检索结果-内蒙古大学图书馆

Fine grain algorithm parallelization on a hybrid control-flow and dataflow processor

JOURNAL OF BIG DATA 2025年第1期12卷 1-25页

作者： Korolija, Nenad Univ Belgrade Sch Elect Engn Belgrade Serbia

The execution time of a high-performance computing algorithm is influenced by various factors, including the algorithm's scalability, the selected hardware for processing elements, and the communication speed between these elements. This study utilizes a hybrid architecture that integrates both control-flow and dataflow hardware. Specifically, the control-flow hardware encompasses both multicore and manycore architectures. Guidance for dataflow programmers is provided to enable them to anticipate the level of acceleration achievable with a hybrid control-flow and dataflow architecture before developing dataflow hardware algorithms. Furthermore, the methodology developed is introduced, offering a structured approach for programmers to decompose algorithms and optimize each segment by leveraging the most suitable architectural type. The prerequisite for a programmer is not to know hardware description languages, but he must be well-versed in estimating the complexity of an algorithm. This study represents the culmination of over a decade of expertise in hybrid control-flow and dataflow architectures. It provides a detailed methodology for decomposing a control-flow algorithm into segments optimized for dataflow architectures and those better suited to control-flow architectures. The Lattice-Boltzmann method is employed as a representative example, implemented on both control-flow and dataflow hardware. The estimated total acceleration factor of the decomposed Lattice-Boltzmann method on the hybrid architecture, relative to execution times using control-flow and dataflow hardware, is approximately two for a given matrix dimension. The findings underscore the advantages of employing a hybrid architecture, demonstrating significant acceleration potential even for algorithms traditionally optimized for dataflow architectures. The primary benefit of the hybrid architecture lies in its capacity to accelerate algorithms where only specific portions are suitable for dataflow

关键词： Control-flow architectures Dataflow architectures Hybrid architectures algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

A survey on partitioning models, solution algorithms and algorithm parallelization for hardware/software co-design

引用

DESIGN AUTOMATION FOR EMBEDDED SYSTEMS 2019年第1-2期23卷 57-77页

作者： Hou, Neng Yan, Xiaohu He, Fazhi Wuhan Univ Sch Comp Sci Wuhan 430072 Hubei Peoples R China

In electronic design automation, hardware/software co-design significantly reduces the time-to-market and improves the performance of embedded systems. With the increasing scale of applications and complexity of hardware architecture of embedded systems, hardware/software co-design is still a research hotspot. As hardware/software co-design is a wide topic, this paper focuses on major developments of three important aspects related to hardware/software partitioning, which has great effects on the performance of embedded systems. Firstly, various partitioning models including hardware architectures and abstract models are surveyed. Secondly, classical and new algorithms for hardware/software partitioning are classified and analyzed. Thirdly, existing parallel algorithms for hardware/software co-design are discussed in details. Finally, possible research directions are pointed out in conclusion.

关键词： Hardware software co-design Hardware software partitioning Partition model Solution algorithm algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

Parallel implementation of GCM on GPUs

引用

ICT EXPRESS 2025年第2期11卷 310-316页

作者： Lee, JaeSeok Kim, DongCheon Seo, Seog Chung Kookmin Univ Dept Informat Secur Cryptog & Math Seoul South Korea Kookmin Univ Dept Financial Informat Secur Seoul South Korea

This paper presents the first fully parallelized optimization of GCM in a GPU environment. As the era of IoT emerges, a large number of clients communicate with servers, necessitating encrypted communications for security. GCM is a type of AEAD and is currently used in various security protocols, including TLS 1.3 and IPsec. Due to the burden of performing encrypted communication with numerous clients, there has been significant research on utilizing GPUs for high-speed parallel processing in encryption. However, to date, there has been no fully parallelized implementation of GCM on GPUs. This paper proposes a method for parallelizing the challenging GHASH computation in GCM mode, leading to a high-speed parallel implementation of AESGCM that can exceed 400Gb/s, meeting the requirements of next-generation communication systems. The proposed approach is algorithm-independent and can be applied to any block ciphers. Our implementation on an RTX 4090 demonstrates a performance improvement of x15.38 compared to the maximum processing throughput of a multi-threaded Intel(R) Core(TM) i7-13700K. It also achieves a x17.87 improvement compared to a hybrid CPU-GPU system. Compared to the most researched FPGA implementation for GCM, specifically Xilinx Ultrascale FPGA, our implementation achieves x1.11 better performance. For not only throughput but also power efficiency also better than other implementation, it achieves x3.33 compared to CPU implementation on Intel Xeon E3-1220, also it achieves x21.09 compared to FPGA implementation for AES on Xilinx Virtex 7 series, which is not including full GCM.

