A paralleldistributed scheme is presented for coding random patterns generated via iteration of contraction mappings. The scheme is implemented by a decentralized computational process an distributed parameter system...
详细信息
ISBN:
(纸本)0818678704
A paralleldistributed scheme is presented for coding random patterns generated via iteration of contraction mappings. The scheme is implemented by a decentralized computational process an distributed parameter system. The distributed parameter system generates a representation of missing probability for attractor as the basis of design of contraction mappings. The local minimals of missing probability is extracted and aggregated as the decentralized representation of contraction mappings. The coding scheme is verified through computer simulation.
This paper is concerned with distributed receding horizon prediction for continuous-time linear stochastic systems with multiple sensors. A distributed fusion with the weighted sum structure is applied to the optimal ...
详细信息
ISBN:
(纸本)9781424443475
This paper is concerned with distributed receding horizon prediction for continuous-time linear stochastic systems with multiple sensors. A distributed fusion with the weighted sum structure is applied to the optimal local receding horizon predictors. The distributed prediction algorithm represents the optimal linear fusion by weighting matrices under the minimum mean square criterion. The algorithm has the parallel structure and allows parallel processing of observations making it reliable since the rest faultless sensors can continue to the fusion estimation if some sensors occur faulty. The derivation of equations for error cross-covariances between the local predictors is the key of this paper. Example demonstrates effectiveness of the distributed receding horizon predictor.
Many organisations are now planning to move their operations from total reliance on centralised databases towards distributed environments which may involve the interoperability of a number of heterogeneous databases....
详细信息
Many organisations are now planning to move their operations from total reliance on centralised databases towards distributed environments which may involve the interoperability of a number of heterogeneous databases. This study looks at a particular case for an international financial institution, with the likely performance of a proposed globally distributed federated database being compared with the performance of the current centralised system. The performance model developed includes submodels for transaction structure and management, user workload, and distributed heterogeneous databases. Simulations focus on response times for a particular class of credit control/deal entry transactions in the presence of a background load. The results demonstrate that the proposed federated database outperforms the current centralised system, and that this is achievable using currently available technology.
In HPC supercomputers, the CPU, memory, network and storage play a critical role in application performance. Established benchmarks measure the theoretical peak performance of these components, but storage benchmarks ...
详细信息
ISBN:
(纸本)9798350369205;9798350369199
In HPC supercomputers, the CPU, memory, network and storage play a critical role in application performance. Established benchmarks measure the theoretical peak performance of these components, but storage benchmarks often focus solely on I/O and lack realism. To address this, we present numio, a benchmark that simulates and evaluates overlapping compute, communication and I/O phases. Furthermore, we present a novel ensemble-driven system benchmarking strategy. This approach involves running multiple benchmarks in parallel to analyse their interactions and assess the system's ability to handle the workload. Using a real-world example, we demonstrate how this approach reveals performance issues in complex HPC systems that remain hidden when using traditional methods using isolated benchmarks on empty systems.
Graph and sparse matrix systems are highly tuned, able to run complex graph analytics in fractions of seconds on billion-edge graphs. For both developers and researchers, the focus has been on computational kernels an...
详细信息
ISBN:
(纸本)9781665435772
Graph and sparse matrix systems are highly tuned, able to run complex graph analytics in fractions of seconds on billion-edge graphs. For both developers and researchers, the focus has been on computational kernels and not end-to-end runtime. Despite the significant improvements that modern hardware and operating systems have made towards input and output, these can still become application bottlenecks. Unfortunately, on high-performance shared-memory graph systems running billion-scale graphs, reading the graph from file systems easily takes over 2000x longer than running the computational kernel. This slowdown causes both a disconnect for end users and a loss of productivity for researchers and developers. We close the gap by providing a simple to use, small, header-only, and dependency-free C++11 library that brings I/O improvements to graph and matrix systems. Using our library, we improve the end-to-end performance for state-of-the-art systems significantly-in many cases by over 40x.
