J. N. Magee, and S.C. Cheung, "Parallel Algorithm Design for Workstation Clusters", Software--Practice and Experience. Vol. 21(3), (235-250), March, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....with an indexing mechanism implemented in C which allows to efficiently embed data driven computations into Prolog. Moreover, the scheduler for rules is written in C, allowing an easier definition of fair scheduling policies [Gas93b] The runtime support is based on the masterworker model [Mag91]. The paper is organised as follows: in Section 2 we introduce Shared Prolog and the programming notation used throughout the paper, in Section 3 we discuss the problems of the current definition of the language, we study the semantics of guard evaluation, and we propose an extension to the ....

....on the first approach. On one side we exploit the same target architecture of the previous implementations of SP based on the same approach, a workstation cluster on an ethernet LAN. On the other side we design a new runtime support. First, we adopt the master worker model with replicated workers [Mag91]: a worker is not associated to a specific agent program, but it is able to execute the goals of all the agents; this allows a better balancing of the machines workload in the network expecially when the number of agents is greather than the number of available processors. Second, we exploit ....

J. N. Magee, and S.C. Cheung, "Parallel Algorithm Design for Workstation Clusters", Software--Practice and Experience. Vol. 21(3), (235-250), March, 1991.

....The first component is an underlying runtime system which supports execution of the parallel programs. The second and third components are MPL and MDF libraries which offer primitives for developing parallel programs in POENDOM environment. 2. 1 Architecture of POENDOM Supervisor Worker model [8] was followed with relevant modifications in the design of the underlying system. Emulation of a parallel machine mainly consists of two components: a Host node and other Remote nodes. The architecture of the POENDOM is shown in Figure 1. The runtime system must be established before submitting a ....

J.N. Magee and S.C.Cheng, Parallel Algorithm Design for Workstation Clusters, Software Practices and Experiences, Vol. 21(3), Mar'90,pp: 235 - 250.

.... similar in form to the CDR strings of Conic[MAG89] similar in form to the CDR strings of Conic[MAG89] The Darwin program for The Darwin program for sieve sieve also contains information on how it should be executed on a also contains information on how it should be executed on a multicomputer [MAG91a] The annotation provided by multicomputer [MAG91a] The annotation provided by pe pe specifies a mapping to an specifies a mapping to an abstract machine consisting of abstract machine consisting of maxprocessor maxprocessor identical processor memory (PE) pairs. PEs are identical ....

.... similar in form to the CDR strings of Conic[MAG89] The Darwin program for The Darwin program for sieve sieve also contains information on how it should be executed on a also contains information on how it should be executed on a multicomputer [MAG91a] The annotation provided by multicomputer [MAG91a] The annotation provided by pe pe specifies a mapping to an specifies a mapping to an abstract machine consisting of abstract machine consisting of maxprocessor maxprocessor identical processor memory (PE) pairs. PEs are identical processor memory (PE) pairs. PEs are numbered from 1 to ....

[Article contains additional citation context not shown here]

J.N. Magee and S.C. Cheung, " Parallel Algorithm Design for Workstation Clusters", Software-Practice and Experience, Vol. 21. March 1991, pp 235-250.

....this approach leads to poor load balancing and thus poor performance. A more sophisticated approach employs a work allocator or supervisor thus poor performance. A more sophisticated approach employs a work allocator or supervisor process to hand out smaller chunks to worker or slave processes [MAG91] A slave process process to hand out smaller chunks to worker or slave processes [MAG91] A slave process computes a chunk and hands it back to the supervisor for display and then gets another chunk computes a chunk and hands it back to the supervisor for display and then gets another chunk to ....

....approach employs a work allocator or supervisor thus poor performance. A more sophisticated approach employs a work allocator or supervisor process to hand out smaller chunks to worker or slave processes [MAG91] A slave process process to hand out smaller chunks to worker or slave processes [MAG91] A slave process computes a chunk and hands it back to the supervisor for display and then gets another chunk computes a chunk and hands it back to the supervisor for display and then gets another chunk to compute until none are left. In the following, chunks are the size of one horizontal line ....

J.N. Magee and S.C. Cheung, " Parallel Algorithm Design for Workstation Clusters", Software-Practice and Experience, Vol. 21. March 1991, pp 235250.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC