• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.6
• /
• pp.80-87
• /
• 2016

We propose a parallel mesh smoothing algorithm using graph coloring and OpenMP library for shared memory many core computer architectures. The proposed algorithm partitions a mesh into independent sets and performs a parallel mesh smoothing using OpenMP library. We study the effect of using various graph coloring and color reordering algorithms on the efficiency of performing the proposed parallel mesh smoothing algorithm. We also investigate the influence of using various OpenMP loop scheduling methods on the parallel mesh smoothing efficiency.

• The Journal of the Korea Contents Association
• /
• v.18 no.6
• /
• pp.27-37
• /
• 2018

Streamline generation is one of the most representative visualization methods to analyze the flow stream of fluid dynamics dataset. It is a challenging problem, however, to determine the seed locations for effective streamline visualization. Meanwhile, it needs much time to compute effective seed locations and streamlines on the massive flow dataset. In this paper, we propose not only an importance based method to determine seed locations for the effective streamline placements but also a parallel streamline visualization method on the distributed visualization system. Moreover, we introduce case studies on the real fluid dynamics dataset using GLOVE visualization system to evaluate the proposed method.

• Proceedings of the Korea Contents Association Conference
• /
• /
• pp.513-516
• /
• 2007

Because many scientific applications require large data processing, the importance of parallel I/O has been increasingly recognized. Collective I/O is one of the considerable features of parallel I/O and enables application programmers to easily handle their large data volume. In this paper we measure and analyze the performance of original collective I/Os and the subgroup method, the way of using collective I/O of MPI effectively. From the experimental results, we found that the two kinds of subgroup method showed different performance. In terms of collective write operation, the subgroup method caused the performance degradation. However, the subgroup method for collective read showed good performance with small data size.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.14 no.3
• /
• pp.287-298
• /
• 2001

Analysis of nonlinear behavior of prestressed concrete frame structures on PC is a time-consuming computing job if the problem size increase to a certain degree. Cluster system has emerged as one of promising computing environments due to its good extendibility, portability, and cost-effectiveness, comparing it with high-end work-stations or servers. In this paper, a parallel nonlinear analysis procedure of prestressed concrete frame structure is presented using cluster computing. Cluster system is configured with readily available pentium III class PCs under Win98 or Linux and fast ethernet. Parallel computing algorithms on element-wise processing parts including the calculation of stiffness matrix, element stresses and determination of material states, check of material failure and calculation of unbalanced loads are developed using MPL. Validity of the method is discussed through typical numerical examples. For the case of 4 node system, maximum speedup is 3.15 and 3.74 for Win98 and Linux, respectively. Important issues for the efficient use of cluster computing system based un PCs and ethernet are addressed.

• Journal of KIISE:Databases
• /
• v.29 no.1
• /
• pp.58-71
• /
• 2002

Generally, multi-dimensional data such as image and spatial data require large amount of storage space. There is a limit to store and manage those large amount of data in single workstation. If we manage the data on parallel computing environment which is being actively researched these days, we can get highly improved performance. In this paper, we propose a parallel high-dimensional index structure that exploits the parallelism of the parallel computing environment. The proposed index structure is nP(processor)-n$\times$mD(disk) architecture which is the hybrid type of nP-nD and lP-nD. Its node structure increases fan-out and reduces the height of a index tree. Also, A range search algorithm that maximizes I/O parallelism is devised, and it is applied to K-nearest neighbor queries. Through various experiments, it is shown that the proposed method outperforms other parallel index structures.

