For more than ten years, GPU hardware (Graphics Processing Unit), which was originally designed for gaming graphics, has been successfully used for the fast solution of various numerical tasks. The performance level is often several times higher than that of comparable CPU applications (Central Processing Unit). Meanwhile, the computing power and the available RAM memory are so large that numerical flow simulation tasks of industrial scope can gain a lot of performance from the General Purpose GPU (GPGPU).
Our SPH-based simulation software PreonLab benefits greatly from the additional computing power for many of its different simulation types. Dr.-Ing. Carsten Schmalhorst, Solution Area Expert, AVL Deutschland GmbH, and Dominik Friedrich, HPC, AI & Quantum Presales Consultant, Atos Information Technology GmbH, will show you how you can benefit from currently available hardware - from single cards to multi GPU applications – and what requirements they place on software operations.
Key topics and takeaways:
• General Purpose GPU – Which approach is pursued? Which hardware is available?
• Performance of PreonLab on different GPGPU systems
• GPGPU operations in the HPC environment
• Rightsizing: The right hardware for your task