Automatic Partitioning of Large Scale Simulation in Grid Computing for Run Time Reduction

2010 ◽  
Vol 1 (2) ◽  
pp. 64-90 ◽  
Author(s):  
Nurcin Celik ◽  
Esfandyar Mazhari ◽  
John Canby ◽  
Omid Kazemi ◽  
Parag Sarfare ◽  
...  
Keyword(s):  
Execution Time ◽  
Large Scale ◽  
Time Synchronization ◽  
Computational Grid ◽  
Time Interval ◽  
Computational Power ◽  
Computing Environment ◽  
Large Scale Systems ◽  
Large Scale Simulations ◽  
Reduce Execution Time

Simulating large-scale systems usually entails exhaustive computational powers and lengthy execution times. The goal of this research is to reduce execution time of large-scale simulations without sacrificing their accuracy by partitioning a monolithic model into multiple pieces automatically and executing them in a distributed computing environment. While this partitioning allows us to distribute required computational power to multiple computers, it creates a new challenge of synchronizing the partitioned models. In this article, a partitioning methodology based on a modified Prim’s algorithm is proposed to minimize the overall simulation execution time considering 1) internal computation in each of the partitioned models and 2) time synchronization between them. In addition, the authors seek to find the most advantageous number of partitioned models from the monolithic model by evaluating the tradeoff between reduced computations vs. increased time synchronization requirements. In this article, epoch- based synchronization is employed to synchronize logical times of the partitioned simulations, where an appropriate time interval is determined based on the off-line simulation analyses. A computational grid framework is employed for execution of the simulations partitioned by the proposed methodology. The experimental results reveal that the proposed approach reduces simulation execution time significantly while maintaining the accuracy as compared with the monolithic simulation execution approach.

Automatic Partitioning of Large Scale Simulation in Grid Computing for Run Time Reduction

2012 ◽  
pp. 225-252
Author(s):  
Nurcin Celik ◽  
Esfandyar Mazhari ◽  
John Canby ◽  
Omid Kazemi ◽  
Parag Sarfare ◽  
...  
Keyword(s):  
Execution Time ◽  
Large Scale ◽  
Time Synchronization ◽  
Computational Grid ◽  
Experimental Results ◽  
Time Interval ◽  
Computational Power ◽  
Large Scale Systems ◽  
Large Scale Simulations ◽  
Reduce Execution Time

Simulating large-scale systems usually entails exhaustive computational powers and lengthy execution times. The goal of this research is to reduce execution time of large-scale simulations without sacrificing their accuracy by partitioning a monolithic model into multiple pieces automatically and executing them in a distributed computing environment. While this partitioning allows us to distribute required computational power to multiple computers, it creates a new challenge of synchronizing the partitioned models. In this article, a partitioning methodology based on a modified Prim’s algorithm is proposed to minimize the overall simulation execution time considering 1) internal computation in each of the partitioned models and 2) time synchronization between them. In addition, the authors seek to find the most advantageous number of partitioned models from the monolithic model by evaluating the tradeoff between reduced computations vs. increased time synchronization requirements. In this article, epoch- based synchronization is employed to synchronize logical times of the partitioned simulations, where an appropriate time interval is determined based on the off-line simulation analyses. A computational grid framework is employed for execution of the simulations partitioned by the proposed methodology. The experimental results reveal that the proposed approach reduces simulation execution time significantly while maintaining the accuracy as compared with the monolithic simulation execution approach.

Automatic Partitioning of Large Scale Simulation in Grid Computing for Run Time Reduction

2012 ◽  
pp. 380-406
Author(s):  
Nurcin Celik ◽  
Esfandyar Mazhari ◽  
John Canby ◽  
Omid Kazemi ◽  
Parag Sarfare ◽  
...  
Keyword(s):  
Execution Time ◽  
Large Scale ◽  
Time Synchronization ◽  
Computational Grid ◽  
Experimental Results ◽  
Time Interval ◽  
Computational Power ◽  
Large Scale Systems ◽  
Large Scale Simulations ◽  
Reduce Execution Time

Simulating large-scale systems usually entails exhaustive computational powers and lengthy execution times. The goal of this research is to reduce execution time of large-scale simulations without sacrificing their accuracy by partitioning a monolithic model into multiple pieces automatically and executing them in a distributed computing environment. While this partitioning allows us to distribute required computational power to multiple computers, it creates a new challenge of synchronizing the partitioned models. In this article, a partitioning methodology based on a modified Prim’s algorithm is proposed to minimize the overall simulation execution time considering 1) internal computation in each of the partitioned models and 2) time synchronization between them. In addition, the authors seek to find the most advantageous number of partitioned models from the monolithic model by evaluating the tradeoff between reduced computations vs. increased time synchronization requirements. In this article, epoch- based synchronization is employed to synchronize logical times of the partitioned simulations, where an appropriate time interval is determined based on the off-line simulation analyses. A computational grid framework is employed for execution of the simulations partitioned by the proposed methodology. The experimental results reveal that the proposed approach reduces simulation execution time significantly while maintaining the accuracy as compared with the monolithic simulation execution approach.

