检索结果-内蒙古大学图书馆

32nd IEEE International Parallel and Distributed Processing Symposium (IPDPS)

作者： Petruzza, Steve Treichler, Sean Pascucci, Valerio Bremer, Peer-Timo Univ Utah Salt Lake City UT 84112 USA Stanford Univ Stanford CA 94305 USA Lawrence Livermore Natl Lab Livermore CA USA

ISBN: (纸本)9781538643686

The rapid growth in simulation data requires large-scale parallel implementations of scientific analysis and visualization algorithms, both to produce results within an acceptable timeframe and to enable in situ deployment. However, efficient and scalable implementations, especially of more complex analysis approaches, require not only advanced algorithms, but also an in-depth knowledge of the underlying runtime. Furthermore, different machine configurations and different applications may favor different runtimes, i.e., MPI vs Charm++ vs Legion, etc., and different hardware architectures. This diversity makes developing and maintaining a broadly applicable analysis software infrastructure challenging. We address some of these problems by explicitly separating the implementation of individual tasks of an algorithm from the dataflow connecting these tasks. In particular, we present an embedded domain specific language (EDSL) to describe algorithms using a new task graph abstraction. This task graph is then executed on top of one of several available runtimes (MPI, Charm++, Legion) using a thin layer of library calls. We demonstrate the flexibility and performance of this approach using three different large scale analysis and visualization use cases, i.e., topological analysis, rendering and compositing dataflow, and image registration of large microscopy scans. Despite the unavoidable overheads of a generic solution, our approach demonstrates performance portability at scale, and, in some cases, outperforms hand-optimized implementations.

关键词： Embedded DSL User productivity In-situ analysis Simulation runtime systems programming models

来源：评论

学校读者我要写书评

暂无评论

Optimizing Collective Communication in UPC 28

Optimizing Collective Communication in UPC

引用

28th IEEE International Parallel & Distributed Processing Symposium Workshops (IPDPSW)

作者： Jose, Jithin Hamidouche, Khaled Zhang, Jie Venkatesh, Akshay Panda, Dhabaleswar K. (DK) Ohio State Univ Dept Comp Sci & Engn Columbus OH 43210 USA

ISBN: (纸本)9781479941162

Message Passing Interface (MPI) has been the defacto programming model for scientific parallel applications. However, data driven applications with irregular communication patterns are harder to implement using MPI. The Partitioned Global Address Space (PGAS) programming models present an alternative approach to improve programmability. PGAS languages like UPC are growing in popularity because of their ability to provide shared-memory programming model over distributed memory machines. However, since UPC is an emerging standard, it is unlikely that entire applications will be re-written with it. Instead, unified communication runtimes have paved the way for a new class of hybrid applications that can leverage the benefits of both MPI and PGAS models. Such unified runtimes need to be designed in a high performance, scalable manner to improve the performance of emerging hybrid applications. Collective communication primitives offer a flexible, portable way to implement group communication operations and are supported in both MPI and PGAS programming models. Owing to their advantages, they are also widely used across various scientific parallel applications. Over the years, MPI libraries have relied upon aggressive software-/hardware-based and kernel-assisted optimizations to deliver low communication latency for various collective operations. However, there is much room for improvement for collective operations in state-of-the-art, open-source implementations of UPC. In this paper, we address the challenges associated with improving the performance of collective primitives in UPC. Further, we also explore design alternatives to enable collective primitives in UPC to directly leverage the designs available in the MVAPICH2 MPI library. Our experimental evaluations show that our designs improve the performance of the UPC broadcast and all-gather operations, by 25X and 18X respectively for 128KB message at 2,048 processes. Our designs improve the performance of the UPC 2D

关键词： UPC Collectives InfiniBand programming models PGAS

来源：评论

学校读者我要写书评

暂无评论

Compiler Techniques for Massively Scalable Implicit Task Parallelism 14

Compiler Techniques for Massively Scalable Implicit Task Par...

引用

International Conference on High Performance Computing, Networking, Storage and Analysis

作者： Armstrong, Timothy G. Wozniak, Justin M. Wilde, Michael Foster, Ian T. Univ Chicago Dept Comp Sci Chicago IL 60637 USA Argonne Natl Lab Math & Comp Sci Div Argonne IL 60439 USA Univ Chicago Computat Inst Chicago IL 60637 USA Argonne Natl Lab Chicago IL USA

ISBN: (纸本)9781479955008

Swift/T is a high-level language for writing concise, deterministic scripts that compose serial or parallel codes implemented in lower-level programming models into large-scale parallel applications. It executes using a data-driven task parallel execution model that is capable of orchestrating millions of concurrently executing asynchronous tasks on homogeneous or heterogeneous resources. Producing code that executes efficiently at this scale requires sophisticated compiler transformations: poorly optimized code inhibits scaling with excessive synchronization and communication. We present a comprehensive set of compiler techniques for data-driven task parallelism, including novel compiler optimizations and intermediate representations. We report application benchmark studies, including unbalanced tree search and simulated annealing, and demonstrate that our techniques greatly reduce communication overhead and enable extreme scalability, distributing up to 612 million dynamically load balanced tasks per second at scales of up to 262,144 cores without explicit parallelism, synchronization, or load balancing in application code.

