Heterogeneous Computer Architectures Research Articles

A parallel and performance portable implementation of a full-field crystal plasticity model

We have developed a parallel implementation of an Elasto-Viscoplastic Fast Fourier Transform-based (EVPFFT) micromechanical solver to enable computationally efficient crystal plasticity modeling for polycrystalline materials. Our primary focus lies in achieving performance portability, allowing a single EVPFFT implementation to run optimally on various homogeneous architectures, including multi-core Central Processing Units (CPUs), as well as on heterogeneous computer architectures comprising multi-core CPUs and Graphics Processing Units (GPUs) from different vendors. To accomplish this goal, we have leveraged MATAR, a C++ software library that simplifies the creation and utilization of multidimensional dense or sparse matrix and array data structures. These data structures are designed to be portable across diverse architectures through the use of Kokkos, a performance-portable library. Additionally, we have employed the Message Passing Interface (MPI) to efficiently distribute the computational workload among processors. The heFFTe (Highly Efficient FFT for Exascale) library is used to facilitate the performance portability of the fast Fourier transforms (FFTs) computation. The computational performance of EVPFFT is evaluated and presented in terms of parallel scalability and simulation runtime on different high-performance computing (HPC) architectures. The utility of the developed framework to efficiently simulate the micro-mechanical fields in polycrystalline microstructures in engineering applications is discussed. Program summaryProgram Title: EVPFFTCPC Library link to program files:https://doi.org/10.17632/2k8579fyyv.1Developer's repository link:https://github.com/lanl/FierroLicensing provisions: BSD 3-Clause LicenseProgramming language: C++External routines/libraries: MPI, Kokkos, MATAR, HeFFTe, HDF5Nature of problem: EVPFFT is a crystal plasticity code designed to compute micro-mechanical fields within a polycrystalline representative volume element (RVE) and predict the macroscale response of the RVE.Solution method: EVPFFT uses the periodic Green's function method in Fourier space to solve the field equations of static stress equilibrium in a periodic spatial domain.

Parallel Static Learning Toward Heterogeneous Computing Architectures

Parallel Static Learning Toward Heterogeneous Computing Architectures

Accelerating Coupled-Cluster Calculations with GPUs: An Implementation of the Density-Fitted CCSD(T) Approach for Heterogeneous Computing Architectures Using OpenMP Directives.

An algorithm is presented for the coupled-cluster singles, doubles, and perturbative triples correction [CCSD(T)] method based on the density fitting or the resolution-of-the-identity (RI) approximation for performing calculations on heterogeneous computing platforms composed of multicore CPUs and graphics processing units (GPUs). The directive-based approach to GPU offloading offered by the OpenMP application programming interface has been employed to adapt the most compute-intensive terms in the RI-CCSD amplitude equations with computational costs scaling as , , and (where NO and NV denote the numbers of correlated occupied and virtual orbitals, respectively) and the perturbative triples correction to execute on GPU architectures. The pertinent tensor contractions are performed using an accelerated math library such as cuBLAS or hipBLAS. Optimal strategies are discussed for splitting large data arrays into tiles to fit them into the relatively small memory space of the GPUs, while also minimizing the low-bandwidth CPU-GPU data transfers. The performance of the hybrid CPU-GPU RI-CCSD(T) code is demonstrated on pre-exascale supercomputers composed of heterogeneous nodes equipped with NVIDIA Tesla V100 and A100 GPUs and on the world's first exascale supercomputer named "Frontier", the nodes of which consist of AMD MI250X GPUs. Speedups within the range 4-8× relative to the recently reported CPU-only algorithm are obtained for the GPU-offloaded terms in the RI-CCSD amplitude equations. Applications to polycyclic aromatic hydrocarbons containing 16-66 carbon atoms demonstrate that the acceleration of the hybrid CPU-GPU code for the perturbative triples correction relative to the CPU-only code increases with the molecule size, attaining a speedup of 5.7× for the largest circumovalene molecule (C66H20). The GPU-offloaded code enables the computation of the perturbative triples correction for the C60 molecule using the cc-pVDZ/aug-cc-pVTZ-RI basis sets in 7 min on Frontier when using 12,288 AMD GPUs with a parallel efficiency of 83.1%.

