Abstract. We investigate the performance characteristics of a numerically enhanced scalar product (dot) kernel loop that uses the Kahan algorithm to compensate for numerical errors, and describe efficient SIMD-vectorized implementations on recent Intel processors. Using low-level instruction analysis and the execution-cache-memory (ECM) performance model we pinpoint the relevant performance bottlenecks for single-core and thread-parallel execution, and predict performance and saturation behavior. We show that the Kahan-enhanced scalar product comes at almost no additional cost compared to the naive (nonKahan) scalar product if appropriate low-level optimizations, notably SIMD vectorization and unrolling, are applied. We also investigate the impact of architectural changes across four generations of Intel Xeon processors.

We investigate the performance characteristics of a numerically enhanced scalar product (dot) kernel loop that uses the Kahan algorithm to compensate for numerical errors, and describe efficient single instruction multiple data-vectorized implementations on recent multi-core and many-core processors. Using low-level instruction analysis and the execution-cache-memory performance model, we pinpoint the relevant performance bottlenecks for single-core and thread-parallel execution and predict performance and saturation behavior. We show that the Kahan-enhanced scalar product comes at almost no additional cost compared with the naive (non-Kahan) scalar product if appropriate low-level optimizations, notably single instruction multiple data vectorization and unrolling, are applied. The execution-cache-memory model is extended appropriately to accommodate not only modern Intel multicore chips but also the Intel Xeon Phi 'Knights Corner' coprocessor and an IBM POWER8 CPU. This allows us to discuss the impact of processor features on the performance across four modern architectures that are relevant for high performance computing. Figure 7. Single-core cycles per CL versus data set size on PWR8: (a) Results for different SMT settings for naive scalar product using SP; (b) Comparison of compiler-generated naive scalar product and manual SIMD Kahan enhanced scalar product using SMT-8. The horizontal lines are ECM model predictions.

Abstract-This paper presents a practical supervised band selection procedure for airborne imaging spectrometers and Maximum Likelihood classification (MLC) as data application. The output band set is optimal in band location, width and number regarding the MLC accuracy of the classification task. The supervised algorithm is based on feature selection and requires a user-defined class set. For two given semi-natural vegetation data and class sets, the selected band sets performed superior to established vegetation band sets used in current satellite and airborne sensors, most noticeably for the first few bands. The algorithm was implemented in IDL TM /ENVI TM . It may also be used for feature selection, the generation of class-discriminate colour composites, the prioritization of already existing band sets, and the determination of the intrinsic discriminant dimensionality of the data set.

In 1993 synthetic aperture radar (SAR) interferometry (InSAR) was introduced to the wider remote sensing community with the publication of the interferogram depicting the ground deformation caused by the Landers earthquake. Although the power of interferometry was demonstrated, the conventional technique has not always been applicable in all operational scenarios. Over the last few years, however, a number of technical developments have emerged that provide a higher precision of motion rates, the extraction of specific motion histories, and precise targeting. This paper examines uses of differential SAR interferometry (DifSAR) for monitoring geohazards. Limitations of DifSAR will be discussed: lack of coherence, atmospheric refraction and targeting. It will be shown how some of these limitations can be overcome with persistent scatterer interferometry (PSI), which detects slow ground motion with annual rates of as little as a few millimetres, reconstructing a motion history based on the European Space Agency's SAR image archive. The technique permits the estimation and removal of the atmospheric phase, achieving higher accuracies than DifSAR. PSI relies on the availability of pre-existing ground features that strongly and persistently reflect back the signal from the satellite. However, in highly vegetated regions, PSI may not be applicable because of the lack of natural scatterers. To ensure motion measurement of the ground or structures at targeted locations, the NPA Group is developing InSAR using artificial radar reflectors, such as Corner Reflectors (CRs) or Compact Active Transponders (CATs). Both reflector types are still undergoing validation tests, but results show a high phase stability in both cases.

AREVA has developed a new coupled neutronics/thermal-hydraulics code system, ARCADIA®. It makes use of modern computing resources to enable more realistic reactor analysis as improved understanding of nuclear reactor behavior is the basis for efficient margin management, i.e. optimization of safety and performance. One of the principal components of this new system is the core simulator, ARTEMIS™. The purpose of this paper is to recall its features, present the latest developments and give a summary of the validation tests.