检索结果-内蒙古大学图书馆

Improving performance portability for GPU-specific Open CL kernels on multi-core/many-core CPUs by analysis-based transformations

引用

Frontiers of Information Technology & Electronic Engineering 2015年第11期16卷 899-916页

作者： Mei WEN Da-fei HUANG Chang-qing XUN Dong CHEN School of Computer National University of Defense Technology National Key Laboratory of Parallel and Distributed Processing

OpenCL is an open heterogeneous programming framework. Although OpenCL programs are func- tionally portable, they do not provide performance portability, so code transformation often plays an irreplaceable role. When adapting GPU-specific OpenCL kernels to run on multi-core/many-core CPUs, coarsening the thread granularity is necessary and thus has been extensively used. However, locality concerns exposed in GPU-specific OpenCL code are usually inherited without analysis, which may give side-effects on the CPU performance. Typi- cally, the use of OpenCL＇s local memory on multi-core/many-core CPUs may lead to an opposite performance effect, because local-memory arrays no longer match well with the hardware and the associated synchronizations are costly. To solve this dilemma, we actively analyze the memory access patterns using array-access descriptors derived from GPU-specific kernels, which can thus be adapted for CPUs by （1） removing all the unwanted local-memory arrays together with the obsolete barrier statements and （2） optimizing the coalesced kernel code with vectorization and locality re-exploitation. Moreover, we have developed an automated tool chain that makes this transformation of GPU-specific OpenCL kernels into a CPU-friendly form, which is accompanied with a scheduler that forms a new OpenCL runtime. Experiments show that the automated transformation can improve OpenCL kernel performance on a multi-core CPU by an average factor of 3.24. Satisfactory performance improvements axe also achieved on Intel＇s many-integrated-core coprocessor. The resultant performance on both architectures is better than or comparable with the corresponding OpenMP performance.

关键词： OpenCL Performance portability Multi-core/many-core CPU Analysis-based transformation

来源：评论

学校读者我要写书评

暂无评论

Locality Protected Dynamic Cache Allocation Scheme on GPUs

Locality Protected Dynamic Cache Allocation Scheme on GPUs

引用

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Yang Zhang Zuocheng Xing Li Zhou Chunsheng Zhu National Laboratory for Parallel and Distributed Processing National University of Defense Technology Changsha China School of Electronic Science and Engineering National University of Defense Technology Changsha China Department of Electrical and Computer Engineering University of British Columbia Vancouver BC Canada

ISBN: (纸本)9781509032068

As we are approaching the exascale era in supercomputing, designing a balanced computer system with powerful computing ability and low energy consumption becomes increasingly important. GPU is a widely used accelerator in most recently applied supercomputers. It adopts massive multithreads to hide long latency and has high energy efficiency. In contrast to its strong computing power, GPUs have few on-chip resources with several MB of fast on-chip memory storage per SM (Streaming Multiprocessors). GPU caches exhibit poor efficiency due to the mismatch of the throughput-oriented execution model and its cache hierarchy design. Since the severe deficiency in on-chip memory, the benefit of high computing capacity of GPUs is pulled down by the poor cache performance dramatically, which limits system performance and energy-efficiency. In this paper, we put forward a locality protected scheme to make full use of the data locality based on the fixed capacity. We present a Locality Protected method based on instruction PC (LPP) to promote GPU performance. Firstly, we use a PC-based collector to collect the reuse information of each cache line. After getting the dynamic reuse information of the cache line, we take an intelligent cache allocation unit (ICAU) which coordinates the reuse information with LRU (Least Recently Used) replacement policy to find out the cache line with the least locality for eviction. The results show that LPP provides an up to 17.8% speedup and an average of 5.5% improvement over the baseline method.

关键词： Graphics processing units System-on-chip computer architecture Instruction sets Resource management Optimization Electronic mail

来源：评论

学校读者我要写书评

暂无评论

An Efficient and Flexible Deterministic Framework for Multithreaded Programs

引用

Journal of computer science & Technology 2015年第1期30卷 42-56页

作者：卢凯周旭王小平 Tom Bergan 陈沉 Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha 410073 China College of Computer National University of Defense Technology Changsha 410073 China Department of Computer Science and Engineering University of Washington Seattle WA 98195-2350 U.S.A.

Determinism is very useful to multithreaded programs in debugging, testing, etc. Many deterministic ap- proaches have been proposed, such as deterministic multithreading （DMT） and deterministic replay. However, these sys- tems either are inefficient or target a single purpose, which is not flexible. In this paper, we propose an efficient and flexible deterministic framework for multithreaded programs. Our framework implements determinism in two steps： relaxed determinism and strong determinism. Relaxed determinism solves data races eificiently by using a proper weak memory consistency model. After that, we implement strong determinism by solving lock contentions deterministically. Since we can apply different approaches for these two steps independently, our framework provides a spectrum of deterministic choices, including nondeterministic system （fast）, weak deterministic system （fast and conditionally deterministic）, DMT system, and deternfinistic replay system. Our evaluation shows that the DMT configuration of this framework could even outperform a state-of-the-art DMT system.

关键词： determinism multithreading framework flexible

来源：评论

学校读者我要写书评

暂无评论

High Performance Interconnect Network for Tianhe System

引用

Journal of computer science & Technology 2015年第2期30卷 259-272页

作者：廖湘科庞征王克非卢宇彤谢旻夏军董德尊所光 College of Computer National University of Defense Technology Changsha 410073 Science and Technology on Parallel and Distributed Processing Laboratory National Changsha 410073 China China University of Defense Technology State Key Laboratory of High Performance Computing National University of Defense Technology Changsha 410073 China

In this paper, we present the Tianhe-2 interconnect network and message passing services. We describe the architecture of the router and network interface chips, and highlight a set of hardware and software features effectively supporting high performance communications, ranging over remote direct memory access, collective optimization, hardwareenable reliable end-to-end communication, user-level message passing services, etc. Measured hardware performance results are also presented.

关键词： Tianhe-2 supercomputer interconnect network router architecture network interface architecture user-level message passing

来源：评论

学校读者我要写书评

暂无评论

parallel pipeline implementation for moving objects detection in traffic video analysis on a heterogeneous platform 4th

Parallel pipeline implementation for moving objects detectio...

引用

4th International Conference on computer Engineering and Networks, CENet2014

作者： Li, Teng Dou, Yong Jiang, Jingfei Qiao, Peng Department of Computer Science National Laboratory for Parallel and Distributed Processing National University of Defense Technology Changsha410073 China

ISBN: (纸本)9783319111032

Moving objects detection is important in traffic video analysis, and many algorithms are being increasingly applied to moving objects detection. Most of these algorithms are time-consuming and cannot satisfy real-time demand in traffic video analysis by using a conventional central processing unit (CPU) sequential method. The emergence of the graphics processing unit (GPU) and multi-core CPU provides a method to accelerate the aforementioned algorithms to meet demand. In this study, we provide a GPU-accelerated implementation of the background subtraction algorithm and the morphological operation algorithm. Then, the connected component labeling algorithm is parallelized on a multi-core CPU with Open Multi-processing (OpenMP). Furthermore, parallel pipeline implementation on a heterogeneous platform is proposed by integrating the aforementioned algorithms. Experimental results show that the proposed implementation achieves a significant speedup of up to 5×, compared with sequential implementation on a CPU. © Springer International Publishing Switzerland 2015.

关键词： Program processors

来源：评论

学校读者我要写书评

暂无评论

Labelwalking nonnegative matrix factorization 40

Labelwalking nonnegative matrix factorization

引用

40th IEEE International Conference on Acoustics, Speech, and Signal processing, ICASSP 2015

作者： Lan, Long Guan, Naiyang Zhang, Xiang Huang, Xuhui Luo, Zhigang Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Changsha China

ISBN: (纸本)9781467369978

Semi-supervised learning (SSL) utilizes plenty of unlabeled examples to boost the performance of learning from limited labeled examples. Due to its great discriminant power, SSL has been widely applied to various real-world tasks such as information retrieval, pattern recognition, and speech separa- tion. Label propagation (LP) is a popular SSL method which propagates labels through the dataset along high density areas defined by unlabeled examples, LP assumes nearby examples should share the same label, thus, it unavoidably pushes the labels to the wrong examples, especially when different la- beled examples are not strictly separated. Seed K-means uses labeled examples to initialize class centers, and avoid getting stuck in poor local optima comparing to traditional K-means, however the hard constraint of each example's membership makes Seed K-means failed in many real world applications. This paper proposes a novel label walking nonnegative matrix factorization method (LWNMF) to handle labeled examples in SSL based on the framework of NMF. LWNMF decomposes the whole dataset into the product of a basis matrix and a coefficient matrix, and to travel labels to unlabeled examples, LWNMF regards the class indicators of labeled examples as their coefficients and iteratively updates both basis matrix and coefficients of unlabeled examples. Since LWNMF learns comprehensive class centroids, labels iteratively walk to unlabeled examples through these significant centroids. © 2015 IEEE.

关键词： K-means Label propagation Nonnegative matrix factorization Semisupervised learning

来源：评论

学校读者我要写书评

暂无评论

Identifying repeated interleavings to improve the efficiency of concurrency bug detection 15th

Identifying repeated interleavings to improve the efficiency...

