Scheduling Of Parallel Tasks Research Articles

Energy-aware parallel task scheduling in a cluster

Reducing energy consumption for high end computing can bring various benefits such as reducing operating costs, increasing system reliability, and environmental respect. This paper aims to develop scheduling heuristics and to present application experience for reducing power consumption of parallel tasks in a cluster with the Dynamic Voltage Frequency Scaling (DVFS) technique. In this paper, formal models are presented for precedence-constrained parallel tasks, DVFS-enabled clusters, and energy consumption. This paper studies the slack time for non-critical jobs, extends their execution time and reduces the energy consumption without increasing the task’s execution time as a whole. Additionally, Green Service Level Agreement is also considered in this paper. By increasing task execution time within an affordable limit, this paper develops scheduling heuristics to reduce energy consumption of a tasks execution and discusses the relationship between energy consumption and task execution time. Models and scheduling heuristics are examined with a simulation study. Test results justify the design and implementation of proposed energy aware scheduling heuristics in the paper.

Adaptive scheduling for parallel tasks with QoS satisfaction for hybrid cloud environments

A hybrid cloud integrates private clouds and public clouds into one unified environment. For the economy and the efficiency reasons, the hybrid cloud environment should be able to automatically max...

Scheduling of hard real-time multi-phase multi-thread (MPMT) periodic tasks

In this paper we study the scheduling of parallel and real-time recurrent tasks on multiprocessor platforms. Firstly, we propose a new parallel task model which allows recurrent tasks to be composed of several phases, each one composed of several threads. Each thread requires a single processor for execution and can be scheduled simultaneously. We then propose an algorithm to transpose popular Fork-Join task model to our MPMT task model. Secondly, we define several kinds of real-time schedulers that can be applied to our parallel task model. We distinguish between two scheduling classes: Hierarchical schedulers and Global Thread schedulers. We present and prove correct an exact schedulability test for each class. Lastly, we also evaluate the performance of our scheduling paradigm in comparison with Gang scheduling by means of simulations. In this work we extend the work of Lupu and Goossens in Scheduling of hard real-time multi-thread periodic tasks (Real-Time and Network Systems, 2011) which considers mono-phase multi-thread task model. We extend their previous results to a Multi-Phase Multi-Thread task model.

A Multi‐granularity Parallel Model for Unified Remote Sensing Image Processing WebServices

AbstractThe growth of the Web has resulted in the Web‐based sharing of distributed geospatial data and computational resources. The Geospatial Processing Web (GeoPW) described here is a set of services that provide a wide array of geo‐processing utilities over the Web and make geo‐processing functionalities easily accessible to users. High‐performance remote sensing image processing is an important component of the GeoPW. The design and implementation of high‐performance image processing are, at present, an actively pursued research topic. Researchers have proposed various parallel strategies for single image processing algorithm, based on a computer science approach to parallel processing. This article proposes a multi‐granularity parallel model for various remote sensing image processing algorithms. This model has four hierarchical interfaces that are labeled the Region of Interest oriented (ROI‐oriented), Decompose/Merge, Hierarchical Task Chain and Dynamic Task interfaces or sub‐models. In addition, interfaces, definitions, parallel task scheduling and fault‐tolerance mechanisms are described in detail. Based on the model and methods, we propose an open‐source online platform named OpenRS‐Cloud. A number of parallel algorithms were uniformly and efficiently developed, thus certifying the validity of the multi‐granularity parallel model for unified remote sensing image processing web services.

가단성 태스크들의 마감시간 스케줄링의 자원추가 분석

In this paper, we deal with the problem of scheduling parallel tasks with deadlines. Parallel tasks can be simultaneously executed on various machines and specially, we consider the malleable tasks, that is, the tasks whose execution time is given by a function of the number of machines on which they are executed. The goal of the problem is to maximize the throughput of tasks completed within their deadlines. This problem is well-known as NP-hard problem. Thus we will find an approximation algorithm, and its performance is compared with that of the optimal algorithm and analyzed by finding the approximation ratio. In particular, the algorithm has more resources, that is, more machines, than the optimal algorithm. This is called the resource augmentation analysis. We propose an algorithm to guarantee the approximation ratio of 3.67 using 1.5 times machines.