关键词： AES GCM GPU GPGPU TLS GHASH algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

parallelization of the B static traffic assignment algorithm

引用

AIN SHAMS ENGINEERING JOURNAL 2022年第2期13卷 101576-101576页

作者： Potuzak, Tomas Kolovsky, Frantisek Univ West Bohemia Fac Appl Sci Dept Comp Sci & Engn Univ 8 Plzen 30614 Czech Republic Univ West Bohemia Fac Appl Sci Dept Geomat Univ 8 Plzen 30614 Czech Republic

A widely used technique for predicting traffic flows in individual roads of a road traffic network is the user-equilibrium (UE) traffic assignment (TA). This technique assigns trips from origins to destinations in a road traffic network so that all trips use the cheapest path. The cost of the path, which consists of roads (edges), is the sum of the roads costs. These costs increase with increasing flow in these roads. In this paper, we describe the parallelization of the B algorithm - a relatively new TA algorithm with a fast convergence to a solution. Since the nature of the algorithm and the nature of its fast convergence complicate the parallelization itself, we considered and implemented three parallel variants and tested them on real road traffic networks to investigate their convergence, usability, and speed. The parallelization is intended for a shared memory parallel computing environment. The description of the parallelization along with the performed tests is the main contribution of this paper. (C) 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.

关键词： algorithm parallelization B algorithm Static traffic assignment User equilibrium Shared memory environment

来源：评论

学校读者我要写书评

暂无评论

Application of a Direct Method for Solving the Boltzmann Equation in Supersonic Flow Computation

引用

COMPUTATIONAL MATHEMATICS AND MATHEMATICAL PHYSICS 2016年第11期56卷 1938-1947页

作者： Aristov, V. V. Rovenskaya, O. I. Russian Acad Sci Dorodnicyn Comp Ctr Fed Res Ctr Comp Sci & Control Moscow 119333 Russia

The Boltzmann kinetic equation is used to numerically study the evolution of separated flows over a backward-facing step at low Knudsen numbers. The Boltzmann equation is solved by applying an explicit-implicit scheme. To improve the efficiency of the solution algorithm, it is parallelized with the help of MPI. The solution obtained with the kinetic equation is compared with those based on continuous medium equations. It is shown that the kinetic approach makes it possible to reproduce the distributions of surface pressure, friction coefficient, and heat transfer, as well as to obtain a flow structure close to experimental data.

关键词： Boltzmann kinetic equation supersonic compressible flow explicit-implicit difference scheme flow instability algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

Scalable Fast Rank-1 Dictionary Learning for fMRI Big Data Analysis 16

Scalable Fast Rank-1 Dictionary Learning for fMRI Big Data A...

引用

22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD)

作者： Li, Xiang Makkie, Milad Lin, Binbin Fazli, Mojtaba Sedigh Davidson, Ian Ye, Jieping Liu, Tianming Quinn, Shannon Univ Georgia Dept Comp Sci Athens GA 30602 USA Univ Michigan Dept Computat Med & Bioinformat Ann Arbor MI 48109 USA Univ Calif Davis Dept Comp Sci Davis CA 95616 USA

ISBN: (纸本)9781450342322

It has been shown from various functional neuroimaging studies that sparsity-regularized dictionary learning could achieve superior performance in decomposing comprehensive and neuroscientifically meaningful functional networks from massive fMRI signals. However, the computational cost for solving the dictionary learning problem has been known to be very demanding, especially when dealing with large-scale data sets. Thus in this work, we propose a novel distributed rank-1 dictionary learning (D-r1DL) model and apply it for fMRI big data analysis. The model estimates one rank-1 basis vector with sparsity constraint on its loading coefficient from the input data at each learning step through alternating least squares updates. By iteratively learning the rank-1 basis and deflating the input data at each step, the model is then capable of decomposing the whole set of functional networks. We implement and parallelize the rank-1 dictionary learning algorithm using Spark engine and deployed the resilient distributed dataset (RDDs) abstracts for the data distribution and operations. Experimental results from applying the model on the Human Connectome Project (HCP) data show that the proposed D-HDL model is efficient and scalable towards fMRI big data analytics, thus enabling data-driven neuroscientific discovery from massive fMRI big data in the future.

