• Journal of Korea Multimedia Society
• /
• v.3 no.3
• /
• pp.309-319
• /
• 2000

The purpose of a parallel scheduling under a multiprocessor environment is to carry out the scheduling with the minimum synchronization overhead, and to perform load balance for a parallel application program. The processors calculate the chunk of iteration and are allocated to carry out the parallel iteration. At this time, it frequently accesses mutually exclusive global memory so that there are a lot of scheduling overhead and bottleneck imposed. And also, when the distribution of the parallel iteration in the allocated chunk to the processor is different, the different execution time of each chunk causes the load imbalance and badly affects the capability of the all scheduling. In the paper. we investigate the problems on the conventional algorithms in order to achieve the minimum scheduling overhead and load balance. we then present a new parallel loop scheduling algorithm, considering the locality of the data and processor affinity.

• Journal of Digital Contents Society
• /
• v.8 no.3
• /
• pp.279-288
• /
• 2007

Recently, Baruah et. al. proposed an optimal Pfair scheduling algorithm in the real-time multiprocessor system environments, and several variants of it were presented. All these algorithms assume the fixed unit quantum size. However, under Pfair based scheduling algorithms that are global scheduling technique, quantum size has direct influence on the scheduling overheads such as task switching and cache reload. We proposed a method for deciding the optimal quantum size[2] and an improved method for the task set whose utilization e is less than or equal to $e\;{\leq}\;p/3+1$[3]. However, these methods use repetitive computation of the task's utilization to determine the optimal quantum size. In this paper, we propose a more efficient method that can determine the optimal quantum size in constant time.

• The KIPS Transactions:PartA
• /
• v.17A no.6
• /
• pp.259-264
• /
• 2010

In case of scheduling tasks with shared resources in multiprocessor systems, Global Earliest Deadline First (GEDF) algorithm, equally applied Earliest Deadline First (EDF) which runs scheduling with deadline criterion, makes schedulability decline because GEDF typically does not have a specific process in order to handle tasks with shared resources. In this paper, we propose Earliest Deadline First with Partitioning (EDFP) for tasks with shared resources which partitions a task into two kinds of subtasks that include critical sections to access to shared resources, gives their own deadline respectively and manages them. As a result of simulations, EDFP shows better performance than GEDF for tasks with shared resources since system load goes up and the number of processor increases.

• Journal of Korea Society of Industrial Information Systems
• /
• v.6 no.2
• /
• pp.38-46
• /
• 2001

This paper presents a new multiprocessor system and corresponding scheduling algorithm that can be applied for implementation of fine grain DSP algorithms such as digital filters. The newly proposed system uses one or more shared buses as the basic interconnection network between processors, and fixed amount of clock-skew is maintained between instruction execution of processors. This system not only can handle the interprocessor communications very efficiently but also can explicitly incorporate the interprocessor communication delay time into the multiprocessor scheduling model. This paper also presents a new scheduling strategy for implementing digital filters expressed in fully-specified flow graphs on the proposed system. The simulation result shows that well-known digital filters can be implemented on proposed multiprocessor in which the implementation satisfies the iteration period bound.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.06a
• /
• pp.213-214
• /
• 2017

홀로그램을 생성하기 위해서 많은 양의 계산을 필요하기 때문에 고속 홀로그램 생성 방법이 필요하다. 본 논문에서는 다중 프로세서와 다중 GPGPU의 스케줄링을 이용하여 고속화 하는 방법을 제안하고 구현하였다. 다중 프로세서를 이용하여 입력과 출력부분을 나누어 동기화 동작을 줄이고, 버퍼를 이용하여 커널과 커널 사이의 대기 시간을 줄일 수 있도록 스케줄링 하였다. nVidia사의 GTX680(Kepler구조) 2개를 이용하여 구현하였을 때, 이전 연구에서 제안한 방법에 비하여 약 70% 정도 계산시간을 줄일 수 있다.

