Several ongoing projects on Navier-Stokes models and Lattice Boltzman methods have shown very large speedups using CUDA-enabled GPUs. This work is highlighted below starting with some charts, followed by links to technical reports. There is also work on weather modeling and ocean modeling using GPUs.
![]() |
![]() |
| Incompressible Navier-Stokes Thibault and Senocak |
Lattice Boltzman Methods Tolke and Krafczyk |
Key CFD ISVs and Applications Using CUDA
| ISV/Application | Supported Features | Expected Speed up* | Release Status |
| Altair AcuSolve | Linear equation solver, multi-GPU | 2x | Released, v1.8a |
| Autodesk Moldflow | Linear equation solver | 1.5x | Released, v2012 |
| FluiDyna Culises for OpenFOAM | Linear equation solvers | 3x Solver | Released, v1.2 |
| Prometech Particleworks | MPS, Particle CFD, multi-GPU | 4x-9x | Released, v2.5 |
| Turbostream Ltd. | Explicit compressible solver, multi-GPU | 19x | Released, v2.0 |
| Vratis SpeedIT eXtreme for OpenFOAM | Linear equation solvers, multi-GPU | 6x Solver | Released, v1.2 |
| FluiDyna LBultra | Lattice-Boltzmann solver, multi-GPU | 20x | Released, v2.0 |
| Vratis ARAEL | Linear equation solvers | 3x Solver | Released, v1.0 |
| ANSYS CFD (FLUENT) | Radiation heat transfer model | 10x | In development, v14.0 |
| ESI CFD-ACE+ | Linear equation solver, multi-GPU | 2x | In development, v2012 |
| Metacomp CFD++ | Linear equation solver | In development | In development, v12 |
| S3D(Sandia NL) | Chemistry kernel | 5x | In development |
| SD++ (Stanford-Jameson) | Explicit compressible solver | 16x | In development |
| FEFLO (GMU-Lohner) | Implicit and explicit solvers | 2x – 10x | In development |
* Expected Speed Up vs a quad-core x64 CPU based system. Speed-ups as per NVIDIA in house testing or application provider documentation.
MATLAB is a registered trademark of The MathWorks, Inc.
Jacket is a trademark of AccelerEyes