A novel approach for extracting coarse-grained parallelism being represented with independent and synchronization-requiring slices is presented. Each slice is composed of dependent iterations of perfectly nested loops...
详细信息
ISBN:
(纸本)9780769529172
A novel approach for extracting coarse-grained parallelism being represented with independent and synchronization-requiring slices is presented. Each slice is composed of dependent iterations of perfectly nested loops. Presented algorithms work for both uniform and non-uniform loops. Our approach, based on operations on relations and sets, requires exact dependence analysis. Examples illustrating the proposed algorithm and results of experiments are presented.
Regular path queries are the building block of almost any mechanism for querying semistructured data. Despite the fact that the main applications of such data are distributed, there are only few works dealing with dis...
详细信息
ISBN:
(纸本)9783540747666
Regular path queries are the building block of almost any mechanism for querying semistructured data. Despite the fact that the main applications of such data are distributed, there are only few works dealing with distributed evaluation of regular path queries. In this paper we present a message-efficient and truly distributed algorithm for computing the answer to regular path queries in a multi-source semistructured database setting. Our algorithm is general as it works for the larger class of weighted regular path queries on weighted (as well) semistructured databases.
The Caravela platform has been proposed by the authors of this paper to perform distributed stream-based computing on general purpose computation. This platform uses a secured execution unit called flow-model that pre...
详细信息
ISBN:
(纸本)9780769529172
The Caravela platform has been proposed by the authors of this paper to perform distributed stream-based computing on general purpose computation. This platform uses a secured execution unit called flow-model that prevents remote users to touch local information in a computer The flow-model is assigned to local or remote processing units that execute its program. This paper is focused on a new execution mechanism that defines a pipeline composed by flow-models, called meta-pipeline, and is designed as a set of additional functions of the Caravela platform. The pipeline is executed automatically by the meta-pipeline runtime environment. This paper describes the execution mechanism and also presents an application example.
Mixed-parallel applications can take advantage of large-scale computing platforms but scheduling them efficiently on such platforms is challenging. In this paper we compare the two main proposed approaches for solving...
详细信息
ISBN:
(纸本)9780769529172
Mixed-parallel applications can take advantage of large-scale computing platforms but scheduling them efficiently on such platforms is challenging. In this paper we compare the two main proposed approaches for solving this scheduling problem on a heterogeneous set of homogeneous clusters. We first modify previously proposed algorithms for both approaches and show that our modifications lead to significant improvements. We then perform a comparison of the modified algorithms in simulation over a wide range of application and platform conditions. We find that although both approaches have advantages, one of them is most likely the most appropriate for the majority of users.
The Evolutionary Geometric Near-neighbor Access Tree (EGNAT) is a recently proposed data structure that is suitable for indexing large collections of complex objects. It allows searching for similar objects represente...
详细信息
ISBN:
(纸本)9780889866379
The Evolutionary Geometric Near-neighbor Access Tree (EGNAT) is a recently proposed data structure that is suitable for indexing large collections of complex objects. It allows searching for similar objects represented in metric spaces. The sequential EGNAT has been shown to achieve good performance in high-dimentional metric spaces with properties (not found in others of the same kind) of allowing update operations and efficient use of secondary memory. Thus, for example, it is suitable for indexing large multimedia databases. However, comparing two objects during a search can be a very expensive operation in terms of running time. This paper shows that parallel computing upon clusters of PCs can be a practical solution for reducing running time costs. We describe alternative distributions for the EGNAT index and their respective parallel search/update algorithms and concurrency control mechanism.
Monitoring distributed applications executed on a computational Grid is challenging since they are executed on several heterogeneous nodes belonging to different administrative domains. An application monitoring servi...
详细信息
ISBN:
(纸本)9780769529172
Monitoring distributed applications executed on a computational Grid is challenging since they are executed on several heterogeneous nodes belonging to different administrative domains. An application monitoring service should provide users and administrators with useful and dependable data on the applications executed on the Grid. We present in this paper a grid application monitoring service designed to handle high availability and scalability issues. The service supplies information on application state, on failures and on resource consumption. A set of transparent monitoring mechanisms are used according to grid node nature to effectively monitor the applications. Experiments on the Grid'5000 testbed show that the service provides dependable information with a minimal cost on Grid performances.
Model checking, logging, debugging, and checkpointing/recovery are great tools to identify bugs in small sequential programs. The direct application of these techniques to the domain of distributed applications, howev...
详细信息
We present a new data partitioning strategy for parallel computing on three interconnected clusters. This partitioning has two advantages over existing partitionings. First it can reduce communication time due to a lo...
详细信息
ISBN:
(纸本)9780769529172
We present a new data partitioning strategy for parallel computing on three interconnected clusters. This partitioning has two advantages over existing partitionings. First it can reduce communication time due to a lower total volume of communication and a more efficient communication schedule. When the network topology is a linear array this partitioning always results in a lower total volume of communication compared to existing partitionings, provided the most powerful node is at the center of the array. When the topology is fully connected this partitioning results in a lower total volume of communication for all but a few power ratios. Second, it allows for the overlapping of communication and computation. These two inherent advantages work together to reduce overall execution time significantly.
Audio search plays an important role in analyzing audio data and retrieving useful audio information. In this paper, a Partially Overlapping Block-parallel Active Search method (POBPAS) is proposed to perform audio qu...
详细信息
Compositional performance analysis iteratively alternates local scheduling analysis techniques and output event model propagation between system components to enable performance analysis of heterogeneous distributed s...
详细信息
暂无评论