To solve the problem that the central guidance system takes too long time to calculate the shortest routes between all node pairs of network which can not meet the real-time demand of central guidance, this paper pres...
详细信息
This paper presents the design, prototype implementation, and evaluation of a runtime management framework for structured adaptive mesh refinement applications. The framework is capable of reactively and proactively m...
详细信息
ISBN:
(数字)9783540362654
ISBN:
(纸本)3540003037
This paper presents the design, prototype implementation, and evaluation of a runtime management framework for structured adaptive mesh refinement applications. The framework is capable of reactively and proactively managing and optimizing application execution using current system and application state, predictive models for system behavior and application performance, and an agent based control network. The overall goal of this research is to enable large-scale dynamically adaptive scientific and engineering simulations on distributed, heterogeneous and dynamic execution environments such as the computational "grid".
Today many applications are developed using distributed technologies such as cluster, cloud and grid computing. These applications demand more resources for computation and storage. They demand flexible scaling and im...
详细信息
ISBN:
(纸本)9781479930807
Today many applications are developed using distributed technologies such as cluster, cloud and grid computing. These applications demand more resources for computation and storage. They demand flexible scaling and improved performance. Application now days can make use of multiple nodes (machines) to get the tasks completed. In this paper we discuss the, implementation details of a grid computing framework known as GridSys. This framework provides a fast and easy way to program a grid. It can easily help the application break the problem to compute intensive tasks. The framework distributes these tasks to different nodes of the grid efficiently and easily aggregate the results of these tasks provide fault tolerance and reliability.
Data parallel frameworks become essential for training machine learning models. The classic Bulk Synchronous parallel (BSP) model updates the model parameters through pre-defined synchronization barriers. However, whe...
详细信息
ISBN:
(纸本)9781450391641
Data parallel frameworks become essential for training machine learning models. The classic Bulk Synchronous parallel (BSP) model updates the model parameters through pre-defined synchronization barriers. However, when a worker computes significantly slower than other workers, waiting for the slow worker will lead to excessive waste of computing resources. In this paper, we propose a novel proactive data-parallel (PDP) framework. PDP enables the parameter server to initiate the update of the model parameter. That is, we can perform the update at any time without pre-defined update points. PDP not only initiates the update but also determines when to update. The global decision on the frequency of updates will accelerate the training. We further propose asynchronous PDP to reduce the idle time caused by synchronizing parameter updates. We theoretically prove the convergence property of asynchronous PDP. We implement a distributed PDP framework and evaluate PDP with several popular machine learning algorithms including Multilayer Perceptron, Convolutional Neural network, K-means, and Gaussian Mixture Model. Our evaluation shows that PDP can achieve up to 20X speedup over the BSP model and scale to large clusters.
Computer architectures for high performance computing have traditionally been based on an assumption of one parallel application running alone on one machine. The current trend is, however, that huge computer installa...
详细信息
ISBN:
(纸本)9781424449217
Computer architectures for high performance computing have traditionally been based on an assumption of one parallel application running alone on one machine. The current trend is, however, that huge computer installations offer compute power to a set of users or customers, each demanding only a subset of the available compute resources. This places new requirements on the architecture, in that it must support dynamic partitioning of the resources into several virtual servers as demand changes. We introduce a novel framework which supports flexible formation of such virtual servers while preventing interference between the communication of different virtual servers. This paper investigates the impacts of a shared interconnection network on applications running on virtual compute servers. We show that the interconnect performance supplied to each job is highly unpredictable, and that a job can experience a performance degradation of 97% when its traffic interferes with the traffic of concurrent jobs. With a minor reduction in the utilization of each processing node, this can be considerably improved through a combination of routing-containment in the interconnection network and a carefully designed resource allocation strategy.
With the development of rural power market in our country, how to improve the operation more correctly and timely of power quantity collection, transmission and processing becomes a problem. The requirement of equipme...
详细信息
ISBN:
(纸本)9780387772509
With the development of rural power market in our country, how to improve the operation more correctly and timely of power quantity collection, transmission and processing becomes a problem. The requirement of equipment in power quantity collection needs new and higher demands. The advantage of using dual-port RAM and main-spare CPU structure, when main CPU running, spare CPU supply monitoring of main CPU status messages, once main CPU break down, spare CPU could instead of main CPU completely. This paper introduces collectivity design in flow chart, hardware design include theory and application in detail. Using dual-port RAM to communicate between the main and spare, CPU not only make sure the transmission efficiency, good anti-jamming performance, improve the speed of disposal, but also reduce the costs, making the operation of rural power network more security, economy and reliability. In rural power terminal system of power quantity collection has broad application prospects.
