Parallel digital processing system using optical interconnection between control sections and data processing sections

• US 5,497,465 A
• Filed: 01/03/1995
• Issued: 03/05/1996
• Est. Priority Date: 03/25/1992
• Status: Expired due to Fees

• Current Assignee: Nippon Sheet Glass Company Limited
• Original Assignee: Nippon Sheet Glass Company Limited
• Inventors: Chin, Danny, Katsuki, Kazuo, Sauer, Donald J., Meyerhofer, Dietrich
• Primary Examiner(s): Lim, Krisna
• Application Number:
  US08/368,242
• Time in Patent Office: 427 Days
• Field of Search: 395/800, 395/325, 395/200, 395/280, 395/200.01, 359/128, 359/109, 385/16, 385/17, 385/31
• US Class Current: 710/100
• CPC Class Codes: G06F 15/17368 non hierarchical topologies
• CPC Class Codes: G06F 15/8007 single instruction multiple data [SIMD] multiprocessors

Abstract

MIMD and pipeline processing is executed by entering data and control signals into processing chips in the form of optical signals, and entering multi-bit information (data and control signals) in parallel and at high speed on the basis of non-coherence of light beams. The efficiency of MIMD processing function has been improved by expanding data transfer buses between processors and output buses in place of data and control signal input buses that have become unnecessary. A processing chip for receiving optical signals consists of a large number of cells dedicated for vector computations, and/or a large number of cells dedicated for vector computations and/or cells dedicated for arithmetic and logical computations. A processing chip for wide applications ranging from vector computations to logical computations by employing a construction combining both processors.

Claims

1. A parallel computer system for executing data processing in parallel and in a pipeline processing mode, said parallel computer system comprising:

a control section including a plurality of controllers;

a processing section including a plurality of processors, each one of the plurality of processors being associated with a respectively different one of the plurality of controllers;

an optical interconnection section for optically coupling each one of said plurality of controllers to the one of said plurality of processors with which it is associated, wherein said optical interconnection section transmits a plurality of optical signals from said one controller to said one processor, said optical signals comprising at least one of a plurality of optical data signals and a plurality of optical control signals;

a first electrical interconnection section, independent of said optical interconnection section, for coupling each one of said plurality of controllers to at least one of said plurality of processors, wherein said first electrical interconnection section transmits electrical control signals from said control section to said processing section;

a second electrical interconnection section, independent of said optical interconnection section and said first electrical interconnection section, coupled to said plurality of processors, for conveying data among said plurality of processors to the substantial exclusion of any control signals, wherein said second electrical interconnection section includes a plurality of inter-processor data transfer buses which interconnect said plurality of processors, wherein said plurality of inter-processor data transfer buses are controlled by a plurality of control signals provided by said control section.

2. A parallel computer system which performs multi-instruction and multi-data flow processing in a pipeline processing mode, said parallel computer system comprising:

a control section including a plurality of controllers and at least one memory;

a processing section including a plurality of processors each of the plurality of processors coupled to a respectively different one of said plurality of controllers;

an optical interconnection section for optically coupling each of said plurality of controllers to the respective one of said plurality of processors; and

an electrical interconnection section, independent of said optical interconnection section, for coupling said control section to said processing section, wherein;

a plurality of inter-processor data transfer buses which interconnect said plurality of processors to convey data signals substantially exclusive of any control signals among said plurality of processors, wherein said plurality of inter-processor data transfer buses are controlled by a plurality of control signals provided by said control section;

said optical interconnection section transmits a plurality of optical signals from each of said plurality of controllers to the respective one of said plurality of processors, wherein said optical signals include at least one of a plurality of optical data signals and a plurality of optical control signals;

said electrical interconnection section transmits electrical signals, including a plurality of control signals, from said control section to said processing section; and

said electrical interconnection section transmits electrical signals, including a plurality of data signals, from said processing section to said control section.

3. A parallel computer system as set forth in claim 2 wherein the plurality of controllers is located on at least one controller chip and the plurality of processors is located on at least one processor chip said at least one controller chip being different from said at least one processor chip.