引用

15th International Conference on Algorithms and Architectures for parallel processing, ICA3PP 2015

作者： Wu, Zhendong Lu, Kai Wang, Xiaoping Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Changsha410073 China

ISBN: (纸本)9783319271392

Detecting concurrency bugs is becoming increasingly important. Many pattern-based concurrency bug detectors focus on the specific types of interleavings that are correlated to concurrency bugs. To detect multiple types of concurrency bugs, general detectors focus on multiple types of interleaving patterns, including data race pattern, atomicity violation pattern, and atomic-set violation pattern. Unfortunately, they suffer from redundant analysis due to repeated interleavings, which may affect the efficiency of concurrency bug detection. Hence, we propose an approach to identify the repeated interleavings. To the best of our knowledge, this is the first approach that can prune repeated interleavings to improve the efficiency of concurrency bug detection. We apply our approach to existing general detectors (PECAN and Maple) to avoid analyzing repeated interleavings. We evaluated the general detectors with and without our approach, respectively. The experimental results show that the bug detection results are not affected. With our approach, the bug detection time of PECAN and Maple are reduced by 40.0% and 44.4%, respectively. Additionally, our approach does not affect the overhead of bug detection, and consumes only a little memory. © Springer International Publishing Switzerland 2015.

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

Cooperative repair based on tree structure for multiple failures in distributed storage systems with regenerating codes 15

Cooperative repair based on tree structure for multiple fail...

引用

12th ACM International Conference on Computing Frontiers, CF 2015

作者： Pei, Xiaoqiang Wang, Yijie Ma, Xingkong Fu, Yongquan Xu, Fangliang Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Changsha Hunan410073 China

ISBN: (纸本)9781450333580

Regenerating codes have been proposed to achieve an optimal trade-off curve between the amount of storage space and the network traffic for repair. However, existing repair schemes based on regenerating codes are inadequate to meet the requirements of small network traffic cost and high efficiency when repairing multiple failures. In this paper, we propose a cooperative repair scheme based on tree structure for multiple failures with regenerating codes, called CTREE. For generality, we propose a two-layer repair framework to support both repairs for single and multiple failures. For high repair efficiency, a parallel tree-structured data transmission technique is proposed to organize the data transmissions between the providers and newcomers. For small network network traffic cost, a core-based data exchange technique is proposed to organize the data exchanges between the coordinator and the other newcomers. To evaluate the performance of CTREE, we conduct experiments on both 30 physical and 200 virtual servers. Numerical analysis and extensive experiments confirm that CTREE can support both single and multiple failure repairs, significantly reduces the network traffic cost and improves the repair efficiency compared with the state-of-the-art approaches under various parameter settings. © Copyright 2015 ACM.

关键词： Repair

来源：评论

学校读者我要写书评

暂无评论

Tree-structured bloom filters for joint optimization of false positive probability and transmission bandwidth

Tree-structured bloom filters for joint optimization of fals...

引用

ACM SIGMETRICS International Conference on Measurement and Modeling of computer Systems, SIGMETRICS 2015

作者： Fu, Yongquan Biersack, Ernst Science and Technology Laboratory of Parallel and Distributed Processing School of Computer National Univ. of Defense Technology China Www.e-biersack.eu France

ISBN: (纸本)9781450334860

Bloom filters are frequently used to perform set queries that test the existence of some items. However, Bloom filters face a dilemma: the transmission bandwidth and the accuracy cannot be optimized simultaneously. This dilemma is particularly severe for transmitting Bloom filters to remote nodes when the network bandwidth is limited. We propose a novel Bloom filter BloomTree that consists of a tree-structured organization of smaller Bloom filters, each one using a set of independent hash functions. BloomTree spreads items across levels that are compressed to reduce the transmission bandwidth need. We investigate in detail under which conditions BloomTree performs better than the compressed Bloom filter and the standard Bloom filter. © Copyright 2015 ACM.

关键词： Hash functions

来源：评论

学校读者我要写书评

暂无评论

Poster: Segmentation Based Online Performance Problem Diagnosis 37

Poster: Segmentation Based Online Performance Problem Diagno...

引用

37th IEEE/ACM International Conference on Software Engineering, ICSE 2015

作者： Zhou, Jingwen Chen, Zhenbang Wang, Ji Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha China College of Computer National University of Defense Technology Changsha China

ISBN: (纸本)9781479919345

Currently, the performance problems of software systems gets more and more attentions. Among various diagnosis methods based on system traces, principal component analysis (PCA) based methods are widely used due to the high accuracy of the diagnosis results and requiring no specific domain knowledge. However, according to our experiments, we have validated several shortcomings existed in PCA-based methods, including requiring traces with a same call sequence, inefficiency when the traces are long, and missing performance problems. To cope with these issues, we introduce a segmentation based online diagnosis method in this poster. © 2015 IEEE.

关键词： Principal component analysis

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：