M. you Wu and D. D. Gajski, "Hypertool: A programming aid for message-passing systems," IEEE Trans. on Parallel and Distributed Systems, vol. 1, pp. 330--343, Jul. 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....a lower priority. Similarly, the larger #### time means that the task can be executed later, also should be assigned lower priority. Unlike #### value, the #### time doesn t change during the scheduling, therefore, it is still computed statically. The algorithm to compute #### can be found in [22]. for(each unscheduled task) check if ready( if(task is ready( finish time = check predecessor( dyna ASAP = search FPGA for space(finish time) dyna priority = dyna ASAP ALAP) Fig. 4. Compute dynamic priority Fig. 4 describes the algorithm of computing dynamic ....

M. you Wu and D. D. Gajski, "Hypertool: A programming aid for message-passing systems," IEEE Trans. on Parallel and Distributed Systems, vol. 1, pp. 330--343, Jul. 1990.

....V and a communication cost c(e) for each node e # E. 3. Related Work A large number of static and dynamic scheduling and partitioning algorithms for scheduling of task graphs to a multiprocessor system are presented in the literature. Some of them use a static technique called list scheduling [4, 7, 10, 6, 5, 9]. A list scheduler keeps a list of tasks that are ready to be scheduled, i.e. all its predecessors have already been scheduled. It selects one of the tasks in the list, by some heuristic, and assigns it to a suitable processor. Another technique is called critical path scheduling [11] The ....

Min-You Wu, Daniel D. Gajski. Hypertool: A Programming Aid for Message-Passing Systems. Transactions on Parallel and Distributed Systems, 1, 1990.

....and fiP strategies are elaborated in Sections 4.2 through 4.5. The efficiency discrepany between PTE and non preemptive task execution (NPTE) is studied in Section 4.6. 4. 1 Two Strategies for Preemptive Execution Recall the normal definition for the non preemptive task execution , as adopted in [3, 13, 16, 17]: a task does not commence execution until it receives all required data; the task then executes without interruption until its completion; the task finally transmits all necessary data to its child tasks. That is to say, all data communication occurs either in the beginning or at the end of each ....

Min You Wu and Daniel Gajski. Hypertool: a programming aid for message-passing systems. IEEE Transactions on Parallel and Distributed Systems, Volume 1, Number 3, pages 330--343, July 1990.

....of a parallel program onto the underlying available physical processors. The scheduling problem has been studied for some time, and has been proved to be NP complete [12] i.e. there is no optimal solution to this problem within polynominal time) Various heuristics [5, 8] and software tools [13, 14] have been proposed to pursue a suboptimal solution within acceptable computation complexity bounds. A detailed survey is found in [1, 4] Most heuristics to date have assumed that the task model remains constant across program executions. However, when tasks (especially task runtime operations ....

Min You Wu and Daniel Gajski. Hypertool: a programming aid for messagepassing systems. IEEE Transactions on Parallel and Distributed Systems, Volume 1, Number 3, pages 330--343, July 1990.

....Australia, February 5 7 1997. mance by decreasing execution time, and increasing system throughput and reliability. Most current research into task mapping environments concentrates on efficient scheduling algorithms to achieve good application and system performance; examples include Hypertool [6] and Pyrros [7] Some other research groups focus on parallel programming support by just incorporating a very simple scheduling algorithm, as done in HeNCE [1] All these environments usually assume that task information, such as task computation time and communication data size, can be precisely ....

Min You Wu and Daniel Gajski. Hypertool: a programming aid for message-passing systems. IEEE Transactions on Parallel and Distributed Systems, Volume 1, Number 3, pages 330--343, July 1990.

....of a parallel program onto the underlying available physical processors. The scheduling problem has been studied for some time, and has been proved to be NP complete [15] i.e. there is no optimal solution to this problem within polynominal time) Various heuristics [5, 10] and software tools [16, 17] have been proposed to pursue a suboptimal solution within acceptable computation complexity bounds. A detailed survey is found in [1, 4] Most heuristics to date have assumed that the task model remains constant across program executions. However, when tasks (especially task runtime operations ....

....actual task models in specific executions. 2.2 Conditional Task Model A software application features a number of interrelated tasks owing to data or control dependencies between tasks which can be represented as a weighted acyclic directed graph, DAG. The DAG representation is widely used [5, 13, 16, 17]. Such a graph can be formally defined as G = T ; E; C u ; Cm ; T s ; T e ) where: ffl T : the set of tasks in the application program. ffl E: the set of task interconnections in the application program. ffl C u : the set of computation times of each task in T , i.e. the time required by ....

Min You Wu and Daniel Gajski. Hypertool: a programming aid for message-passing systems. IEEE Transactions on Parallel and Distributed Systems, Volume 1, Number 3, pages 330--343, July 1990.

....difficult owing to the complexity of partitioning of a program into tasks, task scheduling, task communication and synchronization. Parallel programming environments can significantly reduce a programmer s workload, and increase system and application performance by automating tedious chores [1]. The paper focuses on parallel applications in which runtime task spawn and communication may be conditional; that is, a task may, or may not, be spawned in any one execution as a result of input parameters, user interaction, etc. From the user s perspective, the application software is like a ....

....(termed conditional task scheduling ) Most current research into task scheduling assumes that task information can be precisely obtained prior to execution, and resultant scheduling algorithms do not consider variation of the task model between executions. Such environments include Hypertool [1] and HeNCE [2] However, for some parallel applications, this assumption is impractical, if not impossible, to satisfy especially when conditional branches and loops are included in the user s source code. So far little research into conditional task scheduling has been performed. El Rewini et.al. ....

Min You Wu and Daniel Gajski. Hypertool: a programming aid for message-passing systems. IEEE Transactions on Parallel and Distributed Systems, 1(3):330--343, July 1990.

....local operations. Usually, the prototyping processes have the same behavior and each process runs a private copy of an identical code. Different from general purpose distributed operating systems programming environments where optimal process partitioning and mapping problems must be solved [13] [17], in our rapid prototyping of distributed algorithms, the decomposition and mapping of the prototyping processes are determined by the programmer. The problem of process placement becomes straightforward because the underlying environment is adapted by the programmer to the application so as not ....

Min-You Wu, "Hypertool: A Programming Aid for Message-Passing Systems", IEEE Transactions on Parallel and Distributed Systems, Vol. 1, No.3, July 1990, pp330-343

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