Sustainable Scientific Computing

Workshop Aims

This workshop aims to establish a cross-disciplinary dialogue between experts in scientific computing, programming languages and compilers, numerical verification, and computer architecture to address the urgent challenge of reducing the energy footprint of computational science. Through collaborative discussions, tool exploration, and domain-bridging sessions, participants will identify best practices, explore optimization strategies, and build a foundation for sustainable scientific software and hardware co-design. The workshop fosters an exchange of knowledge to create actionable paths towards energy-efficient, numerically robust scientific computing.

Workshop Description

As the world grapples with the climate crisis, scientific computing faces increasing pressure to balance performance with sustainability. The computational demands of high-performance and data-intensive applications result in significant energy consumption, compounded by hardware constraints and limitations in software design. While the computer architecture community has long addressed energy efficiency, the programming languages and scientific computing communities are only beginning to align with this imperative.

This Lorentz Center workshop brings together researchers from three traditionally siloed communities—scientific computing, programming languages, and computer architecture—to develop a shared understanding of energy-efficient computing practices. By focusing on scientific applications, which are often approximation-heavy and floating-point dominated, the workshop aims to explore novel approaches to optimization that respect correctness constraints.

Key topics include:

• Energy-aware benchmarks and proxy applications suited for cross-domain evaluation
• Metrics and tools for measuring and guiding energy-to-solution
• Optimization strategies across the computing stack, from source code to hardware
• Numerical verification and compiler tools enabling automated energy optimizations

The program features plenary talks, panel discussions, and dynamic breakout sessions to identify opportunities for collaboration, tool integration, and the co-design of sustainable computing solutions. A central outcome of the workshop will be a roadmap outlining shared benchmarks, terminology, and immediate actions for joint research proposals and publications.

This initiative aligns with Dutch and European efforts for sustainable digital infrastructure and seeks to foster a community committed to reducing the environmental impact of computational science.