SEARCH · Engineering Papers

Results for “distributed parallel programming”

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.

SourceResearch connectionFrom yearThrough year

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

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2017

Block-Parallel Data Analysis with DIY2

DIY2 is a programming model and runtime for block-parallel analytics on distributed-memory machines. Its main abstraction is block-structured data parallelism: data are decomposed into blocks; blocks are assigned to processing elements (processes or threads); computation is described as iterations over these blocks, and communication between blocks is defined by reusable patterns. By expressing computation in this general form, the DIY2 runtime is free to optimize the movement of blocks between slow and fast memories (disk and flash vs. DRAM) and to concurrently execute blocks residing in memory with multiple threads. This enables the same program to execute in-core, out-of-core, serial, parallel, single-threaded, multithreaded, or combinations thereof. This paper describes the implementation of the main features of the DIY2 programming model and optimizations to improve performance. DIY2 is evaluated on benchmark test cases to establish baseline performance for several common patterns and on larger complete analysis codes running on large-scale HPC machines.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2015

MapReduceXMT v. Beta 0.1

The MapReduceXMT library ports the MapReduce framework onto the Cray XMT. MapReduce is a programming paradigm and an approach to data management for unstructured problems. It has gained relevance due to its ability to map serial operations onto parallel distributed architectures, significantly improving developer/analyst productivity. The MapReduceXMT implements the key aspects of MapReduce for the Cray XMT, a massively threaded system that is inherently difficult to program. MapReduceXMT allows users to utilize the machine effectively and efficiently without extensive training in multi-threaded programming. The MapReduceXMT library ports the MapReduce framework onto the Cray XMT. MapReduce is a programming paradigm and an approach to data management for unstructured problems. It has gained relevance due to its ability to map serial operations onto parallel distributed architectures, significantly improving developer/analyst productivity. The MapReduceXMT implements the key aspects of MapReduce for the Cray XMT, a massively threaded system that is inherently difficult to program. MapReduceXMT allows users to utilize the machine effectively and efficiently without extensive training in multi-threaded programming.

Goodman, Eric↗

DOE OSTI2015

High-Performance Design Patterns for Modern Fortran

This paper presents ideas for using coordinate-free numerics in modern Fortran to achieve code flexibility in the partial differential equation (PDE) domain. We also show how Fortran, over the last few decades, has changed to become a language well-suited for state-of-the-art software development. Fortran’s new coarray distributed data structure, the language’s class mechanism, and its side-effect-free, pure procedure capability provide the scaffolding on which we implement HPC software. These features empower compilers to organize parallel computations with efficient communication. We present some programming patterns that support asynchronous evaluation of expressions comprised of parallel operations on distributed data. We implemented these patterns using coarrays and the message passing interface (MPI). We compared the codes’ complexity and performance. The MPI code is much more complex and depends on external libraries. The MPI code on Cray hardware using the Cray compiler is 1.5–2 times faster than the coarray code on the same hardware. The Intel compiler implements coarrays atop Intel’s MPI library with the result apparently being 2–2.5 times slower than manually coded MPI despite exhibiting nearly linear scaling efficiency. As compilers mature and further improvements to coarrays comes in Fortran 2015, we expect this performance gap to narrow.

Haveraaen, Magne↗

DOE OSTI2015

System-wide power management control via clock distribution network

An apparatus, method and computer program product for automatically controlling power dissipation of a parallel computing system that includes a plurality of processors. A computing device issues a command to the parallel computing system. A clock pulse-width modulator encodes the command in a system clock signal to be distributed to the plurality of processors. The plurality of processors in the parallel computing system receive the system clock signal including the encoded command, and adjusts power dissipation according to the encoded command.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2023

Distributed out-of-memory NMF on CPU/GPU architectures

We propose an efficient distributed out-of-memory implementation of the non-negative matrix factorization (NMF) algorithm for heterogeneous high-performance-computing systems. The proposed implementation is based on prior work on NMFk, which can perform automatic model selection and extract latent variables and patterns from data. In this work, we extend NMFk by adding support for dense and sparse matrix operation on multi-node, multi-GPU systems. The resulting algorithm is optimized for out-of-memory problems where the memory required to factorize a given matrix is greater than the available GPU memory. Memory complexity is reduced by batching/tiling strategies, and sparse and dense matrix operations are significantly accelerated with GPU cores (or tensor cores when available). Input/output latency associated with batch copies between host and device is hidden using CUDA streams to overlap data transfers and compute asynchronously, and latency associated with collective communications (both intra-node and inter-node) is reduced using optimized NVIDIA Collective Communication Library (NCCL) based communicators. Benchmark results show significant improvement, from 32X to 76x speedup, with the new implementation using GPUs over the CPU-based NMFk. Good weak scaling was demonstrated on up to 4096 multi-GPU cluster nodes with approximately 25,000 GPUs when decomposing a dense 340 Terabyte-size matrix and an 11 Exabyte-size sparse matrix of density 10 -6 .

