We developed a complete GPU based data processing for cone-beam micro CT application which performs not only the reconstruction but also all the correction of the projection images on-the-fly. Test measurements were performed and processing times was compared on different hardware setups. The performance of the GPU together with our modified algorithm allow to process all the calculations within the shortest exposure time of the system for the highest reconstructed volume size thus makes possible the complete data processing in real-time.

In this paper results of an ongoing research on cold-formed C-section members are presented. The research aims the development of design methods for structural arrangements directly not covered by the standard, based on laboratory tests and numerical modelling. In the paper laboratory tests on C-section compression members are presented: arrangement, behaviour modes and load-bearing capacities. The behaviour of the specimens as well as the effect of the non-rigid end support provided by the self-drilling screws used to introduce load in the specimens are discussed. A numerical model capable of geometrically and materially nonlinear, imperfect analysis to reproduce the tests is introduced with a focus on the modelling of the connector elements and imperfections. An approach to model selfdrilling screws and equivalent geometrical imperfections is presented; the applicability of the model is shown. It is shown, that screw stiffness and geometrical imperfections both influence the stiffness of the models. The calibration of the model is carried out using the results of the experimental tests introduced previously; the results of these virtual experiments are presented.

No abstract

In this paper, we present a multi-level programming model for recent GPU-based high performance computing systems. Involving cooperative stream threads and symmetric multiprocessing threads our model gives a computational framework that scales through multi-GPU environments to GPU-cluster systems. Instead of hiding the execution environment from the programmer using compiler extensions or metaprogramming techniques we aim a solution that both enables optimizations and provides abstract problem space mapping with code reusability and virtualization of hardware resources in order to decrease the programming effort. We evaluate an implementation of our model based on CUDA, OpenMP, and MPI2 technologies on a complex practical application scenario and discuss its performance scaling behavior.

No abstract