Flexible Scheduling and Thread Allocation for Synchronous Parallel Tasks

We describe a task model and dynamic scheduling and resource allocation mechanism for synchronous parallel tasks to be executed on SPMD-programmed synchronous shared-memory MIMD parallel architectures with uniform, unit-time memory access and strict memory consistency, also known in the literature as PRAMs (Parallel Random Access Machines). Our task model provides a two-tier programming model for PRAMs that flexibly combines SPMD and fork-join parallelism within the same application. It offers flexibility by dynamic scheduling and late resource binding while preserving the PRAM execution properties within each task, the only limitation being that the maximum number of threads that can be assigned to a task is limited to what the underlying architecture provides. In particular, our approach opens for automatic performance tuning at run-time by controlling the thread allocation for tasks based on run-time predictions. By a prototype implementation of a synchronous parallel task API in the SPMD-based PRAM language Fork and experimental evaluation with example programs on the SBPRAM simulator, we show that a realization of the task model on a SPMD-programmable PRAM machine is feasible with moderate runtime overhead per task.

Energy Efficient Scheduling and Optimization for Parallel Tasks on Homogeneous Clusters

The design of energy-efficient scheduling algorithms has become a hot research topic in high performance computing.To shorten schedule length of parallel tasks with precedence constraints,scheduling algorithms could duplicate tasks on critical paths to avoid communication delay caused by inter-task dependence.However,task duplications incur more energy consumption.In this paper,we propose a heuristic Processor Reduction Optimizing(PRO) approach to reduce the number of processors used to run parallel tasks,thereby decreasing system energy consumption.The PRO approach can find appropriate time slots to accommodate tasks from low-utilized processors according to their earliest start time and earliest complete time.Extensive experimental results show that the proposed PRO approach,compared to existing duplication-based scheduling algorithms,such as Task Duplication Scheduling(TDS),Energy-Aware Duplication(EAD) and Performance-Energy Balanced Duplication(PEBD) algorithms,can effectively decrease the number of used processors and save energy without performance degradation.

A Novel Approach for Task Scheduling in Multiprocessor System

In multiprocessor system, scheduling of tasks to assigned on the number of processors. The major objective of task scheduling is to find minimum execution time of a program. It is well known that the complexity of a general scheduling problem is NP-Complete [9], there are number of heuristic have been developed. Each of which may either find optimal or near optimal scheduling under the different conditions. The task scheduling is represented by a directed acyclic graph (DAG). In this paper, we present a new scheduling algorithm which is called Task Scheduling based on Breath First Search(TSB). The TSB is queue based approach to schedule parallel tasks on the homogenous parallel multiprocessor system. Its performance is evaluated in comparison with Highest Level First with Estimate Time (HLFET) algorithm, Modified Critical Path (MCP) algorithm, Earliest Time First (ETF) algorithm and Dynamic Level Scheduling (DLS) algorithm in terms of Speedup, Efficiency, Load Balance and Normalized Scheduling Length (NSL).

