This paper presents PIT, a library for the parallelization of irregular problems on distributed memory architectures. All the strategies underlying the definition of the library can be expressed in terms of operations...
详细信息
This paper presents PIT, a library for the parallelization of irregular problems on distributed memory architectures. All the strategies underlying the definition of the library can be expressed in terms of operations on a PITree, a parallel version of the tree data structure oriented to irregular problems. We consider the application of PIT to two well known irregular problems: adaptive multigrid and hierarchical radiosity methods. Performance figures that prove the effectiveness of PIT are presented.
This paper identifies the need for multiple clients to concurrently access a device community. A set of embedded devices cooperate via one or more service discovery protocols, such as UPnP, Jini, HAVi or OSGi. We pres...
详细信息
This paper identifies the need for multiple clients to concurrently access a device community. A set of embedded devices cooperate via one or more service discovery protocols, such as UPnP, Jini, HAVi or OSGi. We present an approach that encapsulates additional services into a distributed Access Framework (DAF). The services in DAF are responsible for secure and coordinated access of the distributed devices in the community. The different services of DAF can optionally be implemented and executed on different devices. We specify the DAF and demonstrate an implementation where the devices are organized according to the OSGi specification.
By adopting code mobility techniques, distributed applications can make better use of network resources. This makes logical mobility particularly attractive in a mobile computing scenario, where the level of network r...
详细信息
By adopting code mobility techniques, distributed applications can make better use of network resources. This makes logical mobility particularly attractive in a mobile computing scenario, where the level of network resources changes continuously. However, it is sometimes difficult to express quantitatively the benefits, and also the overhead, introduced by logical mobility. In this paper, we present the performance evaluation of a migration model where the central component of the application, the server, moves towards the physical barycenter of a network made of mobile hosts. We evaluated both application-level metrics, such as the percentage of delivered messages and communication latency, and network-level metrics, such as the number of routed messages. In many cases, logical mobility contributed to a significant improvement of the considered performance indexes.
A distributed object systems is said to be K-tolerant if every object is available after the simultaneous failure of up to K nodes. The problem is that a K-tolerant system, after failures, is no longer K-tolerant;that...
详细信息
ISBN:
(纸本)088986568X
A distributed object systems is said to be K-tolerant if every object is available after the simultaneous failure of up to K nodes. The problem is that a K-tolerant system, after failures, is no longer K-tolerant;that is, subsequent failures may compromise the availability of the objects. A continuously K-tolerant system is one which starting from a K-tolerant configuration, after the failure of up to K nodes, reconfigures itself so to remain K-tolerant. The existing protocols for maintaining continuous K-tolerance do so without regard to the resulting structure of the available data. For example if the distributed set of objects was sorted, this ordering would be most likely lost after restructuring. Analogously, a balanced distribution of the objects among the nodes might also be not achieved in the new distribution after reorganization. In this paper, we present a mechanism for maintaining continuous K-tolerance while keeping the load balanced and the objects sorted. The proposed solution uses minimum amount of replication and has a cost comparable to the one of the known unstructured solutions.
This paper revisits the interconnection of self-stabilization and fault-tolerance. Self-stabilizing algorithms are able to recover from arbitrary system states given that from some point in time on, there are no fault...
详细信息
This paper revisits the interconnection of self-stabilization and fault-tolerance. Self-stabilizing algorithms are able to recover from arbitrary system states given that from some point in time on, there are no faults. Fault-tolerance, on the other hand, refers to algorithms that cope with systems where a (bounded) part of the system (e.g. at most f out of n processes) may fail permanently. In previous work [16] we considered the interconnection of these two paradigms, i.e., algorithms that recover from arbitrary states despite of permanent faults. We have shown that in certain settings, problems as failure detection cannot be solved. This paper presents ways to circumvent this impossibility result.
Modern methods in molecular biology produce a tremendous amount of data. As a consequence, efficient methods have to be developed to retrieve and analyze these data. In this article, a parallel algorithm searching mot...
