- SpeedIT FLOW
- Sparse Linear Algebra
- GPU computing
- Live Cell Imaging
- Computational Fluid Dynamics
- EU-funded projects
Meet us at the workshop organized in the framework of the DISTRICT+ project. Our presentation will be held on 20 September 2012 at 11:00AM in Wroclaw Technology Park.
Now, you can run your OpenFOAM simulations on more than one GPU cards in your system.
The technology was tested at CINECA cluster on up to 64 GPU cards on cases up to 32 millions cells.
Reports on acceleration are accessible at vratis.com/blog.
- Decomposed cases can be submitted to multiple GPU cards.
- MPI is used for subdomain communication.
- Supports Fermi class GPU architectures.
- Good scalability up to 88 GPUs.
GPU Science reports a benchmark of SpeedIT 2.1 at http://gpuscience.com.