We propose a distributed protocol for a queue, called SKUEUE, which spreads its data fairly onto multiple processes, avoiding bottlenecks in high throughput scenarios. SKUEUE can be used in highly dynamic environments...
详细信息
ISBN:
(纸本)9781538643686
We propose a distributed protocol for a queue, called SKUEUE, which spreads its data fairly onto multiple processes, avoiding bottlenecks in high throughput scenarios. SKUEUE can be used in highly dynamic environments, through the addition of JOIN() and LEAVE() requests to the standard queue operations ENQUEUE() and DEQUEUE(). Furthermore SKUEUE satisfies sequential consistency in the asynchronous message passing model. Scalability is achieved by aggregating multiple requests to a batch, which can then be processed in a distributed fashion without hurting the queue semantics. Operations in SKUEUE need a logarithmic number of rounds w.h.p. until they are processed, even under a high rate of incoming requests.
The authors examine the issue of robust transaction routing in distributed database systems. A class of dynamic routing strategies which are estimated response times to make routing decisions is studied in details. Si...
详细信息
ISBN:
(纸本)0818608935
The authors examine the issue of robust transaction routing in distributed database systems. A class of dynamic routing strategies which are estimated response times to make routing decisions is studied in details. Since response time estimation and decision making depend on the assumed transaction model and parameers, it is important to examine the robustness or sensitivity to the inaccuracy in the assumptions and parameter values. Through simulations, the authors find that the dynamic routing strategy based strictly on response time is too aggressive in sharing loads and makes too many nonpreferred system routings. It is robust with respect to change in the number of database calls per transaction, but is relatively sensitive to the distribution of database calls. Two refinements are proposed which improve system performance as well as robustness of routing decisions.
In this paper, the performance and characteristics of the execution of various join-trees on a parallel DBMS are studied. The results of this study are a step into the direction of the design of a query optimization s...
详细信息
In this paper, the performance and characteristics of the execution of various join-trees on a parallel DBMS are studied. The results of this study are a step into the direction of the design of a query optimization strategy that is fit for parallel execution of complex queries. Among others, synchronization issues are identified to limit the performance gain from parallelism. A new hash-join algorithm is introduced that has fewer synchronization constraints than the known hash-join algorithms. Also, the behavior of individual join operations in a join-tree is studied in a simulation experiment. The results show that the introduced Pipelining hash-join algorithm yields a better performance for multi-join queries. The format of the optimal join-tree appears to depend on the size of the operands of the join: A multi-join between small operands performs best with a bushy schedule;larger operands are better off with a linear schedule. The results from the simulation study are confirmed with an analytic model for dataflow query execution.
In the past decade parallel disk systems have been highly, scalable and able to alleviate the problem of disk I/O bottleneck, thereby being widely used to support a wide range of data- intensive applications. Optimizi...
详细信息
ISBN:
(纸本)9780769534251
In the past decade parallel disk systems have been highly, scalable and able to alleviate the problem of disk I/O bottleneck, thereby being widely used to support a wide range of data- intensive applications. Optimizing energy consumption in parallel disk systems has strong impacts on the cost of backup power-generation and cooling equipment, because a significant fraction of the operation cost of data centres is due to energy consumption and cooling. Although a variety of parallel disk systems were developed to achieve high performance and energy efficiency, most existing parallel disk systems lack an adaptive way to conserve energy in dynamically changing work-load conditions. To solve this problem, we develop an adaptive energy-conserving algorithm, or DCAPS, for parallel disk systems using the dynamic voltage scaling technique that dynamically choose the most appropriate voltage supplies for parallel disks while guaranteeing specified performance (i.e., desired response times)for disk requests. We conduct extensive experiments to quantitatively evaluate the performance of the proposed energy-conserving strategy. Experimental results consistently show that DCAPS significantly reduces energy consumption of parallel disk systems in a dynamic environment over the same disk systems without using the DCAPS strategy.
暂无评论