详细信息
Modern methods in molecular biology produce a tremendous amount of data. As a consequence, efficient methods have to be developed to retrieve and analyze these data. In this article, a parallel algorithm searching motifs based on ParSeq, a software tool for motif search, is put forward and its performance analyzed and discussed. Many experiments show that with this parallel algorithm a considerable improvement upon the sequential motif search algorithm is achieved.
This paper introduces SCCB1 cluster system in which SCC PCI boards are implemented. SCCB cluster is built using PCs(32nodes) for processor, and two networks. SCC PCI boards allow to fast barrier synchronization mechan...
详细信息
This paper introduces SCCB1 cluster system in which SCC PCI boards are implemented. SCCB cluster is built using PCs(32nodes) for processor, and two networks. SCC PCI boards allow to fast barrier synchronization mechanism and *** order to perform fast synchronization, each SCC board is connected by a new type of network, and checks completion of barrier by hardware. PGCCS allows us to measure execution times accurately on the Beowulf clusters. In performance evaluation, SCC board plugged into 32-nodes PC cluster achieves barrier latency of 3.2 μ8. Time difference errors of PGCCS were smaller (1/100) than that of a conventional method using MPLBarrier.
This article presents a new distributed approach for generating all prime numbers up to a given limit. From Eratosthenes, who elaborated the first prime sieve (more than 2000 years ago), to the advances of the paralle...
详细信息
This article presents a new distributed approach for generating all prime numbers up to a given limit. From Eratosthenes, who elaborated the first prime sieve (more than 2000 years ago), to the advances of the parallel computers (which have permitted to reach large limits or to obtain the previous results in a shorter time), prime numbers generation still represents an attractive domain of research. Nowadays, prime numbers play a central role in cryptography and their interest has been increased by the very recent proof that primality testing is in P. In this work, we propose a new distributed algorithm which generates all prime numbers in a given finite interval [2,..., n], based on the wheel sieve. As far as we know, this paper designs the first fully distributed wheel sieve algorithm.
The Internet is showing a potential of forming of a supercomputing resource out of networked computers. It is to harness the combined resources of millions of computers connected to the Internet, forming a powerful he...
详细信息
The Internet is showing a potential of forming of a supercomputing resource out of networked computers. It is to harness the combined resources of millions of computers connected to the Internet, forming a powerful heterogeneous computing environment for running coarse-grain parallel applications. The paralleldistributed processing has the high cost-performance ratio application on the Internet. The Internet also has some of unpredictable varieties such as heterogeneity of hosts, various states of hosts, performance of each host to be support continuously, and the number of hosts which are participated in computation and so on. For considering the property of Internet, this paper should provide Grid Service Architecture called paralleldistributed Processing(PDP) which is a parallelcomputing framework implemented with Java for parallelcomputing over the Internet. Therefore, for the parallel processing on the Internet, we propose the strategy of uniform task allocation, CPU performance task allocation, and adaptive task allocation based on CPU performance. This paper also shows an algorithm of dynamic host management, which is changed by the number of host during execution time on PDP. Finally, this paper presents the architecture, the message passing protocol and control flow of PDP and shows the results of performance evaluation for cockpit algorithms on PDP.
distributed deadlock detection is one of the most important problems in distributed systems. Over the two past decades many algorithms have been proposed in the literature to detect deadlock in distributed systems. Al...
详细信息
distributed deadlock detection is one of the most important problems in distributed systems. Over the two past decades many algorithms have been proposed in the literature to detect deadlock in distributed systems. Although many of them fail to detect deadlocks or report deadlocks that does not exist (false deadlocks) and have been proved that are incorrect. The most important and applicable class of distributed deadlock detection algorithms is probe-based algorithms. In this paper existing distributed probe-based deadlock detection algorithms have been classified and a survey including correctness of them and their constraints and conditions have been introduced. Also an improvement for removing of false deadlock with considering process termination or transaction abortion in this class of algorithms has been presented.
暂无评论