• Proceedings of the Korean Information Science Society Conference
• /
• /
• pp.241-243
• /
• 2002

DNA 컴퓨팅은 생체 분자들이 갖는 막대한 병렬성을 정보 처리 기술에 적용한 기술이다. Adleman의 DNA 컴퓨팅은 랜덤한 고정길이의 형태로 문제를 표현하기 때문에 해를 찾지 못하거나 시간이 많이 걸리는 단점을 갖고 있다. 본 논문은 DNA 컴퓨팅에 DNA 코딩 방법을 적용하여 DNA 서열을 효율적으로 표현하고 반응횟수 만큼 합성과 분리 과정을 거쳐 최적의 코드를 생성하는 ACO(Algorithm for Code Optimization)를 제안한다. DNA 코딩 방법은 변형된 유전자 알고리즘으로 DNA 기능을 유지하며, 서열의 길이를 줄일 수 있으므로 최적의 서열을 생성할 수 있는 특징을 갖는다. ACO를 NP-complete 문제 중 Hamiltonian path problem에 적용하여 실험한 결과, Adleman의 DNA 컴퓨팅 보다 초기 문제 표현에서 높은 적합도 값을 갖는 서열을 생성했으며, 경로의 변화에도 능동적으로 대처하여 최적의 결과를 빠르게 탐색할 수 있었다.

• Journal of Digital Contents Society
• /
• v.17 no.1
• /
• pp.51-57
• /
• 2016

Distributed Computing attempts to harness a massive computing power using a great numbers of idle PCs resource distributed linked to the internet and processes a variety of applications parallel way such as bio, climate, cryptology, and astronomy. In this paper, we develop internet-distributed computing environment, so that we can analyze High Resolution Rainfall Prediction application in meteorological field. For analyze the rainfall forecast in Korea peninsula, we used QPM(Quantitative Precipitation Model) that is a mesoscale forecasting model. It needs to a lot of time to construct model which consisted of 27KM grid spacing, also the efficiency is degraded. On the other hand, based on this model it is easy to understand the distribution of rainfall calculated in accordance with the detailed topography of the area represented by a small terrain model reflecting the effects 3km radius of detail and terrain can improve the computational efficiency. The model is broken down into detailed area greater the required parallelism and increases the number of compute nodes that efficiency is increased linearly.. This model is distributed divided in two sub-grid distributed units of work to be done in the domain of $20{\times}20$ is networked computing resources.

• Korean Journal of Remote Sensing
• /
• v.28 no.4
• /
• pp.467-475
• /
• 2012

Cloud computing services are known that they have many advantages from the point of view in economic saving, scalability, security, sharing and accessibility. So their applications are extending from simple office systems to the expert system for scientific computing. However, research or computing technology development in the geo-spatial fields including remote sensing applications are the beginning stage. In this work, the previously implemented smartphone app for image processing was first migrated to mobile cloud computing linked to Amazon web services. As well, parallel programming was applied for improving operation performance. Industrial needs and technology development cases in terms of mobile cloud computing services are being increased. Thus, a performance testing on a satellite image processing module was carried out as the main purpose of this study. Types of implementation or services for mobile cloud varies. As the result of this testing study in a given condition, the performance of cloud computing server was higher than that of the single server without cloud service. This work is a preliminary case study for the further linkage approach for mobile cloud and satellite image processing.

• Journal of Scientific & Technological Knowledge Infrastructure
• /
• /
• pp.41-44
• /
• 2000

중국의 국가적인 슈퍼컴퓨터 개발노력의 총이는 최근 개발되어 소수의 국가기관에 설치되어 운영되기 시작한 Dewning-2000 초병렬형 시스템이다

• The Journal of the Convergence on Culture Technology
• /
• v.5 no.1
• /
• pp.395-399
• /
• 2019

In order to support various parallel computing systems, it is necessary to extend LLVM IR to more efficiently support vector / matrix and to design LLVM IR to machine code as a new algorithm. As shown in the IR example, RISC instruction generation is naturally generated because the RISC instruction is basically composed of the RISC instruction, and the vector instruction is also not supported. There is a need for new IR structures, command generation algorithms and related extensions to support vector / matrix more robustly. To do this, it is important to map each instruction in the LLVM IR to the appropriate instruction in the target architecture (vector / matrix) (instruction selection algorithm). It is necessary to understand the meaning of LLVM IR command, to compare the meaning of each instruction of the target architecture with syntax, and to select the instruction that matches the pattern to make mapping efficient.