Realistic Modeling of Resources for a Large-Scale Computational Grid System

2012 ◽  
Vol 4 (3) ◽  
pp. 1-29
Author(s):  
Uei-Ren Chen ◽  
Yun-Ching Tang
Keyword(s):  
Large Scale ◽  
Simulated Data ◽  
Computational Grid ◽  
Computing Environment ◽  
Resource Model ◽  
Grid Resource ◽  
Normal Distributions ◽  
Realistic Modeling ◽  
Source Data ◽  
Grid Resources

To build a large-scale computational grid resource model with realistic characteristics, this paper proposes a statistical remedy to approximate the distributions of computational and communicational abilities of resources. After fetching the source data of resources including computing devices and networks from the real world, the gamma and normal distributions are employed to approximate probabilities for the ability distribution of grid resources. With this method, researchers can supply the computational grid simulators with required parameters relating to computational and communicational abilities. The experiment result shows that proposed methodology can lead to a better precision according to the error measurement of the simulated data and its source. The proposed method can support existing modern grid simulators with a high-precision resource model which can consider the characteristics of distribution for its computational and communicational abilities in the grid computing environment.

Improved Checkpoint Using the Effective Management of I/O in a Cloud Environment

2019 ◽  
pp. 219-233
Author(s):  
Bakhta Meroufel ◽  
Ghalem Belalem
Keyword(s):  
Fault Tolerance ◽  
Execution Time ◽  
Large Scale ◽  
Fault Tolerant ◽  
Experimental Results ◽  
Cloud Environment ◽  
Large Scale Systems ◽  
Major Disadvantage ◽  
Reliability And Robustness ◽  
Effective Use

One of the most important points for more effective use in the environment of cloud is undoubtedly the study of reliability and robustness of services related to this environment. In this case, fault tolerance is necessary to ensure that reliability and reduce the SLA violation. Checkpointing is a popular fault tolerance technique in large-scale systems. However, its major disadvantage is the overhead caused by the storage time of checkpointing files, which increases the execution time and minimizes the possibility to meet the desired deadlines. In this chapter, the authors propose a checkpointing strategy with lightweight storage. The storage is provided by creating a virtual topology VRbIO and the use of an intelligent and fault tolerant I/O technique CSDS (collective and selective data sieving). The proposal is executed by active and reactive agents and it solves many problems of checkpointing with standard I/O. To evaluate the approach, the authors compare it with a checkpointing with ROMIO as I/O strategy. Experimental results show the effectiveness and reliability of the proposed approach.

An Efficient Specification for System Verification

2006 ◽  
Vol 10 (6) ◽  
pp. 931-938
Author(s):  
Chikatoshi Yamada ◽  
◽  
Yasunori Nagata ◽  
Zensho Nakao ◽  
Keyword(s):  
Model Checking ◽  
Execution Time ◽  
Large Scale ◽  
Large Scale Systems ◽  
Computation Tree Logic ◽  
Computation Tree ◽  
Specification Method ◽  
System Verification ◽  
Temporal Formula ◽  
Temporal Specification

In design of complex and large scale systems, system verification has played an important role. In this article, we focus on specification process of model checking in system verifications. Modeled systems are in general specified by temporal formulas of computation tree logic, and users must know well about temporal specification because the specification might be complex. We propose a method by which specifications with temporal formulas are obtained inductively. We will show verification results using the proposed temporal formula specification method, and show that amount of memory, OBDD nodes, and execution time are reduced.

A Novel Approach to Publishing Tasks for Collaboratively Crowdsourcing Workflows

2019 ◽  
Vol 29 (06) ◽  
pp. 763-790
Author(s):  
Wenjun Tang ◽  
Rong Chen ◽  
Shikai Guo
Keyword(s):  
Execution Time ◽  
Large Scale ◽  
Approximate Solutions ◽  
Constraint Solving ◽  
Total Cost ◽  
Large Scale Problems ◽  
Novel Approach ◽  
Heuristic Strategies ◽  
Time Requirements ◽  
Reduce Execution Time

In recent years, crowdsourcing has gradually become a promising way of using netizens to accomplish tiny tasks on, or even complex works through crowdsourcing workflows that decompose them into tiny ones to publish sequentially on the crowdsourcing platforms. One of the significant challenges in this process is how to determine the parameters for task publishing. Still some technique applied constraint solving to select the optimal tasks parameters so that the total cost of completing all tasks is minimized. However, experimental results show that computational complexity makes these tools unsuitable for solving large-scale problems because of its excessive execution time. Taking into account the real-time requirements of crowdsourcing, this study uses a heuristic algorithm with four heuristic strategies to solve the problem in order to reduce execution time. The experiment results also show that the proposed heuristic strategies produce good quality approximate solutions in an acceptable timeframe.

Decentralized iterative learning control for switched large-scale systems

2018 ◽  
Vol 41 (4) ◽  
pp. 1045-1056
Author(s):  
Panpan Gu ◽  
Senping Tian ◽  
Qian Liu
Keyword(s):  
Iterative Learning Control ◽  
Large Scale ◽  
Contraction Mapping ◽  
Learning Control ◽  
Mapping Method ◽  
Iterative Learning ◽  
Control Law ◽  
Time Interval ◽  
Large Scale Systems ◽  
Numerical Examples

This paper is concerned with the iterative learning control problem for switched large-scale systems. According to the characteristics of the systems, a decentralized D-type iterative learning control law is proposed for such switched large-scale systems. The proposed controller of each subsystem relies only on local output variables, without any information exchanges with other subsystems. By using the contraction mapping method, it is shown that the algorithm can guarantee that the output of each subsystem converges to the desired trajectory over the whole time interval along the iteration axis. Finally, three numerical examples are given to illustrate the effectiveness of the proposed algorithm.

Sign in / Sign up

Export Citation Format

Share Document