Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Computational Load Balancing”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7

Efficient Sparse Matrix-Matrix Multiplication on Heterogeneous High Performance Systems

The efficient implementation of sparse matrix-matrix multiplications on high performance parallel machines poses several challenges: large size of input matrices, compressed representation, density of the output matrices, partitioning and load balancing of matrices that present parts with large differences in density and, thus, in computation times. In this paper we show how, starting from the requirements of such application, we developed a framework that allows its efficient implementation on heterogeneous clusters. We introduce a task based programming model and a runtime supported execution model which provides dynamic load balancing on clusters composed by CPUs and GPUs, allowing better utilization of the system while easing the handling of sparse matrices. Our results show that our solution, which co-designs the application together with the programming model and the runtime system, is able to obtain significant speed-ups due to a virtually perfect load balancing with respect to other programming approaches.

97 MATHEMATICS AND COMPUTING↗

A physics-motivated Centroidal Voronoi Particle domain decomposition method

In this paper, we propose a novel domain decomposition method for large-scale simulations in continuum mechanics by merging the concepts of Centroidal Voronoi Tessellation (CVT) and Voronoi Particle dynamics (VP). The CVT is introduced to achieve a high-level compactness of the partitioning subdomains by the Lloyd algorithm which monotonically decreases the CVT energy. The number of computational elements between neighboring partitioning subdomains, which scales the communication effort for parallel simulations, is optimized implicitly as the generated partitioning subdomains are convex and simply connected with small aspect-ratios. Moreover, Voronoi Particle dynamics employing physical analogy with a tailored equation of state is developed, which relaxes the particle system towards the target partition with good load balance. Since the equilibrium is computed by an iterative approach, the partitioning subdomains exhibit locality and the incremental property. Numerical experiments reveal that the proposed Centroidal Voronoi Particle (CVP) based algorithm produces high-quality partitioning with high efficiency, independently of computational-element types. Thus it can be used for a wide range of applications in computational science and engineering.

97 MATHEMATICS AND COMPUTING↗

Efficient Load Balancing and Data Remapping for Adaptive Grid Calculations

Mesh adaption is a powerful tool for efficient unstructured- grid computations but causes load imbalance among processors on a parallel machine. We present a novel method to dynamically balance the processor workloads with a global view. This paper presents, for the first time, the implementation and integration of all major components within our dynamic load balancing strategy for adaptive grid calculations. Mesh adaption, repartitioning, processor assignment, and remapping are critical components of the framework that must be accomplished rapidly and efficiently so as not to cause a significant overhead to the numerical simulation. Previous results indicated that mesh repartitioning and data remapping are potential bottlenecks for performing large-scale scientific calculations. We resolve these issues and demonstrate that our framework remains viable on a large number of processors.

Oliker, Leonid↗

Towards scaling community detection on distributed-memory heterogeneous systems

Distributed multi-GPU systems pose significant challenges and opportunities for efficient execution of parallel applications. Graph algorithms are generally characterized by irregular memory accesses, low computation to communication ratios, and load balancing problems that are especially hard to address on multi-GPU systems. Graph community detection is an important problem in the emerging domain of graph analytics with numerous applications. In this paper, we present our ongoing work on distributed-memory multi-GPU implementation for graph community detection. Our work parallelizes the widely used (albeit serial) Louvain method on distributed multi-GPU platforms. Supported by an extensive set of experiments on a multi-GPU enabled supercomputer (OLCF Summit) and a single compute node (Nvidia DGX-2®), we demonstrate competitive performance to existing distributed-memory CPU-based implementation, and up to 6.5 better results than Nvidia RAPIDS® CUGRAPH. To the best of our knowledge, this work represents the first effort for community detection on distributed multi-GPU systems. Our approach and related findings can be extended to numerous other iterative graph algorithms on multi-GPU systems.

