Golovina E1, De La Pierre M1, Han L1, Blasco J1
1Do It Now, , Aotearoa / New Zealand
Biography:
Evgeniia Golovina is a Computational Scientist at Do IT Now, a global consulting company providing advanced supercomputing services for scientists and engineers. Evgeniia holds a PhD (2021) in Health Sciences from the University of Auckland, New Zealand. Since May 2024, she is a member of the HPC team at DO IT Now where she is mainly responsible for user support & training, including software stack installation and management on High-Performance Computing (HPC) systems. Evgeniia is curious about new technologies and innovative HPC solutions and enjoys using infrastructure-as-code platforms, software build and installation frameworks and workflow automation tools in her work. In her free time, she is learning languages, practicing hot yoga or going out for a hike.
Abstract:
Scientific computing is becoming increasingly complex as new hardware platforms and scientific software continue to emerge. The traditional model of installing and maintaining software independently on each HPC system is time-consuming, often leads to inconsistent user environments, and can affect performance, portability and reproducibility.
The European Environment for Scientific Software Installations (EESSI) addresses these challenges by offering a unified and reproducible approach to scientific software delivery, Built on EasyBuild, EESSI provides a shared, optimised software framework that can run consistently across large-scale HPC systems, cloud environments, and personal workstations, reducing the need for repeated, site-specific installations.
In this talk, we will introduce the EESSI framework and how it is structured; why a layered design was chosen, which tools support it, and how this structure design enables reproducibility, binary compatibility, performance and efficiency across different environments. We will show how EESSI supports both traditional HPC workloads and modern AI and data-intensive applications allowing researchers to run their complex software stacks without repeatedly rebuilding dependencies or resolving environment-specific issues. We will highlight how EESSI reduces software installation time, improves reproducibility, guarantees microarchitecture-optimised performance, and lowers maintenance burden. Finally, we will discuss how EESSI can strengthen collaboration between research centres and accelerate scientific work.
Overall, EESSI represents an important step towards a unified, reproducible, collaborative, and optimised software infrastructure for scientific computing. By simplifying software delivery and improving consistency across systems, EESSI helps researchers focus on science, discovery, and innovation rather than software setup.
