scheduling parallel tasks represented as a Directed Acyclic Graph (DAG), on a multiprocessor system has been an important research area in the past decades. One of the critical aspects of a class of scheduling algorit...
详细信息
scheduling parallel tasks represented as a Directed Acyclic Graph (DAG), on a multiprocessor system has been an important research area in the past decades. One of the critical aspects of a class of scheduling algorithms, called "List scheduling", is how to decide which task is to be scheduled next. This is achieved by assigning priorities to the nodes or the edges of the input DAG, and thus the task with the highest priority will be scheduled next. This paper proposes a low complexity scheduling algorithm to improve the priority node selection criteria in list scheduling algorithms. The worst case performance of the proposed algorithm is analyzed for general input DAGs. Also, the worst case performance and the optimality conditions are obtained for free structured input DAGs. The performance comparison study shows that the proposed algorithm outperforms existing scheduling algorithms especially for input DAGs with high communication overheads. The performance improvement over existing algorithms becomes larger as the input DAG becomes more dense and the level of parallelism in the DAG is increased.
Distributed real-time systems play an increasingly vital role in our daily life. The most important aspect of such systems is the scheduling algorithm, which must guarantee that every job in the system will meet its d...
详细信息
ISBN:
(纸本)9781424434497
Distributed real-time systems play an increasingly vital role in our daily life. The most important aspect of such systems is the scheduling algorithm, which must guarantee that every job in the system will meet its deadline. In this paper we evaluate by simulation the performance of strategies for the scheduling of parallel jobs (gangs) in a homogeneous distributed real-time system with possible software faults. We provide an alternative version for each scheduling policy, which allows imprecise computations, and we propose a performance metric applicable to our problem. Our simulation results show that the alternative versions of the algorithms exhibit promising performance.
The approximation followed here represents a distributed real-time system (DRTS) as a layered model with a real-time scheduler, based on hierarchical scheduling composition, thus providing the DRTS with larger integri...
详细信息
The approximation followed here represents a distributed real-time system (DRTS) as a layered model with a real-time scheduler, based on hierarchical scheduling composition, thus providing the DRTS with larger integrity and flexibility. The utilization of a metric is proposed in order to determine the overall system performance, therefore helping with the design and configuration decisions.
Array redistribution is required often in programs on distributed memory parallel computers. It is essential to use efficient algorithms for redistribution, otherwise the performance of the programs may degrade consid...
详细信息
Array redistribution is required often in programs on distributed memory parallel computers. It is essential to use efficient algorithms for redistribution, otherwise the performance of the programs may degrade considerably. The redistribution overheads consist of two parts: index computation and interprocessor communication. If there is no communication scheduling in a redistribution algorithm, the communication contention may occur, which increases the communication waiting time. In order to solve this problem, we propose a technique to schedule the communication so that it becomes contention-free. Our approach initially generates a communication table to represent the communication relations among sending nodes and receiving nodes. According to the communication table, we then generate another table named communication scheduling table. Each column of the communication scheduling table is a permutation of receiving node numbers in each communication step. Thus the communications in our redistribution algorithm are contention-free. Our approach can deal with multi-dimensional shape changing redistribution.
This paper concerns the problem of finding the minimum-length TDMA frame of a power-controlled wireless network subject to traffic demands and SINR (signal- to-interference-plus-noise ratio) constraints. We formulate ...
详细信息
This paper concerns the problem of finding the minimum-length TDMA frame of a power-controlled wireless network subject to traffic demands and SINR (signal- to-interference-plus-noise ratio) constraints. We formulate the general joint link scheduling and power control problem as an integer linear programming (ILP) problem. The linear relaxation of the ILP problem has been claimed to be NP-hard in the literature. We present a computationally efficient heuristic algorithm, called the increasing demand greedy scheduling (IDGS) algorithm, to solve the general ILP problem. In addition, we propose using a column generation (CG) method as an augmentation to IDGS to further improve its performance. Simulation results show that integration of IDGS and CG can achieve superior performance in terms of both algorithm run time and solution optimality.
Operation scheduling is one of the most important steps in datapath design. The object of this paper is to propose a new idea of scheduling which may potentially lead to an optimal result. Different from the algorithm...
详细信息
Operation scheduling is one of the most important steps in datapath design. The object of this paper is to propose a new idea of scheduling which may potentially lead to an optimal result. Different from the algorithms proposed in the past, which try to find the control step that is suitable for a selected operation, the new idea tries to find the control step that is not suitable for the operation. An operation is scheduled when all the unsuitable steps are found. Some allocation problems such as functional unit sharing and connection minimisation, are combined into the scheduling procedure. Thus, the allocation procedure is simplified and design quality increases. The experiment show that the optimality of scheduling is distinctly improved comparing to those of related works published before.< >
Since multimedia documents may comprise continuous media, such as audio and video, the presentation of those documents may require a significant amount of processing and network resources. Depending on the system conf...
详细信息
Since multimedia documents may comprise continuous media, such as audio and video, the presentation of those documents may require a significant amount of processing and network resources. Depending on the system configuration and the current system load, it can happen that there are not enough resources to render a multimedia document according to the specification, resulting in a reduced presentation quality. To cope with those situations, documents can be specified as flexible so that they can be adapted to different system configurations and load conditions. We present an adaptive scheduling algorithm which allows to adapt documents conforming to our document model Tiempo in environments with best-effort assignment of resources.
If resource constraints are specified, the false loop free circuit must be built during the scheduling phase. Although the previous approach guarantees to have a false loop free circuit mapping, it does not attempt to...
详细信息
If resource constraints are specified, the false loop free circuit must be built during the scheduling phase. Although the previous approach guarantees to have a false loop free circuit mapping, it does not attempt to minimize the number of control steps. In this paper, we present an effective approach to find a scheduled code, which not only guarantees to have a false loop free circuit mapping but also to minimize the number of control steps. The proposed approach can be easily incorporated into any of existing list scheduling like algorithm. Experimental data shows that the proposed approach achieves very good results in terms of the number of control steps
The previous deadline and budget constrained (DBC) algorithms were proposed to optimize the user's cost in market-based job assignment. However those algorithms are not suitable for unreliable grid environment. An...
详细信息
The previous deadline and budget constrained (DBC) algorithms were proposed to optimize the user's cost in market-based job assignment. However those algorithms are not suitable for unreliable grid environment. An opportunity-cost guided algorithm is proposed for job scheduling to bridge this gap. By employing trust mechanism into the grid market, this approach takes both of resource's explicit cost and the successful opportunity into considerations. Von Neumann-Morgenstern utility function (VNM-UF) from the traditional economy theory is applied to characterize the user's risk bias. A quantitative method is given. Simulations show our approach is more suitable for unreliable environment than many other existing approaches. The job failure rate is reduced to 30% and the total cost saves by 8%~10% averagely.
暂无评论