97 MATHEMATICS AND COMPUTING↗

FOX: A Fault Oblivious Extreme-Scale Execution Environment

The FOX project, funded under the ASCR X-stack I program, developed systems software and runtime libraries for a new approach to the data and work distribution for massively parallel, fault oblivious application execution. Our work was motivated by the premise that exascale computing systems will provide a thousand-fold increase in parallelism and a proportional increase in failure rate relative to today’s machines. To deliver the capability of exascale hardware, the systems software must provide the infrastructure to support existing applications while simultaneously enabling efficient execution of new programming models that naturally express dynamic, adaptive, irregular computation; coupled simulations; and massive data analysis in a highly unreliable hardware environment with billions of threads of execution. Our OS research has prototyped new methods to provide efficient resource sharing, synchronization, and protection in a many-core compute node. We have experimented with alternative task/dataflow programming models and shown scalability in some cases to hundreds of thousands of cores. Much of our software is in active development through open source projects. Concepts from FOX are being pursued in next generation exascale operating systems. Our OS work focused on adaptive, application tailored OS services optimized for multi → many core processors. We developed a new operating system NIX that supports role-based allocation of cores to processes which was released to open source. We contributed to the IBM FusedOS project, which promoted the concept of latency-optimized and throughput-optimized cores. We built a task queue library based on distributed, fault tolerant key-value store and identified scaling issues. A second fault tolerant task parallel library was developed, based on the Linda tuple space model, that used low level interconnect primitives for optimized communication. We designed fault tolerance mechanisms for task parallel computations employing work stealing for load balancing that scaled to the largest existing supercomputers. Finally, we implemented the Elastic Building Blocks runtime, a library to manage object-oriented distributed software components. To support the research, we won two INCITE awards for time on Intrepid (BG/P) and Mira (BG/Q). Much of our work has had impact in the OS and runtime community through the ASCR Exascale OS/R workshop and report, leading to the research agenda of the Exascale OS/R program. Our project was, however, also affected by attrition of multiple PIs. While the PIs continued to participate and offer guidance as time permitted, losing these key individuals was unfortunate both for the project and for the DOE HPC community.

97 MATHEMATICS AND COMPUTING↗

Three-Dimensional High-Order Spectral Volume Method for Solving Maxwell's Equations on Unstructured Grids

A three-dimensional, high-order, conservative, and efficient discontinuous spectral volume (SV) method for the solutions of Maxwell's equations on unstructured grids is presented. The concept of discontinuous 2nd high-order loca1 representations to achieve conservation and high accuracy is utilized in a manner similar to the Discontinuous Galerkin (DG) method, but instead of using a Galerkin finite-element formulation, the SV method is based on a finite-volume approach to attain a simpler formulation. Conventional unstructured finite-volume methods require data reconstruction based on the least-squares formulation using neighboring cell data. Since each unknown employs a different stencil, one must repeat the least-squares inversion for every cell at each time step, or to store the inversion coefficients. In a high-order, three-dimensional computation, the former would involve impractically large CPU time, while for the latter the memory requirement becomes prohibitive. In the SV method, one starts with a relatively coarse grid of triangles or tetrahedra, called spectral volumes (SVs), and partition each SV into a number of structured subcells, called control volumes (CVs), that support a polynomial expansion of a desired degree of precision. The unknowns are cell averages over CVs. If all the SVs are partitioned in a geometrically similar manner, the reconstruction becomes universal as a weighted sum of unknowns, and only a few universal coefficients need to be stored for the surface integrals over CV faces. Since the solution is discontinuous across the SV boundaries, a Riemann solver is thus necessary to maintain conservation. In the paper, multi-parameter and symmetric SV partitions, up to quartic for triangle and cubic for tetrahedron, are first presented. The corresponding weight coefficients for CV face integrals in terms of CV cell averages for each partition are analytically determined. These discretization formulas are then applied to the integral form of the Maxwell equations. All numerical procedures for outer boundary, material interface, zonal interface, and interior SV face are unified with a single characteristic formulation. The load balancing in a massive parallel computing environment is therefore easier to achieve. A parameter is introduced in the Riemann solver to control the strength of the smoothing term. Important aspects of the data structure and its effects to communication and the optimum use of cache memory are discussed. Results will be presented for plane TE and TM waves incident on a perfectly conducting cylinder for up to fifth order of accuracy, and a plane wave incident on a perfectly conducting sphere for up to fourth order of accuracy. Comparisons are made with exact solutions for these cases.