Improved Task Scheduling on Parallel System using Genetic Algorithm

Parallel Processing refers to the concept of speeding-up the execution of a task by dividing the task into multiple fragments that can execute simultaneously, each on its own processor i.e. it is the simultaneous processing of the task on two or more processors in order to obtain faster results. It can be effectively used for tasks that involve a large number of calculations, have time constraints and can be divided into a number of smaller tasks. The scheduling problem deals with the optimal assignment of a set of tasks onto parallel multiprocessor system and orders their execution so that the total completion time is minimized. The efficient execution of the schedule on parallel multiprocessor system takes the structure of the application and the performance characteristics of the proposed algorithm. Many heuristics and approximation algorithms have been proposed to fulfill the scheduling task. It is well known NP-complete problem. This study proposes a genetic based approach to schedule parallel tasks on heterogeneous parallel multiprocessor system. The scheduling problem considered in this study includes next to search for an optimal mapping of the task and their sequence of execution and also search for an optimal configuration of the parallel system. An approach for the simultaneous optimization of all these three components of scheduling method using genetic algorithm is presented and its performance is evaluated in comparison with the First Come First Serve (FCFS), Shortest Job First (SJF), Round Robin (RR), Priority and Largest Job First (LJF) scheduling methods.

О построении расписаний выполнения параллельных задач на группах кластеров с различной производительностью

A classical problem of scheduling the set of tasks optimizing load balancing for a set of given processors was considered in theory in 1966. Graham's algorithm sending each task to a least loaded machine (processor) was the first example of approximate algorithm with guaranteed constant approximation quality. But all tasks and processors (machines) in Graham's example were simple (only one processor was sufficient for each task). On the other hand Graham's algorithm was an on-line one that is each task was sended to processor without knowing future tasks. With wide use of computational clusters very actual became the problem of scheduling parallel tasks on the set of parallel clusters in the class of on-line algorithms.. This problem has very close relation to a two dimensional packing problem, socalled multiple strip packing problem. Few years ago the author proposed an on-line algorithm for this problem and proved that for arbitrary set of parallel tasks (rectangles) the obtained schedule (packing) is within 2e of the optimal. In this paper the author generalized his on-line algorithm for the case of related parallel machines when all clusters have different speeds of processors in different clusters. An on-line algorithm for scheduling parallel tasks on a group of clusters with different speeds of processors is presented and it was proved that its approximation ratio is 2e for any set of parallel tasks. This result is non-trivial because even the case of one-processor related machines was intensively investigated and some algorithms with constant approximation ratio were presented.

Combined Scheduling and Mapping for Scalable Computing with Parallel Tasks

Recent and future parallel clusters and supercomputers use symmetric multiprocessors (SMPs) and multi-core processors as basic nodes, providing a huge amount of parallel resources. These systems often have hierarchically structured interconnection networks combining computing resources at different levels, starting with the interconnect within multi-core processors up to the interconnection network combining nodes of the cluster or supercomputer. The challenge for the programmer is that these computing resources should be utilized efficiently by exploiting the available degree of parallelism of the application program and by structuring the application in a way which is sensitive to the heterogeneous interconnect. In this article, we pursue a parallel programming method using parallel tasks to structure parallel implementations. A parallel task can be executed by multiple processors or cores and, for each activation of a parallel task, the actual number of executing cores can be adapted to the specific execution situation. In particular, we propose a new combined scheduling and mapping technique for parallel tasks with dependencies that takes the hierarchical structure of modern multi-core clusters into account. An experimental evaluation shows that the presented programming approach can lead to a significantly higher performance compared to standard data parallel implementations.

An efficient energy and schedule length model for multiprocessor computers

Multiprocessors have emerged as a powerful computing means for running real-time applications, especially where a uniprocessor system would not be sufficient to execute all the tasks. Such a computing environment requires an efficient algorithm to determine when and on which processor a given task should be executed. An efficient scheduling of parallel tasks onto the multiprocessors is known to be an NP-hard problem. This paper focuses on the combinational optimisation problem, namely the problem of minimising schedule length with energy consumption constraint and the problem of minimising energy consumption with schedule length constraint for independent parallel tasks. These problems emphasise the trade-off between power and performance and are defined such that the power-performance product is optimised by fixing one factor and minimising other and vice versa. The performance of the proposed algorithm with optimal solution is validated analytically and compared with Particle Swarm Optimisation (PSO) and Genetic Algorithm (GA).

