• Journal of the Korea Society of Computer and Information
• /
• v.19 no.6
• /
• pp.101-108
• /
• 2014

The parallel machine scheduling is to schedule each job to exactly one parallel machine so that the total completion time is minimized. It is used in various manufacturing system areas such as steel industries, semiconductor manufacturing and plastic industries. Each job has a setup phase and a processing phase. A removal phase is needed in some application areas. A processing phase is performed by a parallel machine alone while a setup phase and a removal phase are performed by both a server and a parallel machine simultaneously. Most of previous researches used a single server and considered only a setup phase and a processing phase. If a single server is used for scheduling, the bottleneck in the server increases the total completion time. Even though the number of parallel machines is increased, the total completion time is not reduced significantly. In this paper, we have proposed an efficient algorithm for the parallel machine scheduling using multiple servers and considering setup, processing and removal phases. We also have investigated experimentally how the number of servers and the number of parallel machines affect the total completion time.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2010.11a
• /
• pp.1753-1756
• /
• 2010

최근 들어 가상화 기술은 서버의 통합에 뿐만 아니라, 임베디드 시스템에서도 널리 사용되고 있다. 하지만, 가상화 시스템에서는 물리 프로세서가 게스트 운영체제에게 직접 전달되지 않으며, 게스트 운영체제는 가상 프로세서를 통해서 실행할 수 밖에 없다. 따라서, 기존의 처리량 기준의 공평성 스케줄러가 가상머신 모니터에서 동작하는 경우, 실시간 스케줄링이 불가능하다. 본 연구에서는 멀티코어 기반의 가상화 시스템에서 실시간 태스크의 실행을 보장하는 기법을 소개한다. 특히, 본 논문에서는 계층형 스케줄링의 특성과 최대 병렬성 조건을 통하여 멀티코어 가상머신의 스케줄 가능성 분석 기법을 제시한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.10
• /
• pp.2303-2308
• /
• 2012

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.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.49-54
• /
• 2022

With the recent widespread adoption of general-purpose GPUs (GPGPUs) in cloud systems, maximizing the resource utilization through multitasking in GPGPU has become an important issue. In this article, we show that resource allocation based on the workload classification of computing-bound and memory-bound is not sufficient with respect to resource utilization, and present a new thread block scheduling policy for GPGPU that makes use of fine-grained resource utilizations of each workload. Unlike previous approaches, the proposed policy reduces scheduling overhead by separating profiling and scheduling, and maximizes resource utilizations by co-locating workloads with different bottleneck resources. Through simulations under various virtual machine scenarios, we show that the proposed policy improves the GPGPU throughput by 130.6% on average and up to 161.4%.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.10a
• /
• pp.301-303
• /
• 2003

사용자 쓰레드와 시스템 쓰레드간의 1:1 맵핑 모델은 병렬성을 지원하는 장점이 있고, M: N 맵핑은 병렬성 지원과 빠른 문맥교환의 장점이 있다. 리눅스 자바 가상 머신에서는 1:1 맵핑 모델만을 지원한다. 연성 실시간을 보장하기 위해서는 쓰레드간의 문맥교환을 최소화하여 성능 향상시킬 필요가 있다. 이 논문에서는 자바 어플리케이션 레벨에서 경량 프로세스(Light Weight Process, LWP) 개념을 도입하여 리눅스 자바 가상 머신에서 M: N 맵핑을 지원하는 자바 쓰레드 모델을 제안한다. 제안한 모델은 그린 쓰레드 (Green Thread)의 빠른 문맥교환과 네이티브 쓰레드(Native Thread)의 병렬성 지원 장점을 혼합한 것으로 빠른 처리속도와 자바 플랫폼의 독립성을 그대로 유지할 수 있다. 또한, MTR-LS 알고리즘을 경량 프로세스 스케줄링에 채택함으로서, 자바 응용프로그램의 연성 실시간을 보장한다. 1:l 및 M:1 맵핑 모델과의 성능 비교를 통해 제안한 모델이 좋은 성능과 연성 실시간을 보장한다는 것을 보인다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.05a
• /
• pp.123-124
• /
• 2013