Accelerating fully resolved simulation of particle-laden flows on heterogeneous computer architectures

Accelerating fully resolved simulation of particle-laden flows on heterogeneous computer architectures

Accelerating Code Assembly: Exploiting Heterogeneous Computing Architectures for Optimization

Accelerating Code Assembly: Exploiting Heterogeneous Computing Architectures for Optimization

High-Performance Computing Probabilistic Fracture Mechanics Implementation for Gas Turbine Rotor Disks on Distributed Architectures Including Graphics Processing Units

Abstract We present an efficient Monte Carlo-based probabilistic fracture mechanics simulation implementation for heterogeneous high-performance computing (HPC) architectures including central processing units (CPUs) and graphics processing units (GPUs). The specific application focuses on large heavy-duty gas turbine rotor components for the energy sector. A reliable probabilistic risk quantification requires the simulation of millions to billions of Monte Carlo (MC) samples. We apply a modified Runge–Kutta algorithm in order to solve numerically the fatigue crack growth for this large number of cracks for varying initial crack sizes, locations, material, and service conditions. This compute intensive simulation has already been demonstrated to perform efficiently and scalable on parallel and distributed HPC architectures including hundreds of CPUs utilizing the message passing interface (MPI) paradigm. In this work, we go a step further and include GPUs in our parallelization strategy. We develop a load distribution scheme to share one or more GPUs on compute nodes distributed over a network. We detail technical challenges and solution strategies in performing the simulations on GPUs efficiently. We show that the key computation of the modified Runge–Kutta integration step speeds up over two orders of magnitude on a typical GPU compared to a single threaded CPU. This is supported by our use of GPU textures for efficient interpolation of multidimensional tables utilized in the implementation. We demonstrate weak and strong scaling of our GPU implementation, i.e., that we can efficiently utilize a large number of GPUs/CPUs in order to solve for more MC samples, or reduce the computational turn-around time, respectively. On seven different GPUs spanning four generations, the presented probabilistic fracture mechanics simulation tool ProbFM achieves a speed-up ranging from 16.4× to 47.4× compared to single threaded CPU implementation.

Porting hypre to heterogeneous computer architectures: Strategies and experiences

Linear systems are occurring in many applications, and solving them can take a large amount of the total simulation time. The high performance library hypre provides a variety of interfaces and linear solvers, including various multigrid methods, that have achieved good scalability on a variety of homogeneous parallel computer architectures. Heterogeneous architectures with nodes that have both CPUs and accelerators provide new challenges, since they require more fine-grained parallelism and reduced data movement between different memories on a single node as well as across nodes. We will discuss our experiences and strategies to port hypre to heterogeneous computers with accelerators, including the design of a new memory model, the use of abstractions, the BoxLoop macros in the structured and semi-structured interfaces, and the restructuring of algebraic multigrid (AMG) into modular components. We present numerical experiments comparing CPU and GPU performance for several test problems.

A Platform Programming Paradigm for Heterogeneous Systems Integration

To cope with growing computing performance requirements, cyber-physical systems architectures are moving toward heterogeneous high-performance computer architectures and networks. Such architectures, however, incur intricate side effects that challenge traditional software design and integration. The programming paradigm can take a key role in mastering software design, as experience in automotive design demonstrates. To cope with the integration challenge, this industry has started introducing a programming paradigm that efficiently preserves application data flow under platform integration and changes with minimum performance loss. This article will revisit this paradigm that is currently used for lock-free multicore programming and explain its extension to the system level. It will then explore its application to two important developments in industrial design. This article will conclude with an evaluation of its properties, its overhead, and its application toward a robust design process.

Research on highly parallel embedded control system design and implementation method

The pace of technology's advancements is ever-increasing and intelligent systems, such as those found in robots and vehicles, have become larger and more complex. These intelligent systems have a heterogeneous structure, comprising a mixture of modules such as artificial intelligence (AI) and powertrain control modules that facilitate large-scale numerical calculation and real-time periodic processing functions. Information technology expert Professor Masato Edahiro, from the Graduate School of Informatics at the Nagoya University in Japan, explains that concurrent advances in semiconductor research have led to the miniaturisation of semiconductors, allowing a greater number of processors to be mounted on a single chip, increasing potential processing power. 'In addition to general-purpose processors such as CPUs, a mixture of multiple types of accelerators such as GPGPU and FPGA has evolved, producing a more complex and heterogeneous computer architecture,' he says. Edahiro and his partners have been working on the eMBP, a model-based parallelizer (MBP) that offers a mapping system as an efficient way of automatically generating parallel code for multi- and many-core systems. This ensures that once the hardware description is written, eMBP can bridge the gap between software and hardware to ensure that not only is an efficient ecosystem achieved for hardware vendors, but the need for different software vendors to adapt code for their particular platforms is also eliminated.

