Today graphics processors (GPUs) are becoming increasingly programmable, offering the potential for dramatic speedups in the computation of general purpose applications. Indeed, the cost, speed and bandwidth characteristics of these devices make them a special candidate as computing platform for such applications. The increase in performance appears particularly interesting in the aerodynamics field, because the high computational time required to solve the mathematical problem. The object of the work is to accelerate Ansys-Fluent applications, in order to obtain a significant reduction in the time to obtain the solution. The tests are carried out with a single computing node (two Nvidia K40 GPUs) and with a cluster configuration (eight Nvidia K20 GPUs with an Infiniband network between GPUs in different cluster nodes). In this report our experience with using GPUs as fast parallel co-processors are descrive, and examples of capabilities and problems are reported, both for to aeronautical and vehicles aerodynamics.
Impatto dello Sviluppo della Potenza di Calcolo sul progetto aerodinamico: prospettive e problematiche dell’uso dei processori GPU in ANSYS-FLUENT.
LOMBARDI, GIOVANNI;
2015-01-01
Abstract
Today graphics processors (GPUs) are becoming increasingly programmable, offering the potential for dramatic speedups in the computation of general purpose applications. Indeed, the cost, speed and bandwidth characteristics of these devices make them a special candidate as computing platform for such applications. The increase in performance appears particularly interesting in the aerodynamics field, because the high computational time required to solve the mathematical problem. The object of the work is to accelerate Ansys-Fluent applications, in order to obtain a significant reduction in the time to obtain the solution. The tests are carried out with a single computing node (two Nvidia K40 GPUs) and with a cluster configuration (eight Nvidia K20 GPUs with an Infiniband network between GPUs in different cluster nodes). In this report our experience with using GPUs as fast parallel co-processors are descrive, and examples of capabilities and problems are reported, both for to aeronautical and vehicles aerodynamics.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