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2015

SSCA1-K1

SSCA1-K1 is a parallel implementation of kernel 1 of the SSCA1 benchmark suite released by the DARPA HPCS program. This kernel is able to run in parallel on a distributed shared memory system at extreme scales using OpenSHMEM.

Baker, Matthew↗

DOE OSTI2015

SSCA1-K1

Baker, Matthew↗

DOE OSTI2020

GronOR: Massively Parallel and GPU-Accelerated Non-Orthogonal Configuration Interaction for Large Molecular Systems

GronOR is a program package for non-orthogonal configuration interaction calculations for an electronic wave function built in terms of anti-symmetrized products of multi-configuration molecular fragment wave functions. The two-electron integrals that have to be processed may be expressed in terms of atomic orbitals or in terms of an orbital basis determined from the molecular orbitals of the fragments. The code has been specifically designed for execution on distributed memory massively parallel and Graphics Processing Unit (GPU)-accelerated computer architectures, using an MPI+OpenACC/OpenMP programming approach. The task-based execution model used in the implementation allows for linear scaling with the number of nodes on the largest pre-exascale architectures available, provides hardware fault resiliency, and enables effective execution on systems with distinct central processing unit-only and GPU-accelerated partitions. The code interfaces with existing multi-configuration electronic structure codes that provide optimized molecular fragment orbitals, configuration interaction coefficients, and the required integrals. Algorithm and implementation details, parallel and accelerated performance benchmarks, and an analysis of the sensitivity of the accuracy of results and computational performance to thresholds used in the calculations are presented.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2015

Performance analysis of distributed symmetric sparse matrix vector multiplication algorithm for multi-core architectures

In this article, sparse matrix vector multiply (SpMVM) is an important kernel that frequently arises in high performance computing applications. Due to its low arithmetic intensity, several approaches have been proposed in literature to improve its scalability and efficiency in large scale computations. In this paper, our target systems are high end multi-core architectures and we use messaging passing interface + open multiprocessing hybrid programming model for parallelism. We analyze the performance of recently proposed implementation of the distributed symmetric SpMVM, originally developed for large sparse symmetric matrices arising in ab initio nuclear structure calculations. We also study important features of this implementation and compare with previously reported implementations that do not exploit underlying symmetry. Our SpMVM implementations leverage the hybrid paradigm to efficiently overlap expensive communications with computations. Our main comparison criterion is the "CPU core hours" metric, which is the main measure of resource usage on supercomputers. We analyze the effects of topology-aware mapping heuristic using simplified network load model. Furthermore, we have tested the different SpMVM implementations on two large clusters with 3D Torus and Dragonfly topology. Our results show that the distributed SpMVM implementation that exploits matrix symmetry and hides communication yields the best value for the "CPU core hours" metric and significantly reduces data movement overheads.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2014

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↗

DOE OSTI2014

Hierarchical computation in the SPMD programming model

Large-scale parallel machines are programmed mainly with the single program, multiple data (SPMD) model of parallelism. While this model has advantages of scalability and simplicity, it does not fit well with divide-and-conquer parallelism or hierarchical machines that mix shared and distributed memory. In this paper, we define the recursive single program, multiple data model (RSPMD) that extends SPMD with a hierarchical team mechanism to support hierarchical algorithms and machines. We implement this model in the Titanium language and describe how to eliminate a class of deadlocks by ensuring alignment of collective operations. We present application case studies evaluating the RSPMD model, showing that it enables divide-and-conquer algorithms such as sorting to be elegantly expressed and that team collective operations increase performance of conjugate gradient by up to a factor of two. The model also facilitates optimizations for hierarchical machines, improving scalability of particle in cell by 8x and performance of sorting and a stencil code by up to 40% and 14%, respectively.