Physics-Based-Adaptive Plasma Model for High-Fidelity Numerical Simulations

A physics-based-adaptive plasma model and an appropriate computational algorithm are developed to numerically simulate plasma phenomena in high fidelity. %The physics-based-adaptive plasma model is dynamically refined based on the local conditions to provide uniform model fidelity throughout the domain at all times of the simulation. The physics-based-adaptive plasma model can be dynamically refined based on the local plasma conditions to increase model fidelity uniformity throughout the domain at all times of the simulation. %The adaptive plasma model uses continuum representations of the plasma, which include a kinetic Boltzmann model for the highest fidelity, multi-fluid plasma models (13N-moment and 5$N$-moment), and single-fluid MHD models for the lowest fidelity. The adaptive plasma model uses continuum representations of the plasma, which include a kinetic Vlasov model for the highest fidelity, multi-fluid 5$N$-moment plasma model, and single-fluid MHD model for the lowest fidelity. The models include evolution equations for the electromagnetic fields, electron species, ion species, and neutral species. A nodal discontinuous Galerkin finite element method is implemented and is coupled with various implicit and explicit Runge-Kutta methods. Various model coupling techniques are investigated for a 5$N$-moment multi-fluid models with a Vlasov-Maxwell model, and a 5$N$-moment two-fluid model with an MHD model. Continuum plasma models using consistent normalizations and identical spatial representations provide straightforward and accurate coupling between the models. %The solution approach offers the potential for high-order accuracy and computational efficiency. The solution approach offers high-order accuracy and computational efficiency. Target compute platforms are heterogeneous computer architectures using a compute model that minimizes data movement.

Numerical and Experimental Study of Bubble Dynamics in Contact with a Solid Surface

A complex numerical and experimental method is proposed for studying 3D dynamics of a bubble contacting with a surface in the presence of an acoustic field. The numerical approach is based on the boundary element method for potential flows, which is most efficient for solving the problems in a 3D formulation. The use of heterogeneous computer architectures consisting of central graphic processors and becoming more and more popular makes it possible to increase the scale of the problem and sufficiently reduce the calculation time. The mesh destabilization problems are solved using a spherical filter. To describe the contact line dynamics, a semi-empirical law of motion is used. The experimental method is based on high-speed recording and optical microscopy. An air bubble contacts with the inner surface of an experimental cell made from acrylic glass and filled with distilled water. The acoustic field in the cell generated by a disk-shaped acoustic radiator is measured using a hydrophone. The behavior of the bubble contacting with a hydrophillic surface is considered for the cases of a fixed or moving contact line. The shape and volume oscillations of the bubble are investigated. The results of numerical simulations agree qualitatively with the experimental data.

Scalable near-repeat and event chain calculations over heterogeneous computer architecture and systems

As a case of space–time interaction, near-repeat calculation indicates that when an event takes place at a certain location, its immediate geographical surroundings would face an increased risk of experiencing subsequent events within a fairly short period of time. This paper presents an exploratory study that extends the investigation of the near-repeat phenomena to a series of space–time interaction, namely event chain calculation. Existing near-repeat tools can only deal with a limited amount of data due to computation constraints, let alone the event chain analysis. By deploying the modern accelerator technology and hybrid computer systems, this study demonstrates that large-scale near-repeat calculation or event chain analysis can be partially resolved through high-performance computing solutions to advance such a challenging statistical problem in both spatial analysis and crime geography.

A meta-EDSL for distributed web applications

We present a domain-specific language for constructing and configuring web applications distributed across any number of networked, heterogeneous systems. Our language is embedded in Haskell, provides a common framework for integrating components written in third-party EDSLs, and enables type-safe, access-controlled communication between nodes, as well as effortless sharing and movement of functionality between application components. We give an implementation of our language and demonstrate its applicability by using it to implement several important components of distributed web applications, including RDBMS integration, load balancing, and fine-grained sandboxing of untrusted third party code. The rising popularity of cloud computing and heterogeneous computer architectures is putting a strain on conventional programming models, which commonly assume that one application executes on one machine, or at best on one out of several identical machines. With our language, we take the first step towards a programming model better suited for a computationally multicultural future.

Comprehensive curriculum for reconfigurable heterogeneous computer architecture education

As heterogeneity in desktop processor chips was recently promised by the major manufactures, the importance of these new architecture paradigms strongly grows. Especially if programmable CPUs are combined with reconfigurable logic like it has been done in the embedded domain, the complexity to design an energy efficient and powerful system increases. Therefore, heterogeneous system platforms have to be focused even stronger in research and education. Because the design and programming of these architectures is much more complex than using standard processor solutions, it is essential to provide thorough education programs for students. Only then will the engineers of tomorrow be able to deal with the future challenges. As a consequence, the authors restructured their curriculum to especially deal with the upcoming needs. Three courses were oriented towards the common goal of developing a real-world smart camera solution utilising heterogeneous architectures. The new combination provides various synergistic benefits and evaluation results confirm, that the overall orientation of the courses is a step in the right direction. As the basic components are already available at many other universities, their example can encourage to launch similar programmes elsewhere.