Liu, Yen↗

Load Balancing Sequences of Unstructured Adaptive Grids

Mesh adaption is a powerful tool for efficient unstructured grid computations but causes load imbalance on multiprocessor systems. To address this problem, we have developed PLUM, an automatic portable framework for performing adaptive large-scale numerical computations in a message-passing environment. This paper makes several important additions to our previous work. First, a new remapping cost model is presented and empirically validated on an SP2. Next, our load balancing strategy is applied to sequences of dynamically adapted unstructured grids. Results indicate that our framework is effective on many processors for both steady and unsteady problems with several levels of adaption. Additionally, we demonstrate that a coarse starting mesh produces high quality load balancing, at a fraction of the cost required for a fine initial mesh. Finally, we show that the data remapping overhead can be significantly reduced by applying our heuristic processor reassignment algorithm.

Biswas, Rupak↗

PLUM: Parallel Load Balancing for Unstructured Adaptive Meshes

Dynamic mesh adaption on unstructured grids is a powerful tool for computing large-scale problems that require grid modifications to efficiently resolve solution features. Unfortunately, an efficient parallel implementation is difficult to achieve, primarily due to the load imbalance created by the dynamically-changing nonuniform grid. To address this problem, we have developed PLUM, an automatic portable framework for performing adaptive large-scale numerical computations in a message-passing environment. First, we present an efficient parallel implementation of a tetrahedral mesh adaption scheme. Extremely promising parallel performance is achieved for various refinement and coarsening strategies on a realistic-sized domain. Next we describe PLUM, a novel method for dynamically balancing the processor workloads in adaptive grid computations. This research includes interfacing the parallel mesh adaption procedure based on actual flow solutions to a data remapping module, and incorporating an efficient parallel mesh repartitioner. A significant runtime improvement is achieved by observing that data movement for a refinement step should be performed after the edge-marking phase but before the actual subdivision. We also present optimal and heuristic remapping cost metrics that can accurately predict the total overhead for data redistribution. Several experiments are performed to verify the effectiveness of PLUM on sequences of dynamically adapted unstructured grids. Portability is demonstrated by presenting results on the two vastly different architectures of the SP2 and the Origin2OOO. Additionally, we evaluate the performance of five state-of-the-art partitioning algorithms that can be used within PLUM. It is shown that for certain classes of unsteady adaption, globally repartitioning the computational mesh produces higher quality results than diffusive repartitioning schemes. We also demonstrate that a coarse starting mesh produces high quality load balancing, at a fraction of the cost required a fine initial mesh. Results indicate that our parallel load balancing strategy will remain viable on large numbers of processors.

Oliker, Leonid↗

Toucan: A performance portable, scalable implementation of the DECA algorithm

In the field of additive manufacturing (AM), cellular automata (CA) is extensively used to simulate microstructural evolution during solidification. However, while traditional CA approaches are relatively fast, they still require a substantial number of time steps, are limited to moderate volumes, and are relatively difficult to improve through parallelism due to the highly localized nature of the solidification front. Here, to address these issues of time to solution and load balancing, we introduce Toucan, a parallel, performance-portable, and scalable code written in C++ with the Kokkos library that leverages the discrete event inspired cellular automata (DECA) algorithm to perform parallel-in-time (PinT) grain growth simulations. Toucan effectively mitigates load balancing issues by distributing the computational workload more evenly across processors, enhancing scalability and efficiency. We conduct both strong and weak scaling studies on up to 64 GPUs on the Frontier supercomputer, demonstrating that Toucan significantly outperforms the current state-of-the-art, time-stepped CA code, ExaCA, on both single and multi-GPU simulations. Even in AM-specific weak scaling scenarios, Toucan maintains near-ideal scaling, in contrast to the linear increase observed with ExaCA due to the moving laser raster pattern. This study highlights Toucan’s potential to transform microstructural simulations in AM by radically improving both efficiency and scalability over existing methods.