Энергоэффективные вычисления для группы кластеров

The problem of scheduling parallel tasks on a group of geographically distributed clusters optimizing energy-efficiency of computations is considered. Some scheduling algorithms are proposed and experimentally investigated. Methods for reducing the energy consumption of a uniform computer cluster due to flexible control strategies of the node states (waking them up or shutting down) and of the execution order of the awaiting tasks are considered. A software system developed in the Institute for System Programming of the Russian Academy of Sciences (ISP RAS) for the dynamic control of nodes in order to reduce the energy consumption is described. Several strategies for controlling the states of the nodes are proposed and investigated. Our simulation showed that when the average density of tasks is 0.5, the energy saving is about 10%. When the density of the flow of tasks decreases, the effect of using the proposed system drastically increases: when the average density is 0.3, the saving is 30%; when the average density is 0.2, the saving is 50%; and when the average density is 0.1, the saving is 70%.

Resource management of distributed virtual machines

Using Virtual Machines (VMs) as a computing resource within a Service Oriented Architecture (SOA) creates a variety of new issues and challenges. Traditionally parallel task scheduling algorithms only focus on handling CPU resources but with use of a VM there are many more resources properties to monitor and manage. The objective of this paper is to address these challenges of a multi–dimensional scheduling algorithm for VMs within a SOA. To do this we deploy a testbed of SOA environment with VMs that are capable of being registered, indexed, allocated, accessed and controlled by our new parallel task scheduling algorithms.

Partitioned scheduling of parallel real-time tasks on multiprocessor systems

In this paper, we focus on the scheduling of periodic fork-join real-time tasks on multiprocessor systems. Parallel real-time tasks in the fork-join model have strict parallel segments without laxity. We propose a partitioned scheduling algorithm which increases the laxity of the parallel segments and therefore the schedulability of tasksets of this model. A similar algorithm has been proposed in the literature but it produces job migrations. Our algorithm eliminates the use of job migrations in order to create a portable algorithm that can be implemented on a standard Linux kernel. Results of extensive simulations are provided in order to analyze the schedulability of the proposed algorithm, and to provide comparisons with the other algorithm proposed in the literature.

Energy efficient scheduling of parallel tasks on multiprocessor computers

In this paper, scheduling parallel tasks on multiprocessor computers with dynamically variable voltage and speed are addressed as combinatorial optimization problems. Two problems are defined, namely, minimizing schedule length with energy consumption constraint and minimizing energy consumption with schedule length constraint. The first problem has applications in general multiprocessor and multicore processor computing systems where energy consumption is an important concern and in mobile computers where energy conservation is a main concern. The second problem has applications in real-time multiprocessing systems and environments where timing constraint is a major requirement. Our scheduling problems are defined such that the energy-delay product is optimized by fixing one factor and minimizing the other. It is noticed that power-aware scheduling of parallel tasks has rarely been discussed before. Our investigation in this paper makes some initial attempt to energy-efficient scheduling of parallel tasks on multiprocessor computers with dynamic voltage and speed. Our scheduling problems contain three nontrivial subproblems, namely, system partitioning, task scheduling, and power supplying. Each subproblem should be solved efficiently, so that heuristic algorithms with overall good performance can be developed. The above decomposition of our optimization problems into three subproblems makes design and analysis of heuristic algorithms tractable. A unique feature of our work is to compare the performance of our algorithms with optimal solutions analytically and validate our results experimentally, not to compare the performance of heuristic algorithms among themselves only experimentally. The harmonic system partitioning and processor allocation scheme is used, which divides a multiprocessor computer into clusters of equal sizes and schedules tasks of similar sizes together to increase processor utilization. A three-level energy/time/power allocation scheme is adopted for a given schedule, such that the schedule length is minimized by consuming given amount of energy or the energy consumed is minimized without missing a given deadline. The performance of our heuristic algorithms is analyzed, and accurate performance bounds are derived. Simulation data which validate our analytical results are also presented. It is found that our analytical results provide very accurate estimation of the expected normalized schedule length and the expected normalized energy consumption and that our heuristic algorithms are able to produce solutions very close to optimum.