关键词： fMRI Sparse coding Distributed computation algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

algorithm Parallelizing for Classifying the Complex Systems

Algorithm Parallelizing for Classifying the Complex Systems

引用

IEEE International Conference on Quality Management, Transport and Information Security, Information Technologies (IT and QM and IS)

作者： Edgulova, Elizaveta K. Lamerdonov, Zamir G. Khashirova, Tatiana Yu. Moskalenko, Larisa A. Denisenko, Vladimir A. Kabardino Balkarian State Univ Coll Informat Technol & Econ Nalchik Russia Kabardino Balkarian State Agr Univ Dept Environm & Water Construct Nalchik Russia Kabardino Balkarian State Univ Inst Informat Elect & Comp Technol Nalchik Russia Kabardino Balkarian State Univ Ctr Automated Informat Syst Nalchik Russia

The article presents one of the approaches to parallelizing the classification of algorithm for complex systems in condition of uncertainty. The offered variant of the parallelization of the classification of algorith... 详细信息

ISBN: (纸本)9781538667576

关键词： complex system information uncertainty classification of complex systems algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

Multi-band Blending of Aerial Images Using GPU Acceleration 10

Multi-band Blending of Aerial Images Using GPU Acceleration

引用

10th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)

作者： Zhao, Nan Zheng, Xinqi China Univ Geosci Sch Informat Engn Beijing Peoples R China

ISBN: (纸本)9781538619377

Aerial images are densely captured with high resolution. It is an important research topic to efficiently process the aerial images. Benefited from its powerful computation power, GPU acceleration techniques have been widely used in image processing, and many traditional techniques achieved real-time processing through using the GPU acceleration. In this research, we study the aerial image blending problems using GPU acceleration. We parallelized the Multi-band blending algorithm and achieved 30 times acceleration compared with CPU implementation.

关键词： image blending GPU acceleration aerial image image decomposition algorithm parallelization

来源：评论

学校读者我要写书评

暂无评论

CONCURRENT FREQUENT ITEMSETS MINING IN A SHARED PREFIX TREE USING THE APRIORI algorithm

CONCURRENT FREQUENT ITEMSETS MINING IN A SHARED PREFIX TREE ...

引用

SPIE-IEEE-PSP WILGA on Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments

作者： Puscian, Marek Warsaw Univ Technol Fac Elect & Informat Technol Nowowiejska 15-19 PL-00665 Warsaw Poland

ISBN: (数字)9781510622043

ISBN: (纸本)9781510622043

This paper presents a sequential frequent itemsets mining algorithm Apriori that is adapted to concurrent processing. It applies Master Slave scheme to candidate generation and support counting operations performed by threads on a single machine. Two approaches to traversing shared prefix tree and counting support of itemsets are presented and compared. Several optimization methods have been proposed for the multithreaded environment. Proposed enhancements have been successfully implemented using JAVA. This paper discusses results of the performance of concurrent Apriori algorithm against different datasets. Presented approach has been illustrated with many experiments and measurements performed using multiprocessor and multithreaded computer.

关键词： algorithm parallelization shared memory Apriori frequent items mining a measure of speed up

来源：评论

学校读者我要写书评

暂无评论

Aspect-Oriented Set@l Language for Architecture-Independent Programming of High-Performance Computer Systems 5th

Aspect-Oriented Set@l Language for Architecture-Independent ...

引用

5th Russian Supercomputing Days Conference (RuSCDays)

作者： Levin, Ilya I. Dordopulo, Alexey, I Pisarenko, Ivan, V Melnikov, Andrey K. Southern Fed Univ Acad Engn & Technol Inst Comp Technol & Informat Secur Taganrog Russia Supercomp & Neurocomp Res Ctr Taganrog Russia InformInvestGrp CJSC Moscow Russia

ISBN: (纸本)9783030365929;9783030365912

State-of-the-art programming languages for high-performance computer systems require a significant code revision for the porting of the same algorithm between different architectures. To solve this problem, we propose an architecture-independent Set@l programming language, which develops the main concepts of the COLAMO and SETL programming languages. Set@l represents the solution of a computational problem as sets with various attributes and relations between them. Furthermore, the Set@l language is based on the paradigm of aspect-oriented programming and describes an algorithm and features of its implementation as separate modules of a program. The Set@l syntax supports the classification of collections by various criteria, such as the type of parallelism, definiteness and optional user's attributes. To deal with the indefinite collections emerging in the case of the algorithm modification, the Vopenka's classification of collections by definiteness of their elements is introduced into the syntax of the Set@l programming language. In this paper, we consider the example of the Set@l application for the coding of lower-upper decomposition. The code shows that aspects can adapt the parallelizing of the basic algorithm to the implementation on computer systems with reconfigurable and multiprocessor architectures.

关键词： Architecture-independent programming Set@l programming language Aspect-oriented programming paradigm Alternative set theory of Vopenka algorithm parallelization

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：