36 MATERIALS SCIENCE↗

Mota Mapper: A Multi-Objective Topology-Aware Mapper Library

Mota is a library that provides several heuristics for mapping tasks to compute ranks (processes) on a networked computer cluster or supercomputer. It was originally designed for task placement of adaptive mesh refinement codes (AMR). It is multi-objective in the sense that it simultaneously balances (potentially multiple types of) computational load on each rank as well as the communication traffic between the boxes. The heuristics used for mapping include algorithms such as greedy list assignment and space-filling curves, as well as algorithms from graph analysis such as adjacency matrix bandwidth reduction, recursive bisection, and greedy graph placement.

Chan, Cy↗

Computational strategies for three-dimensional flow simulations on distributed computer systems

This research effort is directed towards an examination of issues involved in porting large computational fluid dynamics codes in use within the industry to a distributed computing environment. This effort addresses strategies for implementing the distributed computing in a device independent fashion and load balancing. A flow solver called TEAM presently in use at Lockheed Aeronautical Systems Company was acquired to start this effort. The following tasks were completed: (1) The TEAM code was ported to a number of distributed computing platforms including a cluster of HP workstations located in the School of Aerospace Engineering at Georgia Tech; a cluster of DEC Alpha Workstations in the Graphics visualization lab located at Georgia Tech; a cluster of SGI workstations located at NASA Ames Research Center; and an IBM SP-2 system located at NASA ARC. (2) A number of communication strategies were implemented. Specifically, the manager-worker strategy and the worker-worker strategy were tested. (3) A variety of load balancing strategies were investigated. Specifically, the static load balancing, task queue balancing and the Crutchfield algorithm were coded and evaluated. (4) The classical explicit Runge-Kutta scheme in the TEAM solver was replaced with an LU implicit scheme. And (5) the implicit TEAM-PVM solver was extensively validated through studies of unsteady transonic flow over an F-5 wing, undergoing combined bending and torsional motion. These investigations are documented in extensive detail in the dissertation, 'Computational Strategies for Three-Dimensional Flow Simulations on Distributed Computing Systems', enclosed as an appendix.

Sankar, Lakshmi N.↗

Combined Final Report for Colony II Project

(This report was originally submmited by the lead PI (Terry Jones, ORNL) on October 22, 2013 to the program manager, Lucy Nowell. It is being submitted from University of Illinois in accordance with instructions). HPC Colony II seeks to provide portable performance for leadership class machines. Our strategy is based on adaptive system software that aims to make the intelligent decisions necessary to allow domain scientists to safely focus on their task at hand and allow the system software stack to adapt their application to the underlying architecture. This report describes the research undertaken towards these objectives and the results obtained over the performance period of the project.

97 MATHEMATICS AND COMPUTING↗

Development of a Stiffness-Based Chemistry Load Balancing Scheme, and Optimization of Input/Output and Communication, to Enable Massively Parallel High-Fidelity Internal Combustion Engine Simulations

A closed-cycle gasoline compression ignition engine simulation near top dead center (TDC) was used to profile the performance of a parallel commercial engine computational fluid dynamics code, as it was scaled on up to 4096 cores of an IBM Blue Gene/Q supercomputer. The test case has 9 million cells near TDC, with a fixed mesh size of 0.15 mm, and was run on configurations ranging from 128 to 4096 cores. Profiling was done for a small duration of 0.11 crank angle degrees near TDC during ignition. Optimization of input/output performance resulted in a significant speedup in reading restart files, and in an over 100-times speedup in writing restart files and files for post-processing. Improvements to communication resulted in a 1400-times speedup in the mesh load balancing operation during initialization, on 4096 cores. An improved, “stiffness-based” algorithm for load balancing chemical kinetics calculations was developed, which results in an over 3-times faster run-time near ignition on 4096 cores relative to the original load balancing scheme. With this improvement to load balancing, the code achieves over 78% scaling efficiency on 2048 cores, and over 65% scaling efficiency on 4096 cores, relative to 256 cores.

CFD, Mira, Supercomputing, High Performance Comput↗

EJFAT Scientific Perspective

Presented new computing model to the test by deploying the EJFAT system alongside a data-stream processing framework running the production-level CLAS12 event reconstruction application. In this experiment, a continuous stream of CLAS12 Level-1 identified events was processed in real-time using the EJFAT load balancer, distributing the workload across 90 computing nodes located across the U.S. This marks the first-ever large-scale, real-time distributed data stream processing experiment, demonstrating that scientific data-streaming pipelines can efficiently scale across four dimensions, thanks to EJFAT’s advanced hardware and software capabilities.

