Vogelgesang M., Chilingaryan S., Rolo T.D.S., Kopmann A.
in Proceedings of the 14th IEEE International Conference on High Performance Computing and Communications, HPCC-2012 – 9th IEEE International Conference on Embedded Software and Systems, ICESS-2012 (2012) 824-829, 6332254. DOI:10.1109/HPCC.2012.116
Current synchrotron experiments require state-of-the-art scientific cameras with sensors that provide several million pixels, each at a dynamic range of up to 16 bits and the ability to acquire hundreds of frames per second. The resulting data bandwidth of such a data stream reaches several Gigabits per second. These streams have to be processed in real-time to achieve a fast process response. In this paper we present a computation framework and middleware library that provides re-usable building blocks to implement high-performance image processing algorithms without requiring profound hardware knowledge. It is based on a graph structure of computation nodes that process image transformation kernels on either CPU or GPU using the OpenCL sub-system. This system architecture allows deployment of the framework on a large range of computational hardware, from netbooks to hybrid compute clusters. We evaluated the library with standard image processing algorithms required for high quality tomographic reconstructions. The results show that speed-ups from 7x to 37x compared to traditional CPU-based solutions can be achieved with our approach, hence providing an opportunity for real-time on-line monitoring at synchrotron beam lines. © 2012 IEEE.
Caselle M., Chilingaryan S., Herth A., Kopmann A., Stevanovic U., Vogelgesang M., Balzer M., Weber M.
in 2012 18th IEEE-NPSS Real Time Conference, RT 2012 (2012), 6418369. DOI:10.1109/RTC.2012.6418369
X-ray computed tomography (CT) is a method for non-destructive investigation. Three-dimensional images of internal structure can be reconstructed using a two-dimensional detector. The poly-chromatic high density photon flux in the modern synchrotron light sources offers hard X-ray imaging with a spatio-temporal resolution up to the μm-μs range. Existing indirect X-ray image detectors can be adapted for fast image acquisition by employing CMOS-based digital high speed camera. In this paper, we propose a high-speed visible light camera based on commercial CMOS sensor with embedded processing implemented in FPGA. This platform has been used to develop a novel architecture for a self-event trigger. This feature is able to increase the original frame rate of the CMOS sensor and reduce the amount of the received data. Thanks to a low noise design, high frame rate (kilohertz range) and high speed data transfer, this camera can be employed in modern synchrotron ultra-fast X-ray radiography and computed tomography. The camera setup is accomplished by high-throughput Linux drivers and a seamless integration in our GPU computing framework. Selected applications from life sciences and materials research underline the high potential of this high-speed camera in a hard X-ray micro-imaging approach. © 2012 IEEE.
Chilingaryan S., Mirone A., Hammersley A., Ferrero C., Helfen L., Kopmann A., Dos Santos Rolo T., Vagovic P.
in IEEE Transactions on Nuclear Science, 58 (2011) 1447-1455, 5766797. DOI:10.1109/TNS.2011.2141686
Advances in digital detector technology leads presently to rapidly increasing data rates in imaging experiments. Using fast two-dimensional detectors in computed tomography, the data acquisition can be much faster than the reconstruction if no adequate measures are taken, especially when a high photon flux at synchrotron sources is used. We have optimized the reconstruction software employed at the micro-tomography beamlines of our synchrotron facilities to use the computational power of modern graphic cards. The main paradigm of our approach is the full utilization of all system resources. We use a pipelined architecture, where the GPUs are used as compute coprocessors to reconstruct slices, while the CPUs are preparing the next ones. Special attention is devoted to minimize data transfers between the host and GPU memory and to execute memory transfers in parallel with the computations. We were able to reduce the reconstruction time by a factor 30 and process a typical data set of 20 GB in 40 seconds. The time needed for the first evaluation of the reconstructed sample is reduced significantly and quasi real-time visualization is now possible. © 2006 IEEE.