Dataflow Computing Model - Perspectives, Advantages and Implementation

被引：0

作者：

Ivannikov, A. D. ^{[1
]}

Levchenko, N. N. ^{[1
]}

Okunev, A. S. ^{[1
]}

Stempkovsky, A. L. ^{[1
]}

Zmejev, D. N. ^{[1
]}

机构：

[1] Russian Acad Sci, Inst Design Problems Microelect, IPPM RAS, Moscow, Russia

来源：

2017 IEEE EAST-WEST DESIGN & TEST SYMPOSIUM (EWDTS) | 2017年

关键词：

D O I：

暂无

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

The problems of parallelizing computations and increasing real performance are among the main ones for high-performance computing systems. The article describes a new dataflow computing model and architecture, which allow addressing these problems. Methods for solving other problems of classical dataflow systems and traditional cluster supercomputers by the parallel dataflow computing system are also presented. The experiment results for one of the task are shown. It is concluded that the new dataflow computing model is promising.

引用

页数：4

共 9 条

[1]

Department of Energy (DoE), 2014, TOP 10 EX RES CHALL

[2]

Klimov A.V., 2016, P PROBL PERSP MICR 2, P151

[3]

[Климов Аркадий Валентинович Klimov Ark.V.], 2011, [Информационные технологии и вычислительные системы, Informatsionnye tekhnologii i vychislitel'nye sistemy], P3

[4] DATA-FLOW ARCHITECTURES AND MULTITHREADING [J].

LEE, B ;

HURSON, AR .

COMPUTER, 1994, 27 (08) :27-39

[5]

Maleki Saeed, 2016, P 2016 INT C SUPERCO

[6]

Rauber T., 2013, Parallel Programming: for Multicore and Cluster Systems

[7]

Robic Silc J.B., 1998, PARALLEL DISTRIBUTED, V1, P3

[8]

Stempkovskii A. L., 2008, Information Technologies, P2

[9]

Zmejev D.N., 2014, P PROBL PERSP MICR 4, P79

← 1 →