Scheduling of parallel programs on configurable multiprocessors by genetic algorithms (original) (raw)

Scheduling multiprocessor tasks with genetic algorithms

Parallel and Distributed …, 1999

In this paper, an efficient method based on genetic algorithms is developed to solve the multiprocessor scheduling problem. To efficiently execute programs in parallel on multiprocessor scheduling problem must be solved to determine the assignment of tasks to the processors and the execution order of the tasks so that the execution time is minimized. Even when the target processors is fully connected and no communication delay is considered among tasks in the task graph the scheduling problem is NP-complete. Complexity of scheduling problems dependent of number of processors (P), task processing time T i and precedence constraints. This problem has been known as strong NP-hard intractable optimisation problem when it assumes arbitrary number of processors, arbitrary task processing time and arbitrary precedence constraints. We assumed fixed number of processors and tasks are represented by a directed acyclic graph (DAG) called "task graph".

Genetic algorithm for mapping tasks onto a reconfigurable parallel processor

1995

The authors describe a genetic algorithm for a difficult optimisation problem which arises in the context of parallel processing. The problem is to assign each task in the given task graph T to a processor, so as to minimise the total overall execution time of the tasks. Total execution time is computed with the knowledge of individual run times of tasks and the communication requirements among tasks. The intertask communication time is dependent on the interconnection network which connects the processors. No prior knowledge of the interconnection topology is assumed. The algorithm finds the interconnection architecture that is best suited for the task graph T; this makes sense when the target architecture is reconfigurable through programmable switches, e.g. transputer-based parallel processors. The algorithm is also extended to add heterogeneous platforms, where each task t can be executed on a particular class of processors. The optimisation technique is based on the genetic paradigm. The authors describe an efficient chromosome representation, genetic operators and a fitness measure suitable for the application.

Solving the Parallel Task Scheduling Problem by Means of a Genetic Approach

A parallel program, when running, can be conceived as a set of parallel components (tasks) which can be executed according to some precedence relationship. In this case efficient scheduling of tasks permits to take full advantage of the computational power provided by a multiprocessor or a multicomputer system. This involves the assignment of partially ordered tasks onto the system architecture processing components. This paper shows the problem of allocating a number of non-identical tasks in a multiprocessor or multicomputer system. The model assumes that the system consists of a number of identical processors and only one task may execute on a processor at a time. All schedules and tasks are non-preemptive. The well-known Graham's [8] list scheduling algorithm (LSA) is contrasted with an evolutionary approach using a proposed indirect-decode representation.

Genetic scheduling for parallel processor systems: comparative studies and performance issues

1999

Abstract Task scheduling is essential for the proper functioning of parallel processor systems. Scheduling of tasks onto networks of parallel processors is an interesting problem that is well-defined and documented in the literature. However, most of the available techniques are based on heuristics that solve certain instances of the scheduling problem very efficiently and in reasonable amounts of time.

A Genetic Algorithm for Multiprocessor Task Scheduling

The goal of task scheduling in a multiprocessor system is to schedule dependent tasks on processors such that the processing time is minimized. This ensures optimal usage of the processing systems. However this problem is NP-hard in nature and heuristic based techniques are used to obtain a good schedule in polynomial time. Genetic Algorithms (GA) have been proposed over other heuristics because it can use its genetic processes to find multiple solutions faster. The GA proposed is based on a non-preemptive precedence relation between tasks in the task graph. Tasks assignment is prioritized based on the number of tasks dependencies (NTD) and the earliest start time (EST) of each task. For tasks with multiple possible earliest start times, the minimum earliest start time is chosen for such tasks. Java simulations compared the results obtained using the minimum EST and the maximum EST. Our simulation shows that the proposed algorithm with minimum EST achieves faster processing periods compared with the maximum EST.

A genetic approach using direct representation of solution for the parallel task scheduling problem

1999

In scheduling, a set of machines in parallel is a setting that is important, from both the theoretical and practical points of view. From the theoretical viewpoint, it is a generalization of the single machine scheduling problem. From the practical point of view the occurrence of resources in parallel is common in real-world. When machines are computers, a parallel program can be conceived as a set of parallel components (tasks) which can be executed according to some precedence relationship. In this case efficient scheduling of tasks permits to take full advantage of the computational power provided by a multiprocessor or a multicomputer system. This kind of planning involves the assignment of partially ordered tasks onto the system architecture processing components. This paper shows the problem of allocating a number of non-identical tasks in a multiprocessor or multicomputer system. The model assumes that the system consists of a number of identical processors and only one task may execute on a processor at a time. All schedules and tasks are nonpreemptive. The well-known Graham's list scheduling algorithm (LSA) is contrasted with an evolutionary approach using a direct representation of solutions.

Tasks Scheduling on Parallel Heterogeneous Multi-processor Systems using Genetic Algorithm

International Journal of Computer Applications, 2013

With the increasing use of computers in research contributions, added need for faster processing has become an essential necessity. Parallel Processing refers to the concept of running tasks that can be run simultaneously on several processors. There are conditions that tasks have deadlines for scheduling. Therefore, the tasks should be scheduled before deadlines. May number of tasks before scheduling reached their deadline, Therefore, these tasks lost. These conditions are unavoidable. Thus, parallel multi-processor system tasks should be scheduled in a way, minimizing lost tasks. On the other hand, achieving good response times is necessary. this is an NP-Complete problem. In this article, we introduce a method based on genetic algorithms for scheduling tasks on parallel heterogeneous multi-processor systems for tasks with deadlines. The results of the simulations indicate reduced number of lost tasks in comparison with the LPT and SPT algorithms. Moreover, the response time of the proposed method due to its number of processing tasks, is appropriate, in comparison with the algorithm LPT and SPT.

Genetic Approach to Parallel Scheduling

Task Scheduling is Essential part for proper functioning of parallel processing system. Several approaches have been applied to solve this problem. Genetic algorithms have received much awareness as they are robust and guarantee for a good solution. In this paper a genetic algorithm is developed and implemented and performance of algorithm is measured under altering parameters. Adaptive parameter approach has been applied to enhance the performance of the genetic algorithm.

Multiprocessor Scheduling Based on Genetic Algorithms

environments

This paper presents the development of genetic algorithm approach to schedule tasks on a multiprocessor system. The objective is to minimize the make-span i.e. the completion time of all tasks while maintaining the precedence constraints within the task graph. No inter-processor communication overheads are assumed. The array data structure is employed for string representation and a hybrid selection method for reproduction is adopted. The ability of the geneticbased scheduler to deal with resource failures and apperiodic operations is also explored.