A comparative study of GPU programming models and architectures using neural networks

Recently, General Purpose Graphical Processing Units (GP-GPUs) have been identified as an intriguing technology to accelerate numerous data-parallel algorithms. Several GPU architectures and programming models are beginning to emerge and establish their niche in the High-Performance Computing (HPC) community. New massively parallel architectures such as the Nvidia??s Fermi and AMD/ATi??s Radeon pack tremendous computing power in their large number of multiprocessors. Their performance is unleashed using one of the two GP-GPU programming models: Compute Unified Device Architecture (CUDA) and Open Computing Language (OpenCL). Both of them offer constructs and features that have direct bearing on the application runtime performance. In this paper, we compare the two GP-GPU architectures and the two programming models using a two-level character recognition network. The two-level network is developed using four different Spiking Neural Network (SNN) models, each with different ratios of computation-to-communication requirements. To compare the architectures, we have chosen the two extremes of the SNN models for implementation of the aforementioned two-level network. An architectural performance comparison of the SNN application running on Nvidia??s Fermi and AMD/ATi??s Radeon is done using the OpenCL programming model exhausting all of the optimization strategies plausible for the two architectures. To compare the programming models, we implement the two-level network on Nvidia??s Tesla C2050 based on the Fermi architecture. We present a hierarchy of implementations, where we successively add optimization techniques associated with the two programming models. We then compare the two programming models at these different levels of implementation and also present the effect of the network size (problem size) on the performance. We report significant application speed-up, as high as 1095× for the most computation intensive SNN neuron model, against a serial implementation on the Intel Core 2 Quad host. A comprehensive study presented in this paper establishes connections between programming models, architectures and applications.     

Ferromagnetic mixed tribridged dinuclear copper(II) complex [Cu2(OAc)2(OH)(dpa)2]PF6 · H2O (dpa = 2,2′-dipyridylamine): 3D superstructure based on hydrogen bond and π…π interaction

A new dinuclear complex [Cu₂(OAc)₂(OH)(dpa)₂] PF₆ · H₂O (1) is prepared and structurally and magneto-structurally characterized. The monocationic core contains one acetate in familiar bidentate η¹:η¹:μ₂-bridge and another in the rare monoatomic bridge along with one hydroxo intermediary. 1 packs through N–H…O and O–H…O hydrogen bonds and π…π interaction resulting a 3D supramolecular continuum and displays high-energy intraligand ¹(π − π^*) fluorescence and intraligand ³(π − π^*) phosphorescence in glassy solution.     

The Role of the Prefrontal Cortex in the Maintenance of Verbal Working Memory: An Event-Related fMRI Analysis.

Neuroimaging studies have been inconclusive in characterizing the role of the prefrontal cortex (PFC) for maintaining increasingly larger amounts of information in working memory (WM). To address this question, the authors collected event-related functional MRI data while participants performed an item-recognition task in which the number of to-be-remembered letters was parametrically modulated. During maintenance of information in WM, the dorsolateral and the ventrolateral PFC exhibited linearly increasing activation in response to increasing WM load. Prefrontal regions could not be distinguished from one another on the basis of load sensitivity, but the dorsolateral PFC had stronger functional connectivity with the parietal and motor cortex than the ventrolateral PFC. These results suggest an increasingly important role for the PFC in actively maintaining information as the amount of that information increases. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A study on fitness-for-service assessment for crack-like defects and corrosion in nuclear reactor pressure tubes

This paper looks forward to thoroughly assess zirconium alloy (Zr-2) pressure tubes used in nuclear power reactors for crack like and corrosion type defects. It emphasizes on the use of reliability concept for determining the probability of failure of zirconium alloy pressure tubes based on deuterium ingress. The in depth analysis carried out to calculate the reliability index for the above two problems is based on FORM approach. A steady state corrosion rate is assumed to estimate the growth in the dimensions of corrosion defects. The methodology suggested in this paper is useful for designers and plant engineers in timely detection of the cracks and leaks in the pressure tubes used in nuclear reactors and initiate suitable corrective actions based on their criticality.     