비용대비 작업 처리의 효율을 위해서는 사용자들의 작업 요구사항에 적절한 자원을 선택하고 요구 작업을 적절한 할당된 자원에 스케쥴링하는 플랫폼이 필수적이다. 또한 이러한 플랫폼은 사용자의 SLA 에 따라 작업 처리 기한 안에 요구 비용 이내로 작업을 처리할 수 있도록 결정을 내릴 수 있어야 하고 요구 작업량의 변화에 따라 즉각 대응을 하기 위하여 실시간적인 결정을 내릴 수 있어야 한다. 이러한 복잡한 결정 사항들을 최적 판단으로 대신 처리해주는 미들웨어로 클라우드 자원 브로커 시스템을 사용할 수 있다. 클라우드 자원 브로커 시스템은 작업 스케쥴링과 자원 프로비저닝 등이 가격, 처리시간에 중요한 선택 및 수행을 한다. 기존의 많은 논문들에서의 작업 스케줄링은 다중 테넌트 정책의 클라우드가 제공하는 사용자들간의 가상 머신 독립에 초점을 두어 하나의 가상 머신이 하나의 작업에 한정되도록 처리하는 방식이었다. 이는 병렬화의 정도가 낮은 어플리케이션의 경우 시스템 활용률이 낮아 자원 활용율이 떨어진다. 이를 다수의 작업을 멀티 태스킹, 멀티 스레드의 방법으로 하나의 가상 머신에서 처리하도록 하여 스레드 레벨 병렬화의 이점을 이용해 자원 이용률을 높임으로 효율을 높이고자 한다.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.5
• /
• pp.48-61
• /
• 2017

An IaaS service of a cloud computing environment makes itself attractive for running large scale parallel application thanks to its innate characteristics that a user can utilize a desired number of high performance virtual machines without maintenance cost. The total execution time of a parallel application on a high performance computing environment depends on a task scheduling algorithm. Most studies on task scheduling algorithms on cloud computing environment try to reduce a user cost, and studies on task scheduling algorithms that try to reduce total execution time are rarely carried out. In this paper, we propose a task scheduling algorithm called an HAGD and a performance enhancement method called a group task duplication method of which the HAGD utilizes. The group task duplication method simplifies previous task duplication method, and the HAGD uses the group task duplication method or a task insertion method according to the characteristics of a computing environment and an application. We found that the proposed algorithm provides superior performance regardless of the characteristics in terms of normalized total execution time through performance evaluations.

• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.9
• /
• pp.189-196
• /
• 2020

Recently, GPU cloud computing technology applying GPU(Graphics Processing Unit) devices to virtual machines is widely used in the cloud environment. In a cloud environment, GPU devices assigned to virtual machines can perform operations faster than CPUs through massively parallel processing, which can provide many benefits when operating high-performance computing services in a variety of fields in a cloud environment. In a cloud environment, a GPU device can help improve the performance of a virtual machine, but the virtual machine scheduler, which is based on the CPU usage time of a virtual machine, does not take into account GPU device usage time, affecting the performance of other virtual machines. In this paper, we test and analyze the performance degradation of other virtual machines due to the virtual machine that performs GPGPU(General-Purpose computing on Graphics Processing Units) task in the direct path based GPU virtualization environment, which is often used when assigning GPUs to virtual machines in cloud environments. Then to solve this problem, we propose a GPGPU task management method for a virtual machine.

• Journal of KIISE:Computer Systems and Theory
• /
• v.27 no.5
• /
• pp.450-463
• /
• 2000

Array redistribution is usually required to enhance algorithm performance in many parallel programs on distributed memory multicomputers. GEN_BLOCK redistribution, which is redistribution between different GEN_BLOCKs, is essential for load balancing. However, prior research on redistribution has been focused on regular redistribution, such as redistribution between different CYCLIC(N)s. GEN_BLOCK redistribution is very different from regular redistribution. Message passing in regular redistribution involves repetitions of basic message passing patterns, while message passing for GEN_BLOCK redistribution shows locality. This paper proves that two optimal condition, reducing the number of communication steps and minimizing redistribution size, are essential in GEN_BLOCK redistribution. Additionally, by adding a relocation phase to list scheduling, we make an optimal scheduling algorithm for GEN_BLOCK redistribution. To evaluate the performance of the algorithm, we have performed experiments on a CRAY T3E. According to the experiments, it was proven that the scheduling algorithm shows better performance and that the conditions are critical in enhancing the communication speed of GEN_BLOCK redistribution.