Kamil, A↗

DOE OSTI2019

A memory-distributed quasi-Newton solver for nonlinear programming problems with a small number of general constraints

We approach the problem of parallelizing state-of-the-art nonlinear programming optimization algorithms.Specifically, we focus on parallelizing quasi-Newton interior-point methods that use limited-memory secant Hessian approximations. Such interior-point methods are known to have better convergence properties and to be more effective on large-scale problems than gradient-based and derivative-free optimization algorithms. We target nonlinear and potentially nonconvex optimization problems with an arbitrary number of bound constraints and a small number of general equality and inequality constraints on the optimization variables. These problems occur for example in the form of optimal control, optimal design, and inverse problems governed by ordinary or partial differential equations, whenever they are expressed in a “reduced-space” optimization approach. We introduce and analyze the time and space complexity of a decomposition method for solving the quasi-Newton linear systems that leverages the fact that the quasi-Newton Hessian matrix has a small number of dense blocks that border a low-rank update of a diagonal matrix. This enables an efficient parallelization on memory-distributed computers of the iterations of the optimization algorithm, a state-of-the-art filter line-search interior-point algorithm by Wächter et. al. We illustrate the efficiency of the proposed method by solving structural topology optimization problems on up to 4608 cores on a parallel machine.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2018

Adaptive Runtime Features For Distributed Graph Algorithms

Performance of distributed graph algorithms can benefit greatly by forming rapport between algorithmic abstraction and the underlying runtime system that is responsible for scheduling work and exchanging messages. However, due to their dynamic and irregular nature of computation, distributed graph algorithms written in different programming models impose varying degree of workload pressure on the runtime. To cope with such vastly different workload characteristics, a runtime has to make several trade-offs. One such trade-off arises, for example, when the runtime scheduler has to choose among alternatives such as whether to execute algorithmic work, or progress the network by probing network buffers, or throttle sending messages (termed flow control). This tradeoff decides between optimizing the throughput of a runtime scheduler by increasing the rate of execution of algorithmic work, and reducing the latency of the network messages. Another trade-off exists when a decision has to be made about when to send aggregated messages in buffers (message coalescing). This decision chooses between trading off latency for network bandwidth and vice versa. At any instant, such trade-offs emphasize either on improving the quantity of work being executed (by maximizing the scheduler throughput) or on improving the quality of work (by prioritizing better work). However, encoding static policies for different runtime features (such as flow control, coalescing) can prevent graph algorithms from achieving their full potential, thus can undermine the actual performance of a distributed graph algorithm . In this paper, we investigate runtime support for distributed graph algorithms in the context of two paradigms: variants of wellknown Bulk-Synchronous Parallel model and asynchronous programming model. We explore generic runtime features such as message coalescing (aggregation) and flow control and show that execution policies of these features need to be adjusted over time to make a positive impact on the execution time of a distributed graph algorithm. Since synchronous and asynchronous graph algorithms have different workload characteristics, not all of such runtime features may be good candidates for adaptation. Each of these algorithmic paradigms may require different set of features to be adapted over time. We demonstrate which set of feature(s) can be useful in each case to achieve the right balance of work in the runtime layer. Existing implementation of different graph algorithms can benefit from adapting dynamic policies in the underlying runtime.

Firoz, Jesun S.↗

DOE OSTI2015

ZettaBricks: A Language Compiler and Runtime System for Anyscale Computing

This grant supported the ZettaBricks and OpenTuner projects. ZettaBricks is a new implicitly parallel language and compiler where defining multiple implementations of multiple algorithms to solve a problem is the natural way of programming. ZettaBricks makes algorithmic choice a first class construct of the language. Choices are provided in a way that also allows our compiler to tune at a finer granularity. The ZettaBricks compiler autotunes programs by making both fine-grained as well as algorithmic choices. Choices also include different automatic parallelization techniques, data distributions, algorithmic parameters, transformations, and blocking. Additionally, ZettaBricks introduces novel techniques to autotune algorithms for different convergence criteria. When choosing between various direct and iterative methods, the ZettaBricks compiler is able to tune a program in such a way that delivers near-optimal efficiency for any desired level of accuracy. The compiler has the flexibility of utilizing different convergence criteria for the various components within a single algorithm, providing the user with accuracy choice alongside algorithmic choice. OpenTuner is a generalization of the experience gained in building an autotuner for ZettaBricks. OpenTuner is a new open source framework for building domain-specific multi-objective program autotuners. OpenTuner supports fully-customizable configuration representations, an extensible technique representation to allow for domain-specific techniques, and an easy to use interface for communicating with the program to be autotuned. A key capability inside OpenTuner is the use of ensembles of disparate search techniques simultaneously; techniques that perform well will dynamically be allocated a larger proportion of tests.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Divide and Conquer (DC) BLAST: fast and easy BLAST execution within HPC environments