关键词： Parallel scripting Swift High-performance computing Compiler optimization Data-flow Execution models programming models

来源：评论

学校读者我要写书评

暂无评论

Exascale programing models for extreme data processing 23

Exascale programing models for extreme data processing

引用

20th ACM International Conference on Computing Frontiers (CF)

作者： Garcia-Blas, Javier Carretero, Jesus Univ Carlos III Madrid Madrid Spain

ISBN: (纸本)9798400701405

Extreme Data is an incarnation of Big Data concept distinguished by the massive amounts of data that must be queried, communicated and analysed in (near) real-time by using a very large number of memory/storage elements and Exascale computing systems. Immediate examples are the scientific data produced at a rate of hundreds of gigabits-per-second that must be stored, filtered and analysed, the millions of images per day that must be mined/analyzed in parallel, the one billion of social data posts queried in real-time on an in-memory components database. Traditional disks or commercial storage cannot handle nowadays the extreme scale of such application data. Following the need of improvement of current concepts and technologies, ASPIDE's activities focus on data-intensive applications running on systems composed of up to millions of computing elements (Exascale systems). Practical results will include the methodology and software prototypes that will be designed and used to implement Exascale applications. The ASPIDE project contributed with the definition of a new programming paradigms, APIs, runtime tools and methodologies for expressing data-intensive tasks on Exascale systems, which can pave the way for the exploitation of massive parallelism over a simplified model of the system architecture, promoting high performance and efficiency, and offering powerful operations and mechanisms for processing extreme data sources at high speed and/or real-time. This work is continued now in the ADMIRE EuroHPC project.

关键词： High Performance Data Analysis programming models Monitoring Auto-tuning

来源：评论

学校读者我要写书评

暂无评论

Distributed memory futures for compile-time, deterministic-by-default concurrency in distributed C plus plus applications 4

Distributed memory futures for compile-time, deterministic-b...

引用

IEEE/ACM 4th International Workshop on Extreme Scale programming models and Middleware (ESPM2)

作者： Wilke, Jeremiah J. Hollman, David S. Lewis, Cannada Markosyan, Aram Morals, Nicolas Sandia Natl Labs Scalable Modeling & Anal Livermore CA 94550 USA

ISBN: (纸本)9781728101781

Futures are a widely-used abstraction for enabling deferred execution in imperative programs. Deferred execution enqueues tasks rather than explicitly blocking and waiting for them to execute. Many task-based programming models with some form of deferred execution rely on explicit parallelism that is the responsibility of the programmer. Deterministic-by-default (implicitly parallel) models instead use data effects to derive concurrency automatically, alleviating the burden of concurrency management. Both implicitly and explicitly parallel models are particularly challenging for imperative object-oriented programming. Fine-granularity parallelism across member functions or amongst data members may exist, but is often ignored. In this work, we define a general permissions model that leverages the C++ type system and move semantics to define an asynchronous programming model embedded in the C++ type system. Although a default distributed memory semantic is provided, the concurrent semantics are entirely configurable through C++ constexpr integers. Correct use of the defined semantic is verified at compile-time, allowing deterministic-by-default concurrency to be safely added to applications. Here we demonstrate the use of these "extended futures" for distributed memory asynchronous communication and load balancing. An MPI particle-in-cell application is modified with the wrapper class using this task model, with results presented for a Haswell system up to 64 nodes.

关键词： task-based runtimes template metaprogramming parallel computing programming models execution models

来源：评论

学校读者我要写书评

暂无评论

TURNPIKE THEORY IN DETERMINISTIC DISCRETE DYNAMIC programming WITH DISCOUNT FACTOR GREATER THAN ONE

引用

SIAM JOURNAL ON APPLIED MATHEMATICS 1973年第4期24卷 467-472页

作者： DIRICKX, YMI YALE UNIV NEW HAVENCT 06520

Previously known results concerning the structure of optimal policies will be applied here to derive turnpike theorems for the deterministic discrete dynamic programming model with discount factor greater than one.

