Heterogeneous computing, holds many opportunities for simulation and data analysis applications in the physical sciences. On the desktop, massively parallel calculations are now possible with the use of GPUs. We are currently exploring the capabilities of Nvidia’s CUDA platform on multi-GPU machines, and application to new and existing applications. This project is closely related to our AR/VR project.
The image above shows AntonJr, a dual-CPU (Xeon E5-2680), triple-GPU (GeForce 1080Ti), water-cooled machine.