CU++: an object oriented framework for computational fluid dynamics applications using graphics processing units

The application of graphics processing units (GPU) to solve partial differential equations is gaining popularity with the advent of improved computer hardware. Various lower level interfaces exist that allow the user to access GPU specific functions. One such interface is NVIDIA’s Compute Unified Device Architecture (CUDA) library. However, porting existing codes to run on the GPU requires the user to write kernels that execute on multiple cores, in the form of Single Instruction Multiple Data (SIMD). In the present work, a higher level framework, termed CU++, has been developed that uses object oriented programming techniques available in C++ such as polymorphism, operator overloading, and template meta programming. Using this approach, CUDA kernels can be generated automatically during compile time. Briefly, CU++ allows a code developer with just C/C++ knowledge to write computer programs that will execute on the GPU without any knowledge of specific programming techniques in CUDA. This approach is tremendously beneficial for Computational Fluid Dynamics (CFD) code development because it mitigates the necessity of creating hundreds of GPU kernels for various purposes. In its current form, CU++ provides a framework for parallel array arithmetic, simplified data structures to interface with the GPU, and smart array indexing. An implementation of heterogeneous parallelism, i.e., utilizing multiple GPUs to simultaneously process a partitioned grid system with communication at the interfaces using Message Passing Interface (MPI) has been developed and tested.     

An octree-based,cartesian navier–stokes solver for modern cluster architectures

The Journal of Supercomputing - Adaptive Cartesian mesh approaches have proven useful for multi-scale applications where particular features can be finely resolved within a large solution domain....     

Application and validation of incrementally complex models for wind turbine aerodynamics,isolated wind turbine in uniform inflow conditions

A comparison of several incrementally complex methods for predicting wind turbine performance, aeroelastic behavior, and wakes is provided. Depending on a wind farm's design, wake interference can cause large power losses and increased turbulence levels within the farm. The goal is to employ modeling methods to reach an improved understanding of wake effects and to use this information to better optimize the layout of new wind farms. A critical decision faced by modelers is the fidelity of the model that is selected to perform simulations. The choice of model fidelity can affect the accuracy, but will also greatly impact the computational time and resource requirements for simulations. To help address this critical question, three modeling methods of varying fidelity have been developed side by side and are compared in this article. The models from low to high complexity are as follows: a blade element‐based method with a free‐vortex wake, an actuator disc‐based method, and a full rotor‐based method. Fluid/structure interfaces are developed for the aerodynamic modeling approaches that allow modeling of discrete blades and are then coupled with a multibody structural dynamics solver in order to perform an aeroelastic analysis. Similar methods have individually been tested by researchers, but we suggest that by developing a suite of models, they can be cross‐compared to grasp the subtleties of each method. The modeling methods are applied to the National Renewable Energy Laboratory Phase VI rotor to predict the turbine aerodynamic and structural loads and then also the wind velocities in the wake. The full rotor method provides the most accurate predictions at the turbine and the use of adaptive mesh refinement to capture the wake to 20 radii downstream is proven particularly successful. Though the full rotor method is unmatched by the lower fidelity methods in stalled conditions and detailed prediction of the downstream wake, there are other less complex conditions where these methods perform as accurately as the full rotor method. Copyright © 2014 John Wiley & Sons, Ltd.     

Study of coupled thermal electric behavior of compliant micro-spring interconnects for next generation probing applications

Advances in integrated circuit fabrication have given rise to a need for an innovative, inexpensive, yet reliable probing technology with ultra-fine pitch capability. Flexible micro-spring structures that can exceed the probing needs of the next-generation microelectronic devices have been developed. Highly compliant cantilevered springs have been fabricated at pitches as small as 6/spl mu/m. These micro-springs are designed to accommodate topological variation in probing surfaces while flexing within the elastic regime. Coupled thermal-electric numerical models have been developed to understand the thermal contours and current density developed across these springs. Based on the models and experiments, it is seen that the electrical resistance of the probe spring under study will be less than 1 /spl Omega/. Also, it is seen that the maximum temperature due to Joule heating is localized near the tip of the probe and can be about 93/spl deg/C above the ambient temperature, when temperature dependent bulk material properties are used. Optimization of the spring geometry to reduce this maximum temperature is outlined. In addition, the role of scale effects on the thermal conductivity of the spring material is studied. Based on the work, it can be said that it is possible to design micro-contact springs for probing applications such that the electrical resistance and the temperature increase during probing will be within acceptable limits.     

System-level reliability assessment of mixed-signal convergent microsystems

The next-generation convergent microsystems, based on system-on-package (SOP) technology, require up-front system-level design-for-reliability approaches and appropriate reliability assessment methodologies to guarantee the reliability of digital, optical, and radio frequency (RF) functions, as well as their interfaces. Systems approach to reliability requires the development of: i) physics-based reliability models for various failure mechanisms associated with digital, optical, and RF Functions, and their interfaces in the system; ii) design optimization models for the selection of suitable materials and processing conditions for reliability, as well as functionality; and iii) system-level reliability models understanding the component and functional interaction. This paper presents the reliability assessment of digital, optical, and RF functions in SOP-based microsystems. Upfront physics-based design-for-reliability models for various functional failure mechanisms are presented to evaluate various design options and material selection even before the prototypes are made. Advanced modeling methodologies and algorithms to accommodate material length scale effects due to enhanced system integration and miniaturization are presented. System-level mixed-signal reliability is discussed thorough system-level reliability metrics relating component-level failure mechanisms to system-level signal integrity, as well as statistical aspects.     

Development of indigenous technology for production of titanium sponge by the Kroll process

Titanium has emerged as a major structural metal for a wide range of industrial applications due to its attractive engineering properties. India has a large and rich reserve base for this metal in the beach sands of the eastern and southern regions with well established production facilities for their separation into individual minerals. Research and Development activities for establishing the metal production technology have been underway in the country for over two decades. The Defence Metallurgical Research Laboratory, Hyderabad, has already demonstrated the metal production technology by the conventional Kroll process on 2000 kg/batch scale and is now all set for demonstrating the same by the more advanced, energy efficient combined process route on 4000 kg/batch scale. The paper reviews the R & D efforts undertaken so far in the field of metal extraction with emphasis on the current status of this developmental activity at DMRL.     

A family of phase-variable restriction enzymes with differing specificities generated by high-frequency gene rearrangements

The hsd genes of Mycoplasma pulmonis encode restriction and modification enzymes exhibiting a high degree of sequence similarity to the type I enzymes of enteric bacteria. The S subunits of type I systems dictate the DNA sequence specificity of the holoenzyme and are required for both the restriction and the modification reactions. The M. pulmonis chromosome has two hsd loci, both of which contain two hsdS genes each and are complex, site-specific DNA inversion systems. Embedded within the coding region of each hsdS gene are a minimum of three sites at which DNA inversions occur to generate extensive amino acid sequence variations in the predicted S subunits. We show that the polymorphic hsdS genes produced by gene rearrangement encode a family of functional S subunits with differing DNA sequence specificities. In addition to creating polymorphisms in hsdS sequences, DNA inversions regulate the phase-variable production of restriction activity because the other genes required for restriction activity (hsdR and hsdM) are expressed only from loci that are oriented appropriately in the chromosome relative to the hsd promoter. These data cast doubt on the prevailing paradigms that restriction systems are either selfish or function to confer protection from invasion by foreign DNA.     

Computational personality recognition in social media

A variety of approaches have been recently proposed to automatically infer users’ personality from their user generated content in social media. Approaches differ in terms of the machine learning algorithms and the feature sets used, type of utilized footprint, and the social media environment used to collect the data. In this paper, we perform a comparative analysis of state-of-the-art computational personality recognition methods on a varied set of social media ground truth data from Facebook, Twitter and YouTube. We answer three questions: (1) Should personality prediction be treated as a multi-label prediction task (i.e., all personality traits of a given user are predicted at once), or should each trait be identified separately? (2) Which predictive features work well across different on-line environments? and (3) What is the decay in accuracy when porting models trained in one social media environment to another?     

Numerical modeling of interfacial delamination propagation in anovel peripheral array package

Interfacial delamination, due to the presence of dissimilar material systems, is one of the primary concerns in electronic package designs. The mismatch in the coefficient of thermal expansion between the different layers in the package can generate high interfacial stresses upon heating or cooling of the structure during fabrication, assembly, or in field use. These stresses, if sufficiently large, can compromise the adhesive integrity of the interface. The propagation of the resulting delamination along an interface can degrade or completely destroy the functionality of the system. The focus of this study is to examine the potential for interfacial delamination propagation in current and future versions of a novel peripheral array package. Two-dimensional (2-D) and three-dimensional (3-D) numerical models were constructed of this package with cracks embedded along a critical interface. The energy release rate associated with interfacial fracture was determined by employing the global energy balance and the crack closure technique. The fracture mode mixity was determined using the crack surface displacement method. These critical fracture parameters were compared with experimentally determined interfacial fracture toughness data to determine the possibility of delamination growth. A material parametric study was also completed using the numerical models with pre-existing delaminations to identify material property trends that would lower the potential for failure. Also, the effect of plastic behavior on interfacial crack growth was studied through J-integral calculations