The PE-Router Message Transmisson Mechanism for a Parallel Object-Oriented Total Architecture,A-NET.
スポンサーリンク
概要
- 論文の詳細を見る
The A-NET machine is adesigned as a part of a parallel object-oriented total architecture, which is to consist of a thousand nodes to gain massively parallel processing. Each node includes 2main units, Processing Element (PE) and Router(RT).The PE executes methods according to the messages received from the Router and send/receive messages to/from other PEs via RT. The Router, which acts as message and object forwarder, is connected to other nodes through six 8-bit buses. It also has one link to the Host Machine. The A- NET machine belongs to the direct network system category[2] and all the nodes are not directly connected to each other. Hence it is necessary to define a network topology for node connections. While most of the parallel computers have a predefined network topology ,A-NET machine is completely topology-programmable. As a result, any kind of topology; e.g.2D-mesh,3D-mesh, torus, hypercube, etc.,can be applied to the A-NET. In this paper the whole architecture of the RT is briefly d-escribed and the PE-Router internal-node transmission mechanism, which is performed by the PE-Router Interface(PEIF) unit, will be introduced in more detail.
- 一般社団法人情報処理学会の論文
- 1993-09-27
著者
-
BABA Takanobu
Department, of Information Science, Graduate School of Engineering, Utsunomiya University
-
Baba Takanobu
Department Of Information Science Utsunomiya University
-
Baba Takanobu
Department Of Electrical Engineering Nagaoka University Of Technology
-
YOSHINAGA TSUTOMU
Department of Information Science, Utsunomiya University
-
Pour Hossein
Department of Information Science, Utsunomiya university
-
Pour Hossein
Department Of Information Science Utsunomiya University
-
Yoshinaga Tsutomu
Department Of Information Science Utsunomiya University
関連論文
- Pipelined multithreading with clustered communication on commodity multi-core processors (コンピュータシステム)
- 1A-2 A Compiler Framework for Feedback-Directed Parallelizing Programs on CMP
- A Parallel Navigation Algorithm with Dynamic Load Balancing for OODBMSs
- The PE-Router Message Transmisson Mechanism for a Parallel Object-Oriented Total Architecture,A-NET.
- The Nature of the Specific Heat of Ising Spin Glass
- A Local Operating System for the A-NET Parallel Object-Oriented Computer
- Phase Transition Phenomena in Interconnection Networks of Massively Parallel Computers(Cross-disciplinary physics and related areas of science and technology)
- The Nature of the Specific Heat of Ising Spin Glass