Assessing the Performance of Compute Unified Device Architecture

957 WordsFeb 22, 20184 Pages
The academic paper at hand aimed at assessing the performance of Compute Unified Device Architecture (CUDA) programming for machining simulation. It has three main dimensions, introducing the machining simulation algorithm parallelization, discussing the main constraints related to the use of the CUDA architecture proposing different solutions to the problem of parallelization with CUDA. Authors have argued that the main purpose of machining simulations in Computer Aided Manufacturing (CAM) software is to detect collisions between the tool and the part on the whole part surface of manufactured goods. Such surfaces are employed to provide sealing or friction functions, optical or aero dynamics properties. This task requires specific geometrical deviations to be executed within a given manufacturing process. More specifically, reducing the cycle time of developing a product will rely on how to simulate the evolution of geometrical deviations of surfaces throughout the manufacturing process. The literature reveals that there are many methods to perform machining simulations. For instance, methods based on a partition of the space by lines, by voxels or by planes, methods based on meshes … etc. However, simulations of the machined surfaces in CAM software are very limited, purely geometric and don’t include any characteristic of the actual process that may damage the surface quality during machining. Furthermore, these simulations do not provide the required accuracy within a

More about Assessing the Performance of Compute Unified Device Architecture

Open Document