PSNC as CUDA Research Center

2012-06-29

PSNC as CUDA Research Center
July 26, 2012 marks one year of PSNC activity as the CUDA Research Center in cooperation with NVIDIA. The main area developed in the Research Center is an advanced visualization.
Currently, the Center is working on ‘CaKernel’ framework designed to exa-scale computing together with the Center for Computation & Technology Louisiana State University. It is an innovative solution for parallel computing. The only requirement for the user is to create such a code, which can be executed independently for each grid cell. The CUDA Research Center is working on the massive parallel computing codes for graphics cards using CUDA and OpenCL environments. Parallelized calculations using graphics accelerators allow for much more efficient calculations with a decrease in the amount of energy used. With the CaKernel module in real-life simulations significant acceleration in comparison with standard programming solutions is gained.

On the basis of past experience in the field of image processing and advanced visualization, we decided to create a prototype immersion visualization system – so called. “Cave”. It is planned to construct a system of reconfigurable and portable screens that will allow this equipment to fit specific applications and the constraints of available space and create other unusual display systems. Especially for this task it is planned to purchase a professional NVIDIA Quadro graphics device, providing high efficiency (the pixel rate), respectively needed to handle a sufficient number of projectors and unique function of a distributed multi-GPU synchronization (called GSync) necessary to create and maintain for Cave users to experience the full immersion.

In connection with that cooperation, 23-24.04.2012 Poznan Supercomputing and Networking Center hosted NVIDIA experts in the fields of visualization. The goal of the visit visit was the seminar on “How to build a Viz center” devoted to advanced visualization in the CUDA Research Center.