Statistical motion model based on the change of feature relationships: human gait-based recognition

We offer a novel representation scheme for view-based motion analysis using just the change in the relational statistics among the detected image features, without the need for object models, perfect segmentation, or part-level tracking. We model the relational statistics using the probability that a random group of features in an image would exhibit a particular relation. To reduce the representational combinatorics of these relational distributions, we represent them in a Space of Probability Functions (SoPF), where the Euclidean distance is related to the Bhattacharya distance between probability functions. Different motion types sweep out different traces in this space. We demonstrate and evaluate the effectiveness of this representation in the context of recognizing persons from gait. In particular, on outdoor sequences: (1) we demonstrate the possibility of recognizing persons from not only walking gait, but running and jogging gaits as well; (2) we study recognition robustness with respect to view-point variation; and (3) we benchmark the recognition performance on a database of 71 subjects walking on soft grass surface, where we achieve around 90 percent recognition rates in the presence of viewpoint variation.     

Determination of various impurities in synthetic diamond powder

Impurities in industrial synthetic diamond powder samples were analyzed by X-ray diffraction (XRD) and Flame Atomic Absorption Spectrometry (FAAS). Specimen for FAAS is required in solution form. Diamonds are chemically inert to most acids and alkalies. Carbon was removed as CO₂ on heating and estimated gravimetrically. The remaining residue was fused with di-lithium tetraborate and dissolved in nitric acid. Impurities such as Si, Al, Fe, Ca, Mg, W, Na, Co and Ni were then determined by FAAS. Crystalline phases of major impurities were identified by XRD.     

Design of robust model-based controllers via parametric programming

In this paper a method is presented for deriving the explicit robust model-based optimal control law for constrained linear dynamic systems. The controller is derived off-line via parametric programming before any actual process implementation takes place. The proposed control scheme guarantees feasible operation in the presence of bounded input uncertainties by (i) explicitly incorporating in the controller design stage a set of feasibility constraints and (ii) minimizing the nominal performance, or the expectation of the performance over the uncertainty space. An extension of the method to problems involving target point tracking in the presence of persistent disturbances is also discussed. The general concept is illustrated with two examples.     

Cavitation Erosion Characteristics of Nitrocarburized and HPDL-Treated Martensitic Stainless Steels

This article deals with plasma ion-nitrocarburising and high power diode laser (HPDL) surface treatment of 13Cr4Ni and X10CrNiMoV1222 martensitic stainless steels to enhance their cavitation erosion resistance. These steels are commonly used in hydro turbines and boiler feed pumps. These treated steels have been evaluated for cavitation erosion resistance and it has been observed that the plasma ion-nitrocarburising process has significantly enhanced the cavitation erosion resistance as compared to untreated steel whereas HPDL-treated steels have shown marginal improvement. This is due to formation of high hardness nitrides during nitrocarburising and formation of moderate hardness martensitic phase due to rapid heating and cooling rates involved in HPDL treatment. The cavitation erosion and micro-hardness data of plasma ion-nitrocarburized as well as HPDL-treated steel samples and their comparison with hard deposits such as stellite and HVOF coating form the main part of the article.     

On the polymeric foams: modeling and properties

Polymeric foams are now widely used and researched. The physical properties of polymeric foam can be related to a set of independent structural parameters or variables of the foam. Study of these variables and correlation with commercial FE packages is essential for reliable and faster product development. Some aspects of foam behavior are widely studied while some are little less, like correlation of physical unloading behavior. For example, a lot of work in the area of phenomenological constitutive modeling of uniaxial loading was done, though research in areas of unloading–reloading and their correlation still demands more attention. Increasing number of OEMs and suppliers are moving to computer simulations in the design phase to assess their future products. Hence, different parameters within FE packages play a significant role and also affect the results. Appropriate use of these parameters will narrow down error band and automatically reduce the cycle time and development cost. This brief review is expected to set the perspective for major research work done so far in terms of FE modeling correlation and constitutive modeling of polymeric foam vis-a-vis to its properties.