• Proceedings of the Korean Information Science Society Conference
• /
• 한국정보과학회 2000년도 봄 학술발표논문집 Vol.27 No.1 (A)
• /
• pp.104-106
• /
• 2000

단일프로세서 환경에서 주기 태스크의 시간 제약 조건을 만족시키면서 비주기 태스크의 평균 서비스 시간을 최소화하는 연구가 계속 진행되어 왔으나, 최근의 시스템은 여러 개의 프로세서를 병렬처리하여 프로그램의 처리속도 향상을 꾀하는 멀티프로세서 환경으로 전환되어 가는 추세다. 따라서, 멀티프로세서 환경에서의 태스크 스케쥴링에 관한 다양한 여구가 필요하게 되었다. 일반적으로 멀티프로세서 환경에서는 주기 태스크를 스케쥴링하기 위해서 RMFF(Rate-monotonic First-Fit) 스케쥴링 알고리즘을 사용하는데, 이 알고리즘에서는 비주기 태스크의 스케쥴링을 위한 알고리즘을 추가로 요구한다. 따라서, 본 논문에서는 멀티프로세서 환경에서 주기 태스크와 비주기 태스크가 혼합되어 있을 때, 기존의 RMFF 방식에 슬랙 체크테이블(Slack Check Table;SCT)을 이용하여 비주가 태스크를 효율적으로 스케쥴링하기 위한 방법을 제안한다.

• The Journal of Engineering Research
• /
• 제2권1호
• /
• pp.73-82
• /
• 1997

In this paper, we examine design trends in the development of distributed real-time system. Many real-time systems developed over the last ten years share common characteristics including use of communications, imprecise computations, the object-oriented paradigm, multiprocessor node hardware, dynamic time-driven scheduling and the inclusion of fault tolerance mechanisms. By examining the major design decisions made in previous systems, the best attributes of these systems can be coalesced together to form the "ideal" real-time system. This paper examines such a system and the advantages and disadvantages of the design decisions involved.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• 제37권9호
• /
• pp.66-77
• /
• 2000

This paper presents a fast partitioning heuristic for hardware/software codesign called Best Imaginary Level-Iterative(BILI) partitioning which iteratively applies BIL heterogeneous multiprocessor scheduling heuristic to minimize the cost within the given time constraint. The proposed algorithm solves the partitioning problem with the implementation bin selection problem as well as architectures with multiple software modules. It costs about 15% less than the GCLP and at most about 5% more than the optimal solution obtained by the Integer Linear Programming(ILP) algorithm.

• Proceedings of the Korean Information Science Society Conference
• /
• 한국정보과학회 2003년도 가을 학술발표논문집 Vol.30 No.2 (1)
• /
• pp.280-282
• /
• 2003

많은 산업 분야에서 사용되는 실시간 시스템은 논리적 정확성뿐만 아니라 시간적 정확성을 요구한다. 그래서 실시간 시스템에서 동작하는 작업들은 항상 마감시간을 지키기 위해 작업에 대한 스케줄링이 매우 중요한 요소가 된다. 그리고 시스템의 신뢰도를 높이기 위해서는 고장 감내가 반드시 필요하다. 특히. 현대 기술의 발달로 일부 분야에서 사용되어 지던 실시간 시스템이 실시간 내장형 시스템 형태로 다양한 분야에 사용되어 진다. 내장형 시스템을 개발하는데 있어 자원의 절약 또한 하나의 중요한 요소이다. 따라서 본 논문에서는 실시간 시스템에서의 작업들이 마감 시간을 만족하며 고장 감내를 보장하는 시간 중복 기법과 백업 기법을 혼합하여 주기적으로 동작하는 작업들의 신뢰도와 자원의 효율 및 절약을 높이는 스케줄링 기법을 제안하고 실험한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 한국정보처리학회 2000년도 제13회 춘계학술대회 및 임시총회 학술발표 논문집
• /
• pp.550-556
• /
• 2000

본 논문은 NP-complete문제중의 하나인 순서제약이 있는 병렬프로그램을 멀티프로세서 시스템 상에서 효율적으로 분배하기 위한 유전자 알고리즘 설계 방법을 제안한다. 순서제약 조건을 만족하게 하는 새로운 염색체 코딩방법 및 휴리스틱한 스케줄링 알고리즘으로 정법한 해를 생성하고 프로세서 효율성을 고려한 평가 함수(evaluation function)와 우수한 유전인자를 이용하여 교배하는 교배연산자 등을 제안하였다. 그리고 제안한 알고리즘을 실험한 결과, 순서제약이 있는 다양한 형태(topology)의 병렬프로그램 스케줄링 문제에 대해서 제안한 유전자 알고리즘의 타당성을 확인하였다.