Region-based parallelization of irregular reductions on explicitly managed memory hierarchies

Multicore architectures are evolving with the promise of extreme performance for the classes of applications that require high performance and large bandwidth of memory. Irregular reduction is one of important computation patterns for many complex scientific applications, and it typically requires high performance and large bandwidth of memory. In this article, we propose region-based parallelization techniques for irregular reductions on multicore architectures with explicitly managed memory hierarchies. Managing memory hierarchy in software requires a lot of programming efforts and tends to be error-prone. The difficulties are even worse for applications with irregular data access patterns. To relieve the burden of memory management from programmers, we develop abstractions, particularly targeted to irregular reduction, for structuring parallel tasks, mapping the parallel tasks to processing units and scheduling data transfers between the memory hierarchies. Our framework employs iteration reordering based on regions of data along with dynamic scheduling of parallel tasks. We experimentally evaluate the effectiveness of our techniques for irregular reduction kernels on the Cell processor embedded in a Sony PlayStation3. Experimental results show the speedups of 8 to 14 on the six available SPEs.

A comparison of utility-oriented algorithms for scheduling parallel tasks in multi-cluster grid

Scheduling parallel tasks in multi-cluster grid can be seen as two interdependent problems: cluster allocation and scheduling parallel task on the allocated cluster. In this paper both rigid and moldable parallel tasks are considered. We propose a theoretical model of utility-oriented parallel task scheduling in multi-cluster grid with advance reservations. On the basis of the model we present an approximation algorithm, a repair strategy based genetic algorithm and greedy heuristics MaxMax, T-Sufferage and R-Sufferage to solve the two interdependent problems. We compare the performance of these algorithms in aspect of utility optimality and timing results. Simulation results show on average the (1+?)-approximation algorithm achieves the best trade-off between utility optimality and timing. Genetic algorithm could achieve better utility than greedy heuristics and approximate algorithm at expensive time cost. Greedy heuristics do not perform equally well when adapted to different utility functions while the approximation algorithm shows its intrinsic stable performance.

Time and Cost-Driven Scheduling of Data Parallel Tasks in Grid Workflows

The necessity of identifying suitable computing resources to solve a scientific or engineering problem in a Grid environment requires more and more sophisticated resource management systems: 1) strategies and technologies should be able to master the complexity of modern large-scale networks and computing facilities and 2) the convergence of Grid computing toward the service-oriented approach is fostering a new vision where economic aspects represent central issues to burst the adoption of computing as a utility. In this context, the design and execution of data and compute-intensive applications are often simplified by the adoption of model-driven approaches based on workflows. The execution of Grid workflows can leverage on meta-scheduling systems to automatically and transparently allocate tasks to resources that ensure the fulfillment of functional requirements and quality-of-service (QoS) constraints, specified by the user. This paper presents a time and cost-constrained scheduling strategy that, according to the data parallelism pattern, is able to deploy scientific and business workflow tasks (or other kinds of application tasks) on pools of resources selected with the aim of minimizing the overall execution time. The strategy was implemented as a plug-in in a matchmaker for Grid services and its validity and accuracy were experimentally proved on a real testbed leveraging a framework for the deployment of data parallel tasks. The results show that the tasks deployment is effective and accurate and pave the way for using the Internet as a utility computing facility.

Processor-Minimum Scheduling of Real-Time Parallel Tasks

We propose a polynomial-time algorithm for the scheduling of real-time parallel tasks on multicore processors. The proposed algorithm always finds a feasible schedule using the minimum number of processing cores, where tasks have properties of linear speedup, flexible preemption, arbitrary deadlines and arrivals, and parallelism bound. The time complexity of the proposed algorithm is O(M3· log N) for M tasks and N processors in the worst case.