Data distribution algebras are an abstract notion for the description of parallel programs. Their dynamic execution can be optimized if they are shapely. In this paper we describe a shape analysis which allows compile...
详细信息
Data distribution algebras are an abstract notion for the description of parallel programs. Their dynamic execution can be optimized if they are shapely. In this paper we describe a shape analysis which allows compile-time shapeliness-tests. It operates on the structure of algebraic data types and works for arbitrary functional programs rather than only shapely ones. Besides a first-order calculus we also propose a higher-order version which can handle higher-order functions as well. (C) 2000 Elsevier Science B.V. All rights reserved.
A novel parallel scheduling algorithm known as the parallel solitary-request-first (PSRF) algorithm is proposed to improve the throughput performance of a multiple input-queued (MIQ) switch when thr number of queues i...
详细信息
A novel parallel scheduling algorithm known as the parallel solitary-request-first (PSRF) algorithm is proposed to improve the throughput performance of a multiple input-queued (MIQ) switch when thr number of queues is less than the switch size.
We provide evidence that the saccadic system can simultaneously program two saccades to different goals. We presented subjects with simple visual search displays in which they were required to make a saccade to an odd...
详细信息
We provide evidence that the saccadic system can simultaneously program two saccades to different goals. We presented subjects with simple visual search displays in which they were required to make a saccade to an odd-colored target embedded in an array of distracters. When there was strong competition between target and distracter stimuli (due to color priming from previous trials), subjects were more likely to make a saccade to a distracter. Such error saccades were often followed, after a very short inter-saccadic interval (similar to 10-100 ms), by a second saccade to the target. The brevity of these inter-saccadic intervals suggests that the programming of the two saccades (one to a distracter and one to the target) overlapped in time. Using a saccade-contingent change in the search display, we show that new visual information presented during the initial saccade does not change the goal of the second saccade. This supports the idea that, by the end of the first saccade, programming of the second saccade is already well underway. We also elicited two-saccade responses (similar to those seen in search) using a double-step task, with the first saccade directed to the initial target step and the second saccade directed to the second target step. If the two saccades are programmed in parallel and programming of each saccade is triggered by one of the two target steps, the second saccade should occur at a relatively fixed time after the onset of the second target step, regardless of the timing of the initial saccade. This prediction was confirmed, supporting the idea that the two saccades are programmed in parallel. Finally, we observed that the shortest inter-saccadic intervals typically followed hypometric initial saccades, suggesting that the initial saccade may have been interrupted by the impending second saccade. Using predictions from physiological studies of interrupted saccades, we tested this hypothesis and found that the hypometric initial saccades did no
A number of issues relating to the implementation of a parallel finite-element program are discussed, including choice of programming language (principally FORTRAN90 versus C++), data-communication environment, matrix...
详细信息
A number of issues relating to the implementation of a parallel finite-element program are discussed, including choice of programming language (principally FORTRAN90 versus C++), data-communication environment, matrix partitioner, and parallel preconditioner. Some results computed for an intricately shaped load in a microwave applicator are presented.
A new scheme for parallel process synchronization is presented. For expressing synchronization conditions and associated control flow in a process code an interrupt mechanism is used. As a result, a passive waiting ca...
详细信息
ISBN:
(纸本)076950759X
A new scheme for parallel process synchronization is presented. For expressing synchronization conditions and associated control flow in a process code an interrupt mechanism is used. As a result, a passive waiting can be easily avoided Interrupts are triggered when a specified condition becomes true. The conditions can reflect the status of a whole parallel application. To define the application status processes assign values To "state" variables. The variables can be read by the conditions. Conditions encapsulate the code responsible for synchronization and they let a programmer to construct customized, application specific synchronization primitives.
Most Java-based systems that support portable parallel and distributed computing either require the programmer to deal with intricate low-level details of Java which can be a tedious, time-consuming and error-prone ta...
详细信息
ISBN:
(纸本)0769508960
Most Java-based systems that support portable parallel and distributed computing either require the programmer to deal with intricate low-level details of Java which can be a tedious, time-consuming and error-prone task, or prevent the programmer from controlling locality of data. In this paper we describe Java Symphony, a programming paradigm for distributed and parallel computing that provides a software infrastructure for wide classes of heterogeneous systems ranging from small-scale cluster computing to large scale wide-area meta-computing. The software infrastructure is written entirely in Java and runs on any standard compliant Java virtual machine. In contrast to most existing systems, Java Symphony provides the programmer with the flexibility to control data locality and load balancing by explicit mapping of objects to computing nodes. Virtual architectures are specified to impose a virtual hierarchy on a distributed system of physical computing nodes. Objects can be mapped and dynamically migrated to arbitrary components of virtual architectures. A high-level API to hardware/software system parameters is provided to control mapping, migration, and load balancing of objects. Objects can interact through synchronous asynchronous and one-sided method invocation. Selective remote class loading may reduce the overall memory requirement of an application. Moreover;objects can be made persistent by explicitly storing and loading objects to/from external storage. A prototype of the Java Symphony software infrastructure has been implemented. Preliminary experiments on a heterogeneous cluster of workstations are described that demonstrate reasonable performance values.
We present parallel algorithms for the solution of microfluidic equations. The algorithm consists of the multigrid method in combination with adaptive finite elements. These methods give good algorithmic and parallel ...
详细信息
ISBN:
(纸本)0966613570
We present parallel algorithms for the solution of microfluidic equations. The algorithm consists of the multigrid method in combination with adaptive finite elements. These methods give good algorithmic and parallel scalability and are thus very efficient. Model test problems are taken from Electroosmotic flow simulations.
Recent advances in high-speed networks, rapid improvements in microprocessor design, and availability of highly performing clustering software implementations enables cost-effective high-performance parallel computing...
详细信息
ISBN:
(纸本)076950759X
Recent advances in high-speed networks, rapid improvements in microprocessor design, and availability of highly performing clustering software implementations enables cost-effective high-performance parallel computing on clustered low cost workstations and PCs. Such clusters are very attractive because they rely on available of the shelf hardware and software technologies, To simplify programming in network environments and to realise component-based software architectures, many models have emerged as possible standards, namely, RPC: Remote Procedure Call, DCE: Distributed Computing Environment, DCOM: Distributed Component Object Model, CORBA: Common Object Request Broker Architecture, PVM: parallel Virtual Machine and MPI: Message Passing Interface. The MPI approach is considered one of the most mature methods currently used in parallel programming. MPI implementations on emerging cluster interconnects are an important requirement for useful parallel processing on cost-effective computer clusters. This paper offers an overview of different parallel computing environments and models. Functionality and performance of MPI running on ho,homogeneous and heterogeneous workstation clusters are compared. The project goals, laboratory environment, performance measurements, future work, and a conclusion are presented.
In this work we show the research track and the current results about the application of structured parallel programming tools to develop scalable data-mining applications. We discuss the exploitation of the divide an...
详细信息
ISBN:
(纸本)354067442X
In this work we show the research track and the current results about the application of structured parallel programming tools to develop scalable data-mining applications. We discuss the exploitation of the divide and conquer nature of the well known C4.5 classification algorithm in spite of its in-core memory requirements. The opportunity of applying external memory techniques to manage the data is advocated. Current results of the experiments axe reported.
暂无评论