Gild Technology is the great progress of network after Internet, since gild enables the integrated and collaborative use of distributedcomputing resources owned and managed by multiple organizations, available over a...
详细信息
ISBN:
(纸本)9780769527369
Gild Technology is the great progress of network after Internet, since gild enables the integrated and collaborative use of distributedcomputing resources owned and managed by multiple organizations, available over a local or wide area network. In this paper, we provided a Data Grid platform with integrated storage to solve such data gridproblems by using PVFS2 to increase the data storage space. the experimental results presented show the effectiveness of such proposed combination of technologies.
In the applications of global computing like SETI@home, the scheduling problem of computation components is an important issue to improve the performance. In this paper, we propose a theoretical model of global comput...
详细信息
ISBN:
(纸本)9780769527369
In the applications of global computing like SETI@home, the scheduling problem of computation components is an important issue to improve the performance. In this paper, we propose a theoretical model of global computing application concerning the scheduling of computation components. Based on the model, we evaluate the performance of different scheduling mechanisms to choose a proper scheduling strategy for the application according to the corresponding network environment and computing resources. Finally we make the comparisons among different scheduling strategies and suggestions to improve the performance of the application.
In this paper we present a new distributed Group Mutual Exclusion(DGME) based on Clients/Servers model, and uses a dynamic data structures. Several processes (Clients) can access simultaneously to a same opened sessio...
详细信息
ISBN:
(纸本)9780769527369
In this paper we present a new distributed Group Mutual Exclusion(DGME) based on Clients/Servers model, and uses a dynamic data structures. Several processes (Clients) can access simultaneously to a same opened session (Server). the algorithm ensures that, at any time, at most one session is opened, and any requested Session will be opened in a finite time. the number of messages is between 0 and m, where m is the number of session in the network. In the average case, O(Log(m)) messages are necessary to open a session. the maximum concurrency is n, where n is the number of processes in the network.
this paper proposes a novel task scheduling algorithm to exploit the potential of parallel processing, allowing for system heterogeneity in distributed heterogeneous computing environments. Its goal is to achieve maxi...
详细信息
ISBN:
(纸本)3540290672
this paper proposes a novel task scheduling algorithm to exploit the potential of parallel processing, allowing for system heterogeneity in distributed heterogeneous computing environments. Its goal is to achieve maximizing parallelization and minimizing Communication. Due to that the algorithm avoids from the max-min anomaly in the parallelization problem and exploits schedule holes, it could produce better schedules than those obtained by existing algorithms. Experimental results are presented to verify the preceding claims. three comparative algorithms are applied to demonstrate the proposed algorithm's effectiveness. As the system's heterogeneity increases, the performance improvement of the proposed algorithm becomes more outstanding than that of others. therefore, the proposed scheduling algorithm may be used in designing efficient parallel environments for those situations where the system heterogeneity is the system performance bottleneck.
the proceedings contain 128 papers. the topics discussed include: fast and reliable random number generators for scientific computing;large-scale computations withthe unified Danish Eulerian model;a chemical engineer...
详细信息
ISBN:
(纸本)3540290672
the proceedings contain 128 papers. the topics discussed include: fast and reliable random number generators for scientific computing;large-scale computations withthe unified Danish Eulerian model;a chemical engineering challenge problem that can benefit from interval methods;interval based Markov Decision processes;a verification method for solutions of linear programming problems;on the approximation of interval functions;the distributed interval geometric machine model;new algorithms for statistical analysis of interval data;on efficiency of tightening bonds in interval global optimization;applying software testing matrices to lapack;parallel algorithms for balanced truncation model reduction of sparse systems;and applying high performance computing techniques in astrophysics.
Grid computing is a new paradigm for distributedcomputing, and service has become building block of grid applications. However, current approaches can not free developers from low-level laborious work when building g...
详细信息
ISBN:
(纸本)9780769527369
Grid computing is a new paradigm for distributedcomputing, and service has become building block of grid applications. However, current approaches can not free developers from low-level laborious work when building grid applications. We propose a service-oriented virtual machine called Abacus Virtual Machine to simplify the task of grid application development. As a language level virtual machine, it provides a service-oriented instruction set to abstract the operations on the services of a grid application. It also virtualizes services and creates a virtual global system image for grid applications, thus services can be transparently distributed and shared In this way, Abacus Virtual Machine hides the cumbersome underlying details from programmers and reduces the complexity greatly in grid application development.
the knapsack problem is a famous NP-complete problem. It is very important in the research on cryptosystem and number theory. After its proposed parallel algorithms are analyzed deeply, a new parallel algorithm by sam...
详细信息
ISBN:
(纸本)9780769527369
the knapsack problem is a famous NP-complete problem. It is very important in the research on cryptosystem and number theory. After its proposed parallel algorithms are analyzed deeply, a new parallel algorithm by sampling is proposed based on MIMD supercomputers in the paper . then performance analysis and comparisons are illuminated Finally the experimental results of the knapsack instances randomly generated on IBM P690 supercomputer are given. the results show: the parallel efficiency can be over 60% when solving the larger scale knapsack instances (n >= 40). thus it is proved that the proposed parallel algorithm for the knapsack problem is feasible and efficient on MIMD scalable supercomputers.
Withthe development of mobile devices and wireless communication technologies, users and devices will face an environment full of changes. It is indispensable for a system to adapt to the environment changes, which w...
详细信息
ISBN:
(纸本)9780769527369
Withthe development of mobile devices and wireless communication technologies, users and devices will face an environment full of changes. It is indispensable for a system to adapt to the environment changes, which we call context-awareness. this paper introduces the adaptive mechanisms for embedded systems from the point of view of the data. the data in a context-aware system are in three different forms in different stage: the sensor data, the context, the situation. this paper introduces the context model which consists of three layers and the adaptive mechanisms in different layer to enable the system adaptive to contexts. Besides, we develop an example to show how the clustering and reasoning technology are used to recognize and derive the higher level context.
One important way of solving 110 bottleneck is to develop parallel file system. Research showed that the strided file access pattern in scientific applications combined with current file system methods to perform thes...
详细信息
ISBN:
(纸本)9780769527369
One important way of solving 110 bottleneck is to develop parallel file system. Research showed that the strided file access pattern in scientific applications combined with current file system methods to perform these accesses lead to unacceptable performance for large data access. To enhance performance of discontiguous data access, we have created a new high performance 110 method: the user-defined file view and combining 110 requests (UFCR). We have implemented our ideas on WPFS. Experimentation results showed that our method can substantially enhance the performance of scientific applications.
the technology and performance improvements of computational architectures and communication infrastructure, together withthe increase of the demand for a greater and more easily usable computing power, generate new ...
ISBN:
(纸本)3540290672
the technology and performance improvements of computational architectures and communication infrastructure, together withthe increase of the demand for a greater and more easily usable computing power, generate new sophisticated requirements for distributed application development. the methodologies, infrastructures and tools traditionally used to develop distributed applications disclose a variety of limitations when applied to these new and evolving scenarios. the goal of this minisymposium is to investigate these issues, presenting some innovative approaches in managing the complexity in the design, development and performance analysis of distributed systems. Papers included in this minisymposium primary face the challenges of using the computing Grids as a commonly accepted general purpose computing platform, but the focus is also on paradigms and tools to support the programmers in developing efficient parallel solutions using MPI and OpenMP libraries. Furthermore, particular emphasis is given to some significant aspects of industrial and commercial distributed applications such as, the mobility, the dependability and the security.
暂无评论