Background of the Invention
1. Field of the Invention
The present invention relates to a combination of processors that have a uniform transmission processing function, and more specifically to a combined computer system that is constituted in a one-dimensional to three-dimensional manner.
2. Description of the Prior Art
Network systems for connecting a plurality of processors (computers) have heretofore been either a bus structure, a loop structure, a star structure, and a mesh structure. The first three of these structures can not be constructed in a three-dimensional configuration. In a mesh structure in which processors are connected through bidirectional transmission lines, routing is complex. In a star structure, the processor at the center of the structure is different from the other processors. In a mesh structure, a particular processor must work to cope with any change in the structure of the system or with any abnormal condition that may occur, thereby making it difficult to expand and maintain the system.
Summary of the Invention
To preclude the defects inherent in the conventional art, the present invention provides a method for constituting processors that serve as constituent units of a system in which very uniform constituent units are connected through a plurality of looped transmission lines, which can be constructed in two-dimensional and three-dimensional configurations, and to which a simply constructed transmission system can be adapted.
For this purpose, the present invention deals with a structure which comprises a first group of looped transmission lines consisting of a plurality of looped transmission lines, a second group of looped transmission lines that consist of at least one looped transmission line and that are arranged to intersect said first group of looped transmission lines, transmission lines that couple said intersecting transmission lines together, and data processing and controlling means that control the transmission of data along the transmission lines and that process the data, and wherein, as required, the number of building blocks is increased or decreased for each of the looped transmission lines.
Brief Description of the Drawings
FIGS. 1(a), 1(b) and 1(c) are diagrams of constituent processor units as employed by the present invention;
FIG. 2 is a diagram showing the connection of a plurality of processor units in the form of a network;
FIG. 3 is a diagram showing a plurality of processor units connected in a different form of network;
FIG. 4 is a diagram showing a plurality of processor units connected in another form of network;
FIG. 5 is a diagram showing a plurality of processor units connected in another form of network;
FIG. 6 is a diagram showing a plurality of processor units connected in another form of network;
FIG. 7 is a diagram showing a plurality of processor units connected in another form of network; FIG. 8 is a diagram showing a plurality of processor units connected in another form of network;
FIG. 9 shows a plurality of processor units connected in another form of network;
FIG. 10 shows a plurality of processor units connected in another form of network;
FIG. 11 shows a plurality of processor units connected in another form of network;
FIG. 12 shows a plurality of processor units connected in another form of network;
FIG. 13 is a diagram showing a plurality of processor units connected in another form of network; and
FIG. 14 is a schematic diagram of a data processing and controlling unit according to the present invention.
Description of the Preferred Embodiments
Three kinds of constituent elements as shown in FIGS. 1(a), 1(b) and 1(c) may be combined in the form of building blocks to constitute a system. The constituent unit consists of a processor and unidirectional transmission lines for transmitting data therefrom to other processors. Transmission lines of neighboring processors are connected to the unidirectional transmission lines so as to form pairs and so as to receive to data therefrom. The constituent units shown in FIGS. 1(a), 1(b) and 1(c) consist of processors 10, 20 and 30, and two, three and four transmission lines (101, 102), (201, 202, 203), and (301, 302, 303, 304), respectively. To the transmission lines 101, 102, 201, 202, 203, 301, 302, 303 and 304 are connected transmission lines 10k, 10l, 20k, 20l, 20m, 30k, 30l, 30m and 30n of the neighboring processors to form processor pairs. FIGS. 2 to 13 show transmission lines of the processors forming pairs, that are connected successively. The constituent units are all the same, hence, each of the transmission lines constitutes a portion of a closed loop transmission line. Here, the constituent unit may have any number of unidirectional transmission lines in addition to the above-mentioned numbers.
FIGS. 2 to 5 illustrate examples of connecting the constituent units that have two unidirectional transmission lines. Namely, FIG. 2 illustrates an example in which the processors are divided into two groups consisting of (10, 11, 12) and (13, 14, 15) which are coupled through a loop (101, 111, 121) and a loop (131, 151, 141). The processors (10, 13), (11, 14) and (12, 15) on these different loops are coupled via a loop (102, 132), a loop (112, 142) and a loop (122, 152).
FIG. 3 shows an example in which all of the processors are coupled through the same loops.
FIG. 4 illustrates an example in which the processors are divided into three groups (10b, 11b, 12b, 13b), (14b, 15b, 16b, 17b) and (18b, 19b, 20b, 21b) which are coupled in the form of loops. Unlike the example of FIG. 2, however, processors (10b, 11b), (12b, 13b), (18b, 19b) and (20b, 21b) on the same loops are coupled. In this regard, the example of FIG. 4 resembles the example of FIG. 3. However, other processors (14b, 15b, 16b, 17b) are connected to constitute loops that traverse the above-mentioned three loops (101b to 131b), (141b to 151b) and (181b to 211b). In this regard, the example of FIG. 4 is the same as the connection method of FIG. 2.