Space division multiplexing (SDM) is a promising solution with the scaling potential to overcome the possible capacity crunch problem in optical backbone networks. The key idea behind SDM is to exploit the spatial dim...
详细信息
ISBN:
(纸本)9783901882937
Space division multiplexing (SDM) is a promising solution with the scaling potential to overcome the possible capacity crunch problem in optical backbone networks. The key idea behind SDM is to exploit the spatial dimension to provide a significant increase in the transmission system capacity. In SDM, optical signals are transmitted in parallel through spatial resources (fibers, cores or modes), thus co-propagating in the same optical fiber structure. The goal of this paper is twofold. First, we propose and evaluate several versions of a greedy algorithm for optimization of flexible-grid SDM optical networks, including comparison with optimization results yielded by the CPLEX solver. Second, using best found algorithm settings, we complement our study with an analysis of SDM network performance in terms of the spectrum usage. We report and discuss results of numerical experiments run on a representative network topology with realistic physical assumptions. The key observation is the low scalability of the CPLEX solver, while the greedy algorithm is capable of generating solutions for larger network scenarios. However, even the relatively simple heuristic experiences quite large execution times.
This paper, as well as coupled paper [2], deal with various aspects of scheduling algorithms dedicated for processing in parallelcomputing environments. In this paper, for the exemplary problem, namely the flow-shop ...
详细信息
ISBN:
(纸本)9783540695721
This paper, as well as coupled paper [2], deal with various aspects of scheduling algorithms dedicated for processing in parallelcomputing environments. In this paper, for the exemplary problem, namely the flow-shop scheduling problem with makespan criterion, there are proposed original methods for parallel analysis of a solution as well as a group of concentrated and/or distributed solutions, recommended for the use in metaheuristic approaches with single-thread trajectory. Such methods examine in parallel consecutive local sub-areas of the solution space, or a set of distributed solutions called population, located along the single trajectory passed through the space. Supplementary multithread search techniques applied in metaheuristics have been discussed in complementary our paper [2].
The encryption process of LUC Cryptosystem is V-e(P,1)(mod N), while the decryption process is V-d(C,1)(mod N). An N is a product of two relatively primes p and q, P is a message and C is the ciphertext. To compute V-...
详细信息
ISBN:
(纸本)9780769532639
The encryption process of LUC Cryptosystem is V-e(P,1)(mod N), while the decryption process is V-d(C,1)(mod N). An N is a product of two relatively primes p and q, P is a message and C is the ciphertext. To compute V-e and V-d, this system used Lucas Function. The Lucas Functions is a special form of second order linear recurrence relation using a large public integer. Recently, methods for fast LUC Cryptosystem computation in sequential and parallel techniques have been developed. We are introducing an important property of Lucas Function, V2n+1=PV2n-QV(2n-1). This property could increase the performance of LUC Cryptosystem computation and reduce the computation time. We are implementing the method in parallel on distributed memory multiprocessor machine using message passing interface. In order to show a new parallel solution is better than the previous parallel solution;we will compare the simulation time, speedup and efficiency of both solutions.
Most of the current video cooperative surveillance strategies upload all video clips the camera takes, which will cause great data redundancy and bandwidth waste. In this paper, we combine temporal similarity with spa...
详细信息
ISBN:
(纸本)9783031213946;9783031213953
Most of the current video cooperative surveillance strategies upload all video clips the camera takes, which will cause great data redundancy and bandwidth waste. In this paper, we combine temporal similarity with spatial similarity and introduce the concept of spatiotemporal similarity. In particular, we design a framework to calculate spatial-temporal similarity to reduce the complexity of the collection and the transmission of source data. Besides, we model the problem of minimum spatiotemporal similarity with the bandwidth limitation into a knapsack problem and propose a dynamic programming-based algorithm to determine the selection of video uploading. The results show the framework can make a 10% data redundancy reduction and bandwidth saving.
暂无评论