[HTML][HTML] CUBENS: A GPU-accelerated high-order solver for wall-bounded flows with non-ideal fluids

PC Boldini, R Hirai, P Costa, JWR Peeters… - Computer Physics …, 2025 - Elsevier
We present a massively parallel GPU-accelerated solver for direct numerical simulations of
transitional and turbulent flat-plate boundary layers and channel flows involving fluids in non …

The 2DECOMP&FFT library: an update with new CPU/GPU capabilities

S Rolfo, C Flageul, P Bartholomew, F Spiga… - Journal of Open Source …, 2023 - hal.science
The 2DECOMP&FFT library is a software framework written in modern Fortran to build large-
scale parallel applications. It is designed for applications using three-dimensional structured …

[HTML][HTML] CaLES: A GPU-accelerated solver for large-eddy simulation of wall-bounded flows

M **ao, A Ceci, P Costa, J Larsson… - Computer Physics …, 2025 - Elsevier
We introduce CaLES, a GPU-accelerated finite-difference solver designed for large-eddy
simulations (LES) of incompressible wall-bounded flows in massively parallel environments …

Differentiable Cosmological Hydrodynamics for Field-Level Inference and High Dimensional Parameter Constraints

B Horowitz, Z Lukic - arxiv preprint arxiv:2502.02294, 2025 - arxiv.org
Hydrodynamical simulations are the most accurate way to model structure formation in the
universe, but they often involve a large number of astrophysical parameters modeling …

A pencil-distributed finite-difference solver for extreme-scale calculations of turbulent wall flows at high Reynolds number

RD Sanhueza, J Peeters, P Costa - arxiv preprint arxiv:2502.06296, 2025 - arxiv.org
We present a computational method for extreme-scale simulations of incompressible
turbulent wall flows at high Reynolds numbers. The numerical algorithm extends a popular …

WaterLily. jl: A differentiable fluid simulator in Julia with fast heterogeneous execution

GD Weymouth, B Font - arxiv preprint arxiv:2304.08159, 2023 - arxiv.org
Integrating computational fluid dynamics (CFD) software into optimization and machine-
learning frameworks is hampered by the rigidity of classic computational languages and the …

CaNS-Fizzy: A GPU-accelerated finite difference solver for turbulent two-phase flows

G Lupo, P Costa, P Wellens - arxiv preprint arxiv:2502.04189, 2025 - arxiv.org
CaNS-Fizzy--Fizzy for short--is a GPU-accelerated numerical solver for massively-parallel
Direct Numerical Simulations (DNS) of incompressible two-phase flows. A DNS enables …

Characterization of NCCL and Unified Memory Under Normal and Oversubscribed Memory Conditions

R Strina - 2024 - search.proquest.com
Abstract The NVIDIA Collective Communications Library (NCCL) is a multi-GPU
communication library widely used in applications such as deep learning, molecular …

[PDF][PDF] GPU-Accelerated Atmospheric Large Eddy Simulation

CAA Jungbacker - repository.tudelft.nl
With the completion of this thesis, my time as a student at TU Delft has come to an end.
During my BSc in civil engineering, I developed a passion for fluid dynamics, numerical …

[PDF][PDF] Towards Enabling Automatically Differentiable High Performance ComputingForCosmologicalN-bodySimulations

W Kabalan, F Lanusse, A Boucaud, E Aubourg… - moriond.in2p3.fr
A series of recent works highlighted the potential of so-called full-field cosmological
inference for the analysis of upcoming weak lensing and galaxy clustering surveys, with the …