T.Kalinowski, Scheduling Parallel Program Graphs onto Multi-Processor Systems with Dynamic Interconnection Networks, In: Proc. of the Second Int. Conf. on Massively Parallel Computer Systems, IEEE Computer Society,1996, pp. 287-295.  Home/Search   Document Not in Database   Summary   Related Articles

....the list of available tasks is empty or there is no free processor, the global time is increased until some processors finish execution of their assigned tasks. Then, the list of available tasks and the set of free processors are updated. We propose to apply list scheduling heuristic described in [11] which constructs schedules for the system model presented in section 2. It well reflects features of real parallel systems by accounting for link contention caused by limited number of links connected to each processor and limited link capacity. Please note that the system model applied in [2] ....

....the parallel genetic algorithm with GRADE We propose to implement a PGA solving the task scheduling problem in the following way. The first step is implementation of a serial GA which searches a space of possible task priorities and task assignments for the list scheduling algorithm presented in [11]. Then, we use the GRED editor which supports GRAPNEL language [9] to design a parallel genetic algorithm. In GRADE, parallel programs are constructed in a hierarchical way. At the highest level, called application level, processes, their communication ports and connections between ports must be ....

T.Kalinowski, Scheduling Parallel Program Graphs onto Multi-Processor Systems with Dynamic Interconnection Networks, In: Proc. of the Second Int. Conf. on Massively Parallel Computer Systems, IEEE Computer Society,1996, pp. 287-295.

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