We present Calyx, a new intermediate language (IL) for compiling high-level programs into hardware designs. Calyx combines hardware-like structural with software-like control flow representation loops and conditionals. This split enables class of hardware-focused optimizations that require both information which are crucial programming models design. The compiler lowers constructs using finite-state machines generates synthesizable descriptions. have implemented in an optimizing translates...

Field-programmable gate arrays (FPGAs) provide an opportunity to co-design applications with hardware accelerators, yet they remain difficult program. High-level synthesis (HLS) tools promise raise the level of abstraction by compiling C or C++ accelerator designs. Repurposing legacy software languages, however, requires complex heuristics map imperative code onto structures. We find that black-box in HLS can be unpredictable: changing parameters program should improve performance...

Embedded applications extract the best power-performance trade-off from digital signal processors (DSPs) by making extensive use of vectorized execution. Rather than handwriting many customized kernels these use, DSP engineers rely on auto-vectorizing compilers to quickly produce effective code. Building is a large and error-prone investment, each new architecture or application-specific ISA customization must repeat this effort derive high-performance compiler.

Summary This paper describes an efficient numerical scheme for nonisothermal compositional flow coupled to chemistry. An iterative implicit-pressure/explicit-composition (IMPEC) method is applied solve the problem using a volume-balance-convergence criterion. A backward-Euler mixed finite-element (FEM) with lowest-order RT0 elements pressure equation, and component local mass-preserving explicit used update concentrations. Chemical reactions are solved Runge-Kutta (RK)...

Abstract This paper describes an efficient numerical scheme for non-isothermal compositional flow coupled to chemistry. An iterative IMPEC method is applied solve the problem using a volume balance convergence criterion. A backward Euler mixed FEM with lowest order RT0 elements pressure equation and component local mass preserving explicit used update concentrations. Chemical reactions are solved Runge-Kutta ODE integration schemes. higher Godunov thermal advection conduction, respectively,...

This paper presents an efficient numerical scheme for multiphase compositional flow coupled with subsurface heat transfer. The equations are first presented followed by a brief discussion of the equation state (EOS) and description two-phase flash algorithm. An implicit-pressure, explicit-concentrations (IMPEC) sequential algorithm is then applied iteratively to enforce non-linear volume balance saturation) constraint. pressure solved using mixed FEM, while concentrations updated consistent...

Summary This paper presents a multiblock-discretization method—the enhanced-velocity mixed-finite-element method (EVMFEM) (Wheeler et al. 2002)—for coupled multiphase flow and reactive-species-transport modeling in porous-media applications. The provides local mass balance continuous approximation of fluxes across interfaces elements subdomains. It can treat nonmatching grids, allowing for flexible choice grid refinements. Further, by distributing the blocks among processors such that each...

Abstract This paper presents multiblock methods for coupled multiphase flow and species transport modeling in porous media applications. These provide local mass conservation a continuous approximation of fluxes across inter-element faces sub-domain (inter-block) interfaces can treat non-matching grids, allowing flexible choice grid refinements. Furthermore, they lend themselves naturally to parallel implementations multiphysics, multinumerics, multiscale applications transport. The briefly...

This paper presents a novel framework for history matching using the concept of simulation-based optimization with guided search sampling, multiscale resolution and incremental metamodel (surrogate model) generation, aimed to mitigate computational burden large-scale matching.<br><br>The initial stage consists treatment permeability field through successive wavelet transformations. The coarsest grid which represents highly constrained parameter space, is sampled aid...

The main objective of the present work is to numerically determine how sensor information may aid in reducing ill‐posedness associated with permeability estimation via 4‐D seismic history matching. These sensors are assumed provide timely pressures, concentrations and fluid velocities at given locations a reliable fashion. This incorporated into an function that additionally includes production components mismatched between observed predicted data. In order efficiently perform large‐scale...

We present a lock-free, linearizable, and NUMA-aware data structure that implements sets, maps, priority queue abstract types (ADTs), based on using thread-local, sequential maps are used to "jump" suitable positions in linearizable variant of skip graph. Our graph is suitably constrained height subjected partition scheme reduces contention increases NUMA locality. developed an additional variant, which we call sparse graph, causes our thread-local as well shared become more sparse. Compared...

The main objective of the present work is to propose and evaluate a learning computational engine for history matching, which based on hybrid multilevel search methodology. According this methodology, parameter space globally explored sampled by simultaneous perturbation stochastic approximation (SPSA) algorithm at given resolution level. This estimation followed further analysis using neural evaluating sensitiveness function with respect variations each individual model in vicinity...

Geological media exhibit permeability fields and porosities that differ by several orders of magnitude across highly varying length scales. Computational methods used to model flow through such should be capable treating rough coefficients grids. Further, the adherence these basic physical properties as local mass balance continuity fluxes is great importance. Both discontinuous Galerkin (DG) mixed finite element (MFE) satisfy can accurately treat The appropriate choice models numerical...

High-performance simulations and parallel frameworks often rely on highly scalable, concurrent data structures for system scalability. With an increased availability of NUMA architectures, we present a technique to promote NUMA-aware parallelism inside structure, bringing significant quantitative qualitative improvements locality, as well reduced contention synchronized memory accesses. Our architecture is based data-partitioned, skip graph indexed by thread-local sequential maps. We...

Integral and expansion techniques for Fourier transforming position-space wave functions containing interelectron coordinates are presented. Of the two techniques, integral methods more powerful as exponential terms can be handled.

Seismic history matching has been used to reduce uncertainty and increase the accuracy in reservoir characterization. In this paper we propose a joint inversion scheme for quantitative petrophysics characterization inside fluid flow imaging, which integrates as many data sources available, such time‐lapse seismic data, production sensor information. We demonstrate that integrated framework leads accurate parameter estimation incorporation of does help speed up convergence process. addition,...

GexSi1−x/Si heteroepitaxial thin films have been grown using the low-temperature remote plasma-enhanced chemical vapor deposition (RPCVD) approach, in which substrate is kept from glow discharge, and an Ar plasma employed to indirectly activate reactant gases (SiH4 GeH4) drive reactions. Secondary ion mass spectroscopy (SIMS), plan-view cross-sectional transmission electron microscopy (TEM), situ reflection high-energy diffraction (RHEED) analyze with different Ge mole fractions thicknesses....