainability, and access come together. Toubkal shows that it is possible to run a world-class system in a way that is energy-efficient and regionally empowering. Expanding HPC infrastructure in Africa will allow local researchers to participate in global science, contribute to solutions for local challenges, and help position the continent as an active player in data-intensive research.