Bioinformatics is currently faced with very large-scale data sets that lead to computational jobs, especially sequence similarity searches, that can take absurdly long times to run. For example, the National Center for Biotechnology Information (NCBI) Basic Local Alignment Search Tool (BLAST and BLAST+) suite, which is by far the most widely used tool for rapid similarity searching among nucleic acid or amino acid sequences, is highly central processing unit (CPU) intensive. While the BLAST suite of programs perform searches very rapidly, they have the potential to be accelerated. In recent years, distributed computing environments have become more widely accessible and used due to the increasing availability of high-performance computing (HPC) systems. Therefore, simple solutions for data parallelization are needed to expedite BLAST and other sequence analysis tools. However, existing software for parallel sequence similarity searches often requires extensive computational experience and skill on the part of the user. In order to accelerate BLAST and other sequence analysis tools, Divide and Conquer BLAST (DCBLAST) was developed to perform NCBI BLAST searches within a cluster, grid, or HPC environment by using a query sequence distribution approach. Scaling from one (1) to 256 CPU cores resulted in significant improvements in processing speed. Thus, DCBLAST dramatically accelerates the execution of BLAST searches using a simple, accessible, robust, and parallel approach. DCBLAST works across multiple nodes automatically and it overcomes the speed limitation of single-node BLAST programs. DCBLAST can be used on any HPC system, can take advantage of hundreds of nodes, and has no output limitations. Thus, this freely available tool simplifies distributed computation pipelines to facilitate the rapid discovery of sequence similarities between very large data sets.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2024

Custom Accessors: Enabling Scalable Data Ingestion, (Re-)Organization, and Analysis on Distributed Systems

The emerging class of high velocity and high volume data analytic workflows comprise interwoven data ingestion, organization, and processing stages, with ingestion and organization steps often contributing comparable or even higher computational costs than actual processing steps. Since complex workflows consist of a variety of phases that view and use data differently, being able to construct efficient, scalable, distributed data structures (arrays, vectors, sets, maps, and multi-maps) is essential and requires custom methods to extend and shrink containers, analyze and position data, and, maintain globallyconsistent meta-data. In this paper, we propose a novel datastructure access paradigm based on the concept of Accessors. At a high level, accessors are customizable callable objects that can modify the behavior of insert, read, update, and delete operations for distributed containers while preserving atomicity guarantees. Accessors provide a very clean and natural way to implement a variety of programming patterns, e.g., conditional insertion/deletion and cascading computations, which would be otherwise hard (or even impossible) to express in parallel and distributed settings without using locks. We demonstrate the practicality and usefulness of our approach with two representative use cases and study the performance of these applications on a distributed High-Performance Computing system. Our analysis highlights that our proposed abstraction allows for an effective overlapping and concurrent execution of different workflow steps (e.g., data ingestion and analysis), which in a conventional analytics pipeline would execute sequentially, contributing cumulatively to the overall latency.

Castellana, Vito G. [BATTELLE (PACIFIC NW LAB)] (O↗

DOE OSTI2025

Parallel Programming in MCNP6

Monte Carlo N-Particle (MCNP)1 is a general-purpose Monte Carlo particle transport code developed by Los Alamos National Laboratory (LANL). To efficiently handle long simulations, MCNP version 6 (MCNP6) supports parallel execution using two primary programming models: • Shared-memory task-based threading using OpenMP (Open Multi-Processing), and • Distributed-memory calculations using MPI (Message Passing Interface). The OpenMP and MPI programming models enable MCNP6 to scale from desktop systems to high-performance computing (HPC) clusters, allowing users to run MCNP in one of three parallel modes: • OpenMP-only, • MPI-only, and • Hybrid (MPI + OpenMP). The choice of parallelization mode depends on the underlying computer architecture and the characteristics of the simulation problem.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Parallel computing for power system climate resiliency: Solving a large-scale stochastic capacity expansion problem with mpi-sppy

Here we propose a nodal stochastic generation and transmission expansion planning model that incorporates the output from high-resolution global climate models through load and generation availability scenarios. We implement our model in Pyomo and perform computational studies on a realistically-sized test case of the California electric grid in a high performance computing environment. We propose model reformulations and algorithm tuning to efficiently solve this large problem using a variant of the Progressive Hedging Algorithm. We utilize the parallelization capabilities and overall versatility of mpi-sppy, exploiting its hub-and-spoke architecture to concurrently obtain inner and outer bounds on an optimal expansion plan. Initial results show that instances with 360 representative days on a system with over 8,000 buses can be solved to within 5% of optimality in under 4 h of wall clock time, a first step towards solving a large-scale power system expansion planning problem across a wide range of climate-informed operational scenarios.

24 POWER TRANSMISSION AND DISTRIBUTION↗