检索结果-内蒙古大学图书馆

multithreaded derivative computation with generated libraries

JOURNAL OF computationAL SCIENCE 2010年第2期1卷 89-97页

作者： Buntinas, Darius Malozemoff, Alexis J. Utke, Jean Argonne Natl Lab Math & Comp Sci Div MCS Argonne IL 60439 USA McGill Univ Montreal PQ H3A 2T5 Canada Univ Chicago Computat Inst Chicago IL 60637 USA

The computation of derivatives via automatic differentiation is a valuable technique in many science and engineering applications. While the implementation of automatic differentiation via source transformation yields the highest efficiency results, particularly for gradient computations, the implementation via operator overloading remains a viable alternative for some application contexts. Examples include the computation of higher order derivatives or cases where C++ as the language of choice still proves to be too complicated for the currently available source transformation tools. In this paper, we utilize a code generator to create libraries that overload intrinsics for derivative computation, and discuss approaches to improve the efficiency of the generated code. We first discuss the use of limited loop unrolling, but the main focus of the paper is multithreaded derivative computation, in particular an asynchronous scheme for C++ and a synchronous scheme for Fortran. We present test results obtained with a proof-of-concept implementation. (C) 2010 Elsevier B.V. All rights reserved.

关键词： Automatic differentiation Operator overloading Generated code Loop unrolling multithreaded computation

来源：评论

学校读者我要写书评

暂无评论

Performance Evaluation of multithreaded computations for CPU Bounded Task 3

Performance Evaluation of Multithreaded Computations for CPU...

引用

3rd International Conference on Platform Technology and Service (PlatCon)

作者： Kim, Jin Yu, Sung Yun Park, Jinsoo Korea Inst Sci & Technol Informat Global Sci Expt Data Hub Ctr Daejeon South Korea Korea Res Inst Biosci & Biotechnol Natl Res Safety Headquarters Daejeon South Korea Yong In Univ Dept Management Informat Syst Yongin South Korea

ISBN: (纸本)9781467386852

The development of multicore processors causes the enhancement of computing performances. In general, these processors give rise to the whole performance improvement via simultaneous processing of separated modules such as logical module and graphical module. In the single algorithm such as the calculation of a mathematical formula, a multithreaded program can also get the same effect as a distributed processing system. The computing speed can be improved even though there is still the storage limitation. Using the multithreaded programming, this research evaluates the computing performance in the single algorithm. In the multicore processor environment, implications for the computation performance of synchronization algorithm in the multithreaded computation will also be mentioned.

关键词： Multicore processor multithreaded computation CPU bounded task

来源：评论

学校读者我要写书评

暂无评论

On the efficiency of localized work stealing

引用

INFORMATION PROCESSING LETTERS 2016年第2期116卷 100-106页

作者： Suksompong, Warut Leiserson, Charles E. Schardl, Tao B. Stanford Univ Dept Comp Sci Stanford CA 94305 USA MIT Comp Sci & Artificial Intelligence Lab Cambridge MA 02139 USA

This paper investigates a variant of the work-stealing algorithm that we call the localized work-stealing algorithm. The intuition behind this variant is that because of locality, processors can benefit from working on their own work. Consequently, when a processor is free, it makes a steal attempt to get back its own work. We call this type of steal a steal-back. We show that the expected running time of the algorithm is T-1/P + 0(T infinity P), and that under the "even distribution of free agents assumption", the expected running time of the algorithm is T-1/P + 0(T(infinity)lg P). In addition, we obtain another running-time bound based on ratios between the sizes of serial tasks in the computation. If M denotes the maximum ratio between the largest and the smallest serial tasks of a processor after removing a total of 0(P) serial tasks across all processors from consideration, then the expected running time of the algorithm is T-1/P + 0 (T infinity M). (C) 2015 Elsevier B.V. All rights reserved.

关键词： Parallel algorithms multithreaded computation Work stealing Localization

来源：评论

学校读者我要写书评

暂无评论

Cluster computing with Java

引用

COMPUTING IN SCIENCE & ENGINEERING 2005年第2期7卷 34-39页

作者： Hatcher, P Reno, M Antoniu, G Bougé, L Univ New Hampshire Durham England Ecole Normale Super IRISA Cachan France

Java could be a new lingua franca for uniting disparate computing worlds. In this article, the authors explore two approaches for Java's support of cluster computing-as single and multiple virtual machines-and eva... 详细信息

关键词： Java multi-threading multiprocessing systems virtual machines workstationclusters Java Web services cluster computing multiple virtual machines multiprocessorcomputing multithreaded computation peer-to-peer applications uniprocessor computing Hyperion