4. A parallel computer system as set forth in claim 2 comprising a plurality of computer units defined by said plurality of controllers and said plurality of processors, wherein pipeline processing is executed in each of said computer units by transmitting a plurality of data signals and a plurality of instruction-control signals through said optical interconnection section from at least one of said plurality of controllers to at least one of said plurality of processors, and transmitting a plurality of data output control signals from said controllers to said processors.

5. A parallel computer system as set forth in claim 2 including a plurality of computer units in which said plurality of controllers and said plurality of processors are interconnected, and pipeline processing is executed in each of said computer units by transmitting at least one of said plurality of data signals and at least one of said plurality of instruction-control signals through the optical interconnection section from the control section to the processor section, and transmitting at least one of said plurality of interconnection-control signals and at least one of said plurality of data signals through said electrical connection section from said control section to said processor section.

6. A parallel computer system as set forth in claim 2, said parallel computer system further including a plurality of computer units, each formed by one of said plurality of controllers interconnected with the respective one of said plurality of processors, wherein pipeline processing is executed in each of said computer units by transmitting different data from each of said plurality of controllers to the respective one of said plurality of processors through said optical interconnection section, and by transmitting a plurality of instruction-control signals from said plurality of controllers to said plurality of processors.

7. A parallel computer system as set forth in claim 3 wherein said plurality of inter-processor data transfer buses are wired to connect every processor on said processor chip with the adjacent processors in the left and right, and in the up and down directions;

in which one of said transfer buses is used to read data and another one of said transfer buses is used to write data, said transfer buses including circuits in the peripheral part of said processor, wherein read and write operations in said processor are executed by a tri-state switch located between said processors and gated by control logic circuitry in said processors which are operated by a plurality of optical control signals transmitted from said control section to said processing section.

8. A parallel computer system as set forth in claim 7 wherein said read and write inter-processor buses can directly connect any two of said processors located in one row or in one column of said chip by use of bypass switches in the peripheral part of each of said processors, said bypass switches being controlled by said control logic circuit operated by a plurality of optical control signals transmitted from said control section to said processor section.

9. A parallel computer system as set forth in claim 8 wherein additional control buses between said processors on said processor chip are used to request a processor to perform data transfer between other than its nearest processors and to return acknowledge signals from said requested processor, to make master-slave relations among said processors by turning on said bypass switches in the intermediate processors, located between said master processor and said slave processor, and to transfer signals to operate the tri-state switches of said inter-processor data transfer buses from said master processors to said slave processors and to said intermediate processors, wherein control signals making said master-slave relations among said processors on said processor chip are transferred to said control logic circuits in said processors in response to optical control signals provided by said control section to said processing section.

10. A parallel computer system having a control section including a plurality of controllers, a processing section including a plurality of processors each of said plurality of processors being interconnected to a respectively different one of said plurality of controllers, an optical interconnection conveying a plurality of optical signals, including at least one of a plurality of optical data signals and a plurality of optical control signals, between each of said plurality of controllers and the respective one of said plurality of processors, a first electrical interconnection means, independent of said optical interconnection, for interconnecting said control section and said processing section, and a second electrical connection means, independent of said optical interconnection and said first electrical interconnection means, distinct from the first electrical connection means for conveying data among said plurality of processors wherein each of the processors is operating an individual program, and wherein the coordination of the operation of the programs by said plurality of processors is executed by receiving as flag signals a state of program execution transmitted from one of said plurality of processors to said control section.

11. A system control block as set forth in claim 10 wherein said flag signals are transmitted when one of said plurality of processors enter into relations in terms of program execution with others of said plurality of processors, and interrupt the execution of said programs being operated by said plurality of processors.

12. A system control block as set forth in claim 10 wherein said flag signals are reset to cause all of said processors to restart when all of said processors generate said flag signals and stop operating.

13. A system control block as set forth in claim 10 wherein system control function of said flag signal creation is distributed to said plurality of controllers in said control section.