• The KIPS Transactions:PartA
• /
• v.16A no.6
• /
• pp.437-442
• /
• 2009

For multiprocessor systems, Earliest Deadline First (EDF) based on deadline and Least Laxity First (LLF) based on laxity are not suitable for practical environment since EDF has low schedulability and LLF has high context switching overhead. As a combining of EDF and LLF to improve the performance, Earliest Deadline Zero Laxity (EDZL) was proposed. EDZL is basically the same as EDF. But if the laxity of a task becomes zero, its priority is promoted to the highest level. In this paper, we propose Least Laxity Zero Laxity (LLZL) which is based on LLF. But context switching is allowed only if the laxity of a task on rady queue becomes zero. Simulation results show that LLZL has high schedulability approaching to LLF and low context switching overhead similar to EDF. In comparison with EDZL, LLZL has better performance in both of schedulability and context switching overhead.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2012.10a
• /
• pp.733-735
• /
• 2012

Real-Time Aperiodic Tasks Scheduling Using Synthetic Utilization on Multiprocessor Systems has a drawback in that if some tasks, even though they are completed and have no more execution times, are included in the current invocation set, their execution times and deadlines are added to the synthetic utilization. This may lead to a problem in which actually schedulable tasks are decided not to be schedulable. In this paper, we recognize the above mentioned problem and propose an improved synthetic utilization method that can be used to schedule aperiodic tasks more efficiently on multiprocessor systems.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.79-81
• /
• 2001

논리적 정확성과 시간적 제약의 만족을 중요시하는 실시간 시스템은 시간이 경과함에 따라 그 기능이 다른 것들로 변화되는 mode change를 요구할 수도 있으며, 그 시간적 제약의 엄격함에 따라 경성 실시간 시스템과 연성 실시간 시스템으로 나눌 수 있다. 유닉스 계열의 운영체제를 확장한 운영체제인 Linux는 연성 실시간을 지원하도록 개발되었으며 이 Linux에 최소의 변경을 가해 경성 실시간을 지원하기 위한 RTLinux가 개발되었다. RTLinux 버전 2.0은 다중 프로세서 시스템을 지원할 수 있도록 재 설계되었지만, 다중 프로세서 환경에서 주어진 태스크들을 각 프로세서에서 정확하고 효율적으로 실행시키기 위한 구체적인 프로세서할당 및 스케줄링 기능은 제공되지 않고 있다. 이러한 사항들을 만족시키기 위해 본 논문에서는 다중 프로세서 환경에서 mode change되는 태스크들이 각각의 mode 범위 내에서 정확하고 효율적으로 실행될 수 있도록 프로세서를 할당하고 스케줄링하는 RTLinux스케줄러를 제시하고 구현하였다.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.3
• /
• pp.147-153
• /
• 2007

Task scheduling is an essential part of any computer system for allocating tasks to a processor of the system among various competitors. As we know, in real-time system, the failure of scheduling a hard real-time task my lead to disastrous consequence. Besides efficiency, resource and speed, real-time system has to take time constraint in serious consideration. This paper proposes a priority-driven scheduling algorithm for real-time multiprocessor system. which is called L-RE coordinates algorithm. L-RE coordinates is a new way of describing the task scheduling problem. In the algorithm, we take both deadline and laxity into consideration for allocating the priority. The simulation result shows that the new algorithm is viable and performance better than EDF and LLF algorithm on schedulability and context switch respectively.

• The Journal of the Korea Contents Association
• /
• v.7 no.7
• /
• pp.39-49
• /
• 2007

Since the PF scheduling algorithm[13], which is optimal in the hard real-time multiprocessor environments, several scheduling algorithms have been proposed. All these algorithms assume the fixed unit quantum size, and this assumption has problems in the mode change environments. To settle the problem, we already proposed a method for deciding the optimal quantum size[2]. In this paper, we propose improved methods considering the task set whose utilization e is less than or equal to p/3+1. As far as the numbers of computations used to determine the optimal quantum size are concerned, newly proposed methods are proved to be more efficient than our previous ones.