关键词： Toll roads Dynamic programming Determinism Algorithms programming models

来源：评论

学校读者我要写书评

暂无评论

A Distributed Stream Library for Java 8

引用

IEEE TRANSACTIONS ON BIG DATA 2017年第3期3卷 262-275页

作者： Chan, Yu Wellings, Andy Gray, Ian Audsley, Neil Univ York Dept Comp Sci York YO10 5DD N Yorkshire England

Java 8 has introduced new capabilities such as lambda expressions and streams which simplify data-parallel computing. However, as a base language for Big Data systems, it still lacks a number of important capabilities such as processing very large datasets and distributing the computation over multiple machines. This paper gives an overview of the Java 8 Streams API and proposes extensions to allow its use in Big Data systems. It also shows how the API can be used to implement a range of standard Big Data paradigms. Finally, it compares performance with that of Hadoop and Spark. Despite being a proof-of-concept implementation, results indicate that it is a lightweight and efficient framework, comparable in performance to Hadoop and Spark, and is up to 5 times faster for the largest input sizes tested.

关键词： Big data Java distributed computing programming models

来源：评论

学校读者我要写书评

暂无评论

PERFECT AGGREGATION AND DISAGGREGATION OF COMPLEMENTARITY-PROBLEMS - COMMENT

引用

AMERICAN JOURNAL OF AGRICULTURAL ECONOMICS 1981年第4期63卷 753-755页

作者： GUCCIONE, A University of WindsorOntario

Comments on a study which proposed a macromodel for the aggregation of programming models. Aggregation of von Thunen's model; Comparison of the macromodel with the macromodel proposed by Richard Day to consolidate... 详细信息

关键词： Aggregation Miniaturization Industrial production programming models Industrial market Industrial output Production functions Transportation costs

来源：评论

学校读者我要写书评

暂无评论

Multi-processor SoC-based design methodologies using configurable and extensible processors

引用

JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY 2008年第1-2期53卷 113-127页

作者： Martin, Grant Tensilica Inc Santa Clara CA USA

The growing interest in multiprocessor system-on-chip (MPSoC) design, or 'multicore' processors, has resulted in some confusion between the various types of multiprocessor architectures and their suitability in different application spaces. In particular, there are clear differences between the general-purpose, symmetric multiprocessor (SMP) approaches, and the application-specific, asymmetric multiprocessor (AMP) architectures. Configurable and extensible processors are especially suited for the AMP approach, yet their flexibility means that new design methodologies and tools must be developed to allow effective utilisation of multiple instruction-set processors in a complex design. Configurable and extensible processors are especially well suited for data-intensive computational tasks, such as are found in many signal and image processing applications, including audio, video, and wireless and wired networking. A design methodology for such applications must pay careful attention to the right programming models, and dataflow styles of processing seem a natural fit to the application space. In this paper, we describe a design methodology, flow and tools for MPSoC design using configurable and extensible processors that is especially interesting for data-intensive dataflow style applications. Some of the issues involved in this design approach are used to highlight opportunities for ongoing research.

关键词： MPSoC multiprocessor system-on-chip configurable processor instruction-set extension electronic system-level design programming models dataflow

来源：评论

学校读者我要写书评

暂无评论

Apollo: Reusable models for Fast, Dynamic Tuning of Input-Dependent Code 31

Apollo: Reusable Models for Fast, Dynamic Tuning of Input-De...

引用

31st IEEE International Parallel and Distributed Processing Symposium (IPDPS)

作者： Beckingsale, David Pearce, Olga Laguna, Ignacio Gamblin, Todd Lawrence Livermore Natl Lab Ctr Appl Sci Comp Livermore CA 94550 USA

ISBN: (纸本)9781538639146

Increasing architectural diversity makes performance portability extremely important for parallel simulation codes. Emerging on-node parallelization frameworks such as Kokkos and RAJA decouple the work done in kernels from the parallelization mechanism, allowing for a single source kernel to be tuned for different architectures at compile time. However, computational demands in production applications change at runtime, and performance depends both on the architecture and the input problem, and tuning a kernel for one set of inputs may not improve its performance on another. The statically optimized versions need to be chosen dynamically to obtain the best performance. Existing auto-tuning approaches can handle slowly evolving applications effectively, but are too slow to tune highly input-dependent kernels. We developed Apollo, an auto-tuning extension for RAJA that uses pre-trained, reusable models to tune input-dependent code at runtime. Apollo is designed for highly dynamic applications;it generates sufficiently low-overhead code to tune parameters each time a kernel runs, making fast decisions. We apply Apollo to two hydrodynamics benchmarks and to a production multi-physics code, and show that it can achieve speedups from 1.2x to 4.8x.

关键词： autotuning Machine learning performance programming models

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：