ExaFlow

Turbulent flows are widely observed in nature and various industries, representing the most common flow regime that can be visually perceived by the human eye. These flows exhibit captivating three-dimensional, chaotic, and multi-scale dynamics. Direct Numerical Simulations (DNS) of the Navier-Stokes equations have revolutionized turbulence research by revealing its dynamics in full detail. However, several fundamental challenges still persist, particularly regarding the flow dynamics at high speeds (high Reynolds numbers), which can now be tackled through high-fidelity simulations made possible by advancing computing power and computational tools. This application aims to support a DNS solver that utilizes massively-parallel High- Performance Computing resources based on CPUs/GPUs, enabling the resolution of future challenges. The focus is twofold: Firstly, enhancing workflows for the sustainable development of our solvers through continuous integration. Secondly, ensuring performance portability to achieve high efficiency across multiple nodes with varying hardware configurations while maintaining a sustainable codebase.