• Journal of Internet Computing and Services
• /
• 제1권2호
• /
• pp.19-28
• /
• 2000

The scheduling problem in parallel processing systems has been a challenging research issue for decades. The problem is defined as finding an optimal schedule which minimizes the parallel execution time of an application on a target multiprocessor system. Duplication Based Scheduling (DBS) is a relatively new approach for solving the problem. The DBS algorithms are capable of reducing communication overhead by duplicating remote parent tasks on local processors. Most of DBS algorithms assume an availability of the unlimited number of processors in the system. Since the assumption may net hold in practice, the paper proposes a new scalable DBS algorithm for a target system with limited number of processors. It Is shown that the proposed algorithm with N available processors generates the same schedule as that obtained by the algorithm with unlimited number of processors, where N is the number of input tasks. Also, the performance evaluation reveals that the proposed algorithm shows a graceful performance degradation as the number of available processors in the system is decreased.

• KIPS Transactions on Computer and Communication Systems
• /
• 제6권5호
• /
• pp.219-230
• /
• 2017

General-Purpose Graphics Processing Units (GPGPUs) build massively parallel architecture and apply multithreading technology to explore parallelism. By using programming models like CUDA, and OpenCL, GPGPUs are becoming the best in exploiting plentiful thread-level parallelism caused by parallel applications. Unfortunately, modern GPGPU cannot efficiently utilize its available hardware resources for numerous general-purpose applications. One of the primary reasons is the inefficiency of existing warp/thread block schedulers in hiding long latency instructions, resulting in lost opportunity to improve the performance. This paper studies the effects of hardware thread scheduling policy on GPGPU performance. We propose a novel warp scheduling policy that can alleviate the drawbacks of the traditional round-robin policy. The proposed warp scheduler first classifies the warps of a thread block into two groups, warps with long latency and warps with short latency and then schedules the warps with long latency before the warps with short latency. Furthermore, to support the proposed warp scheduler, we also propose a supplemental technique that can dynamically reduce the number of streaming multiprocessors to which will be assigned thread blocks when encountering a high contention degree at the memory and interconnection network. Based on our experiments on a 15-streaming multiprocessor GPGPU platform, the proposed warp scheduling policy provides an average IPC improvement of 7.5% over the baseline round-robin warp scheduling policy. This paper also shows that the GPGPU performance can be improved by approximately 8.9% on average when the two proposed techniques are combined.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• 제1권2호
• /
• pp.159-168
• /
• 1997

This paper presents a mechanism which supplies tasks with fast turn-around time on real-time multimedia environments. Tasks are classified into periodic and aperiodic tasks according to their executing period, and the types of them are classified into three groups : critical tasks, essential tasks and common tasks by the degree of its urgency. In the case of periodic tasks, we defer the execution of it within the extent to keep the deadline as long as possible and serve the aperiodic tasks, and provide aperiodic tasks with fast turn-around time. Changing the priority of each task is allowed within the same type and it is scheduled by using the dynamic priority. The emergency tasks are executed within deadline in any circumstances, and the least laxity one is served first when many real-time tasks are waiting for execution. The result of simulation shows that the proposed mechanism is better than the EDZL, known as suboptimal in multiprocessor systems, in the point of rum-around time.

• Journal of KIISE:Computing Practices and Letters
• /
• 제10권1호
• /
• pp.49-56
• /
• 2004

Load imbalance is a serious impediment to achieving good performance in parallel processing. Global load balancing schemes cannot adequately manage to balance parallel tasks generated from a single application. Dynamic loop scheduling methods are known to be useful in balancing parallel loops on shared-memory multiprocessor machines. However, their centralized nature causes a bottleneck for the relatively small number of processors in a network of workstations because of order-of-magniture differences in communication overheads. Moreover, improvements of basis loops scheduling methods have not effectively dealt with irregularly distributed workloads in parallel loops, which commonly occur in applications for a network of workstation. In this paper, we present a new reconfigurable and decentralized balancing method for parallel loops on a network of workstations. Since our method supplements performance balancing with those tranditional load balancing methods, it minimizes the overall execution time.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• 제21권11호
• /
• pp.2801-2809
• /
• 1996

Scheduling problems in real-time systems are known to be NP-hard. the heuristic approaches aregenerally aplied to solve a certain class of systems. One of such cases is to allocate periodic tasks to multiprocessors while the moethod assures the requirement of the deadine constraints of real-time systems. The study on the allocation of periodic taks includes RMNF, RMFF, FFDUF and Next-Fit-M algorithms, which make a set of task grups first and then allocate to processors. This papre proposes the various algorithms which are based on the Next-Fit-M. To analyze the four proposed methods, simulation was carried on, in which the sample tasks are randomly generated with the various time intervals. The proposed algorithms reduce the number of processors compared with the conventional methods.