Imagine trying to run a modern app on a computer that is a million times faster, but the app is written in a way that creates a traffic jam in the memory. This project finds and funds the smartest ways to rewrite those instructions so they can actually use all that power. It's like upgrading a city's roads from single lanes to twenty lanes so the traffic actually moves.
Existing software cannot keep up with the speed of new supercomputers because data moves too slowly and algorithms aren't designed for GPU-CPU combinations.
A funding and management mechanism that produced a series of published algorithms designed to reduce time-to-solution and energy-to-solution on Exascale systems.
If you are a drug discovery firm dealing with slow molecular simulations — this project developed new algorithms that reduce the time-to-solution. This allows for faster screening of compounds using European Exascale resources.
If you are an aircraft design manufacturer dealing with massive fluid dynamics calculations — this project developed efficient software mapping for heterogeneous hardware. This helps in reducing the energy-to-solution for complex wing simulations.
If you are a weather forecasting service dealing with data bottlenecks on supercomputers — this project developed algorithms to bridge the gap between compute power and data movement. This enables more precise, large-scale environmental models.
Based on available project data, the project focused on funding the development of algorithms via a 4.1M€ call; specific commercial pricing for the resulting software is not listed.
Yes, the project specifically targets Exascale and post-Exascale computers to ensure the most successful algorithms are taken up by science and industry for performance gains.
Based on available project data, the results include the publication of novel algorithms, but specific licensing terms for industrial use are not detailed.
The innovation studies funded by the project lasted from February/March 2024 until the end of February 2025.
They are designed to map computations onto heterogeneous hardware, specifically combining multi-core CPUs with graphics processing units (GPUs).
The consortium is lean and specialized, consisting of 4 partners from 3 countries (BE, DE, ES). It features a 25% industry ratio with one SME, balanced by two major European HPC research centers and one university. This structure suggests a strong focus on technical execution and academic rigor with a direct link to industrial application.
Contact the Barcelona Supercomputing Center (BSC)
Contact us to identify which of the 27 funded algorithms fits your computational bottleneck.
