December 09, 2024 - by CSCS
Thirty-five proposals were submitted for the new period 2025 to 2028, requesting a total of CHF 16.8 million. Fourteen were finally approved with a total budget of CHF 6.0 million. The submitted projects were required to address the broad availability, quality, and performance of software on GPU-accelerated supercomputing platforms. They can focus on simulation or data analysis using approaches from computational science or the machine learning field and must lead to application software related to significant scientific problems that require high-end GPU-accelerated supercomputers to be solved.
The new allocation period in PASC starts on 1 January 2025. The selected projects cover various scientific disciplines and rely on interdisciplinary collaboration between domain scientists, mathematicians, computer scientists, computational scientists and software developers. The projects will receive funding from the Swiss High-Performance Computing and Networking (HPCN) initiative, which will finance around 22 full-time positions for the entire duration of the projects. All projects must provide matching funding. In addition, every project is awarded a development project on the “Alps” infrastructure with 15,000 node hours per year in compute resources.
The funded projects are the following:
- Boosting large-scale quantum transport simulations through GPU-based dedicated libraries (BoostQT), led by Mathieu Luisier, Professor at ETH Zurich.
- Correlated electrons on accelerated architectures from frequency-dependent response functions, led by Nicola Colonna, Tenure track scientist at PSI.
- Modular and user-friendly machine learning frameworks for atomistic modelling at scale, led by Michele Ceriotti, Professor at EPFL.
- Alpenglue - Advances in lattice applications for the next generation of large and fine discrete Quantum ChromoDynamics ensembles, led by Urs Wenger, Professor at the University of Bern.
- openQxD: Efficient QCD+QED Simulations with Various Boundary Conditions on GPUs, led by Marina Krstic Marinkovic, Professor at ETH Zurich.
- HiRAD-Gen - High-Resolution Atmospheric Downscaling Using Generative Machine Learning Models, led by Sebastian Schemm, Professor at ETH Zurich.
- Next-Generation Radio Interferometry II, led by Jean-Paul Kneib, Professor at EPFL.
- XSES-FSI: towards eXtreme scale Semi-Structured discretizations for Fluid-Structure Interaction, led by Patrick Zulian, Post-doctoral Researcher at UniDistance Suisse.
- OPAL-X – Object-oriented Parallel Accelerator Library for eXascale, led by Andreas Adelmann, Head of Laboratory for Simulation and Modelling at PSI.
- RandESC: Novel algorithms based on randomization and mixed precision for electronic-structure calculations, led by Laura Grigori, Professor at EPFL.
- Toward the next-generation simulations of the tokamak boundary, led by Paolo Ricci, Professor at EPFL.
- ∂GPU4GEO: Differentiable multi-physics solvers for extreme-scale geophysics simulations on GPUs, led by Ludovic Räss, Computational Geoscientist at the University of Lausanne.
- Accelerated Radiative Transfer Simulations for SKAO, led by Alexandre Refregier, Professor at ETH Zurich.
- Establishing the Portable Model for Multi-Scale Atmospheric Prediction, led by Heini Wernli, Prof. at ETH Zurich.