Gyurjyan, Vardan [Thomas Jefferson National Accele↗

Dynamic load balance scheme for the DSMC algorithm

The direct simulation Monte Carlo (DSMC) algorithm, devised by Bird, has been used over a wide range of various rarified flow problems in the past 40 years. While the DSMC is suitable for the parallel implementation on powerful multi-processor architecture, it also introduces a large load imbalance across the processor array, even for small examples. The load imposed on a processor by a DSMC calculation is determined to a large extent by the total of simulator particles upon it. Since most flows are impulsively started with initial distribution of particles which is surely quite different from the steady state, the total of simulator particles will change dramatically. The load balance based upon an initial distribution of particles will break down as the steady state of flow is reached. The load imbalance and huge computational cost of DSMC has limited its application to rarefied or simple transitional flows. In this paper, by taking advantage of METIS, a software for partitioning unstructured graphs, and taking the total of simulator particles in each cell as a weight information, the repartitioning based upon the principle that each processor handles approximately the equal total of simulator particles has been achieved. The computation must pause several times to renew the total of simulator particles in each processor and repartition the whole domain again. Thus the load balance across the processors array holds in the duration of computation. The parallel efficiency can be improved effectively. The benchmark solution of a cylinder submerged in hypersonic flow has been simulated numerically. Besides, hypersonic flow past around a complex wing-body configuration has also been simulated. The results have displayed that, for both of cases, the computational time can be reduced by about 50%.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Parallel rendering

This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concepts and the issues which arise in the design of parallel rendering algorithms and systems. We examine the different types of parallelism and how they can be applied in rendering applications. Concepts from parallel computing, such as data decomposition, task granularity, scalability, and load balancing, are considered in relation to the rendering problem. We also explore concepts from computer graphics, such as coherence and projection, which have a significant impact on the structure of parallel rendering algorithms. Our survey covers a number of practical considerations as well, including the choice of architectural platform, communication and memory requirements, and the problem of image assembly and display. We illustrate the discussion with numerous examples from the parallel rendering literature, representing most of the principal rendering methods currently used in computer graphics.

Crockett, Thomas W.↗

On the Impact of Execution Models: A Case Study in Computational Chemistry

Efficient utilization of high-performance computing (HPC) platforms is an important and complex problem. Execution models, abstract descriptions of the dynamic runtime behavior of the execution stack, have significant impact on the utilization of HPC systems. Using a computational chemistry kernel as a case study and a wide variety of execution models combined with load balancing techniques, we explore the impact of execution models on the utilization of an HPC system. We demonstrate a 50 percent improvement in performance by using work stealing relative to a more traditional static scheduling approach. We also use a novel semi-matching technique for load balancing that has comparable performance to a traditional hypergraph-based partitioning implementation, which is computationally expensive. Using this study, we found that execution model design choices and assumptions can limit critical optimizations such as global, dynamic load balancing and finding the correct balance between available work units and different system and runtime overheads. With the emergence of multi- and many-core architectures and the consequent growth in the complexity of HPC platforms, we believe that these lessons will be beneficial to researchers tuning diverse applications on modern HPC platforms, especially on emerging dynamic platforms with energy-induced performance variability.

Execution models, computational chemistry, MPI and↗

Indirect addressing and load balancing for faster solution to Mandelbrot Set on SIMD architectures

SIMD computers with local indirect addressing allow programs to have queues and buffers, making certain kinds of problems much more efficient. Examined here are a class of problems characterized by computations on data points where the computation is identical, but the convergence rate is data dependent. Normally, in this situation, the algorithm time is governed by the maximum number of iterations required by each point. Using indirect addressing allows a processor to proceed to the next data point when it is done, reducing the overall number of iterations required to approach the mean convergence rate when a sufficiently large problem set is solved. Load balancing techniques can be applied for additional performance improvement. Simulations of this technique applied to solving Mandelbrot Sets indicate significant performance gains.

Tomboulian, Sherryl↗