来源：评论

学校读者我要写书评

暂无评论

Performance driven distributed scheduling of parallel hybrid computations

引用

THEORETICAL COMPUTER SCIENCE 2011年第32期412卷 4212-4225页

作者： Narang, Ankur Shyamasundar, Rudrapatna K. IBM Res India Inst Area New Delhi 110070 India Tata Inst Fundamental Res Bombay 400005 Maharashtra India

Exascale computing is fast becoming a mainstream research area. In order to realize exascale performance, it is necessary to have efficient scheduling of large parallel computations with scalable performance on a large number of cores/processors. The scheduler needs to execute in a pure distributed and online fashion, should follow affinity inherent in the computation and must have low time and message complexity. Further, it should also avoid physical deadlocks due to bounded resources including space/memory per core. Simultaneous consideration of these factors makes affinity driven distributed scheduling particularly challenging. We attempt to address this challenge for hybrid parallel computations which contain tasks that have pre-specified affinity to a place and also tasks that can be mapped to any place in the system. Specifically, we address two scheduling problems of the type P(m)[M(j), prec]C(max). This paper presents online distributed scheduling algorithms for hybrid parallel computations assuming both unconstrained and bounded space per place. We also present the time and message complexity for distributed scheduling of hybrid computations. To the best of our knowledge, this is the first time that distributed scheduling algorithms for hybrid parallel computations have been presented and analyzed for time and message bounds under both unconstrained space and bounded space. (C) 2011 Elsevier B.V. All rights reserved.

关键词： Distributed scheduling multithreaded computation Asymptotic complexity Deadlock free scheduling Work stealing Online algorithm

来源：评论

学校读者我要写书评

暂无评论

Distributed Scheduling of Parallel Hybrid computations

Distributed Scheduling of Parallel Hybrid Computations

引用

20th International Symposium on Algorithms and computations (ISAAC 2009)

作者： Agarwal, Shivali Narang, Ankur Shyamasundar, Rudrapatna K. IBM India Res Lab New Delhi India Tata Inst Fundamental Res Bombay Maharashtra India

ISBN: (纸本)9783642106309

Multicore computing is fast becoming the norm. Improving parallel programming productivity without compromising performance on multicores is a serious challenge facing research community and systems vendors. Towards this end, efficient run-time scheduling of parallel programs helps programmer by dynamically mapping tasks onto processors and scheduling them in appropriate order. Distributed scheduling of parallel computations on multiple places' while ensuring low time and message complexity in bounded space is a very challenging problem. We attempt to address this challenge for hybrid parallel computations which contain tasks that have pre-specified affinity to a place and also tasks that can be mapped to any place in the system. This paper presents online distributed scheduling algorithms for hybrid parallel computations assuming both unconstrained and bounded space per place. We also present the time and message complexity for distributed scheduling of hybrid computations. To the best of our knowledge, this is the first time distributed scheduling algorithms for hybrid parallel computations have been presented and analyzed for time and message bounds under both unconstrained space and bounded space.

关键词： Work Stealing Scheduling multithreaded computation Algorithm

来源：评论

学校读者我要写书评

暂无评论

Performance driven distributed scheduling of parallel hybrid computations

Performance driven distributed scheduling of parallel hybrid...

引用

20th International Symposium on Algorithms and computations (ISAAC 2009)

作者： Narang, Ankur Shyamasundar, Rudrapatna K. IBM Res India Inst Area New Delhi 110070 India Tata Inst Fundamental Res Bombay 400005 Maharashtra India

关键词： Distributed scheduling multithreaded computation Asymptotic complexity Deadlock free scheduling Work stealing Online algorithm

来源：评论

学校读者我要写书评

暂无评论

Implementation of Symmetric Encryption by State Standard 28147-89 Algorithm on a Graphical Processor

引用

AUTOMATIC CONTROL AND COMPUTER SCIENCES 2016年第8期50卷 792-795页

作者： Sukhoparov, M. E. Lebedev, I. S. OAO NPK TRISTAN St Petersburg Branch St Petersburg Russia St Petersburg Natl Res Univ Informat Technol Mech St Petersburg Russia

A new method for implementation of symmetric encryption by the GOST (State Standard) 28147-89 algorithm using the technology of NVIDIA CUDA was suggested. The highest-performance option and system parameters providing maximal performance are defined. The effectiveness of the suggested solution as well as a comparative analysis of the given approach with existing solutions is assessed.

关键词： symmetric encryption information security graphical processor multithreaded computation

来源：评论

学校读者我要写书评

暂无评论

Mit :: Lcs :: Tr :: Mit-Lcs-Tr-677