http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Tree 구조의 Data Flow Machine에서의 Grain-size 결정에 관한 연구
유관종,신현산 충남대학교 1985 忠南科學硏究誌 Vol.12 No.2
This dissertation discusses the grain-size of tree-structured data flow machine, AMPS & DDM1. We know that the factors which decide grain-size are each structure, communication overhead, and load balancing. Above factors affect the structural characteristics of tree-structured date flow machine. Therefore, we need to decide the grain-size appropriate to each data flow machine. As a result, we aim at Coarse-grain in AMPS and Fine-grain in DDM1.
Prodution Language로 表現된 Parser Generator의 設計에 관한 硏究
유관종,박두순 충남대학교 자연과학연구소 1983 忠南科學硏究誌 Vol.10 No.1
In general, a parser which is written by Floyd-Evans production language can be mechanically constructed from a weak procedence grammar. But all grammars are not weak precedence grammars. So, in this paper an algorithm for the mechanical generation of productions for a context-free grammar is presented.
유관종,최종석,신지수,변승환 충남대학교 자연과학연구소 1989 忠南科學硏究誌 Vol.16 No.1
A concept of data flow computer is a radical departure from that of Von Neumann type computers since the central concept of sequential execution is discarded. Data flow model offers an entirely new way of view the execution of program. This paper describes the data flow graph partitioning algorithm for enhancing the efficiency of the data flow architecture which consists of Processing Elements, Distributor, Arbitrator and Global Memory, and evaluate the performance of the proposed data flow system.
유관종 충남대학교 1991 忠南科學硏究誌 Vol.18 No.1
In this paper, we describe the existing scheduling algorithms and propose new heuristic schduling algorithm. This algorithm accepts program graph as input, and assign the nodes of graph, i. e. tasks, into processors and generates scheduled chart which describes the state of assignment of tasks and the processing overview of multiprocessor system. Program graph is a another representation of partitioned parallel program and we assume the execution time of each task and the communication time between tasks are given before scheduling with various values.
Clustered 데이타 흐름 그래프의 성능평가에 관한 연구
유관종,변승환 충남대학교 1991 忠南科學硏究誌 Vol.18 No.1
This paper is to evaluate the performance for the Partitioning Data Flow Graph. For the simulation, the input is a Partitioned Data Flow Graph and the target machine is a modified MIT Data Flow System. The goal of the performance evaluation is to compare between clustering data flow graph and no clustering data flow graph and to see the throughput between one PE and n PE in clustered data flow graph.