Iterative load-balancing method with multigrid level relaxation for particle simulation with short-range interactions

We developed dynamic load-balancing algorithms for Particle Simulation Methods (PSM) involving short-range interactions, such as Smoothed Particle Hydrodynamics (SPH), Moving Particle Semi-implicit method (MPS), and Discrete Element method (DEM). These are needed to handle billions of particles modeled in large distributed-memory computer systems. Our method utilizes flexible orthogonal domain decomposition, allowing the sub-domain boundaries in the column to be different for each row. The imbalances in the execution time between parallel logical processes are treated as a nonlinear residual. Load-balancing is achieved by minimizing the residual within the framework of an iterative nonlinear solver, combined with a multigrid technique in the local smoother. Our iterative method is suitable for adjusting the sub-domain frequently by monitoring the performance of each computational process because it is computationally cheaper in terms of communication and memory costs than non-iterative methods. Numerical tests demonstrated the ability of our approach to handle workload imbalances arising from a non-uniform particle distribution, differences in particle types, or heterogeneous computer architecture which was difficult with previously proposed methods. We analyzed the parallel efficiency and scalability of our method using Earth simulator and K-computer supercomputer systems.

Hybrid massively parallel fast sweeping method for static Hamilton–Jacobi equations

The fast sweeping method is a popular algorithm for solving a variety of static Hamilton–Jacobi equations. Fast sweeping algorithms for parallel computing have been developed, but are severely limited. In this work, we present a multilevel, hybrid parallel algorithm that combines the desirable traits of two distinct parallel methods. The fine and coarse grained components of the algorithm take advantage of heterogeneous computer architecture common in high performance computing facilities. We present the algorithm and demonstrate its effectiveness on a set of example problems including optimal control, dynamic games, and seismic wave propagation. We give results for convergence, parallel scaling, and show state-of-the-art speedup values for the fast sweeping method.

Taking the Human Out of the Loop: A Review of Bayesian Optimization

Big Data applications are typically associated with systems involving large numbers of users, massive complex software systems, and large-scale heterogeneous computing and storage architectures. The construction of such systems involves many distributed design choices. The end products (e.g., recommendation systems, medical analysis tools, real-time game engines, speech recognizers) thus involve many tunable configuration parameters. These parameters are often specified and hard-coded into the software by various developers or teams. If optimized jointly, these parameters can result in significant improvements. Bayesian optimization is a powerful tool for the joint optimization of design choices that is gaining great popularity in recent years. It promises greater automation so as to increase both product quality and human productivity. This review paper introduces Bayesian optimization, highlights some of its methodological aspects, and showcases a wide range of applications.

A hybrid parallel cellular automata model for urban growth simulation over GPU/CPU heterogeneous architectures

As an important spatiotemporal simulation approach and an effective tool for developing and examining spatial optimization strategies (e.g., land allocation and planning), geospatial cellular automata (CA) models often require multiple data layers and consist of complicated algorithms in order to deal with the complex dynamic processes of interest and the intricate relationships and interactions between the processes and their driving factors. Also, massive amount of data may be used in CA simulations as high-resolution geospatial and non-spatial data are widely available. Thus, geospatial CA models can be both computationally intensive and data intensive, demanding extensive length of computing time and vast memory space. Based on a hybrid parallelism that combines processes with discrete memory and threads with global memory, we developed a parallel geospatial CA model for urban growth simulation over the heterogeneous computer architecture composed of multiple central processing units (CPUs) and graphics processing units (GPUs). Experiments with the datasets of California showed that the overall computing time for a 50-year simulation dropped from 13,647 seconds on a single CPU to 32 seconds using 64 GPU/CPU nodes. We conclude that the hybrid parallelism of geospatial CA over the emerging heterogeneous computer architectures provides scalable solutions to enabling complex simulations and optimizations with massive amount of data that were previously infeasible, sometimes impossible, using individual computing approaches.

Power-Aware Job Scheduling on Heterogeneous Multicore Architectures

This paper presents a power-aware scheduling algorithm based on efficient distribution of the computing workload to the resources on heterogeneous CPU-GPU architectures. The scheduler manages the resources of several computing nodes with a view to reducing the peak power. The algorithm can be used in concert with adjustable power state software services in order to further reduce the computing cost during high demand periods. Although our study relies on GPU workloads, the approach can be extended to other heterogeneous computer architectures. The algorithm has been implemented in a real CPU-GPU heterogeneous system. Experiments prove that the approach presented reduces peak power by 10 percent compared to a system without any power-aware policy and by up to 24 percent with respect to the worst case scenario with an execution time increase in the range of 2 percent. This leads to a reduction in the system and service costs.

The Impact of Heterogeneous Computer Architectures on Computational Physics

Guest editor Barry I. Schneider describes what went into creating this special issue of CiSE magazine, which focuses on high-performance computing and its applications to current subfields of physics and chemistry.