FIG. 5 shows an example in which processors are divided into three groups (10c, 11c, 12c, 13c, 14c, 15c), (16c, 17c, 18c, 19c, 20c, 21c) and (22c, 23c, 24c, 25c, 26c, 27c) which are coupled in the form of loops. Unlike the example of FIG. 4, however, the processors 11c, 14c and the processors 23c, 26c exist on the loops (101c to 121c), (221c to 231c) and are not adjacent to each other.
In the arrangement of FIG. 2, if the system is to be expanded without changing the constitution of the double loop consisting of a loop (101, 111, 121) and a loop (131, 151, 141), it is necessary to increment the number of processors by one for each of these loops. Hence, it is necessary to increase the number of processors by two. If it is based on loops (102, 131), (112, 142) and (122, 152), the processors can be incremented by one for each of the loops. That is, the processors can be increased by a total of three.
In the arrangement of FIG. 3 based upon a double loop structure, the processors can be incremented by a unit of one. Based upon the same idea as that of FIG. 2, the processors in the arrangement of FIG. 4 can be increased by a unit of six or by a unit of four. In the arrangement of FIG. 5, in the same manner, the processors can be increased by a unit of six processors. Increase in the number of processors changes the structure of system. Therefore, the number of processors to be increased may change thereafter.
FIGS. 6 to 10 illustrate methods for connecting constituent units that have three unidirectional transmission lines. In FIG. 6, the three processors 20, 21 and 22 are connected by loops (202, 212), (211, 221), (203, 223) that connect between the two processors, and by loops 201, 213, 222 of single processors.
In FIG. 7, the four processors are connected by triple loops.
FIG. 8 shows an example in which a processor is added to the structure of FIG. 6. FIG. 9 shows an example in which three processors are added to the structure of FIG. 6. FIG. 10 illustrates an example in which three processors are added to the structure of FIG. 9. In the structures of FIGS. 7 to 10, the processors can be incremented by any integer greater than one.
FIGS. 11 to 13 illustrate methods of connecting constituent units that have four unidirectional transmission lines. FIG. 11 shows an example in which nine processors are arranged in a two-dimensional manner, FIG. 12 shows an example in which three processors are arranged in a one-dimensional configuration, and FIG. 13 shows an example in which 32 processors are arranged in a three-dimensional configuration. FIG. 13 is an expansion of the structure of FIG. 11.
FIG. 14 shows the internal structure of the constituent units 10, 20 and 30. Responsive to the numbers 2, 3, 4 of unidirectional transmission lines of the constituent units 10, 20 and 30, however, there exist two, three or four interfaces connected to the loop transmission lines. The processor consists of a terminal 3000 and a data processing and controlling unit 300 which controls the transmission or reception of information relative to the loop transmission lines.
The terminal 3000 consists of a terminal device 3005 and interfaces 3001 to 3004 between the terminal device and the data processing and controlling unit. Here, the four interfaces 3001 to 3004 make connection with different data processing and controlling units, and may be provided in any number greater than one.
Described below is a process in which data generated from the terminal device 3005 is transmitted to a predetermined terminal via a loop transmission line.
Data generated from the terminal device 3005 is sent to data processing and controlling units via interfaces 3001 to 3004. Data sent from the terminal device via an interface 322 is then stored in a sending buffer 317. Data is provided with an address (receiving address) of a terminal where the data will be received or a code (content code) corresponding to the content of the data. In a routine table 323 a program has been written which determines in which loop direction the data should be transmitted corresponding to the address or the content code. The program has been written onto the routine table 323 prior to sending the data. When data exists in the sending buffer 317, the operational unit 315 sends that data from any one of the interfaces 311 to 314 relying upon the routine table 323. If the data thus sent returns after having circulated through the loop, the operational unit 315 judges that the data is properly transmitted, and erases the data from the sending buffer 317. Here, the data that is to be sent is provided with an address of the interface that sends the data. The address of the interface has been stored beforehand in the ROM of interface address table 321.
When data arrives on the loop, the data processing and controlling unit 300 introduces the data through, for example, an interface 311, reads the address of the interface 311 from the interface address table 321, attaches the interface address, and stores the data in a receiving buffer 316. If the data received is not the data that is generated from the data processing and controlling unit 300, the data is then sent through the interface 311 that has received the data. If the data is the data that is generated from the data processing and controlling unit 300, that data is erased from the receiving buffer 316 and from the sending buffer 317 as mentioned above.
As the data is taken into the receiving buffer 316, the operational unit 315 transfers the data to the terminal via the interface 322 provided the receiving address or the content code of the data is in agreement with that of a receiving function code table and address 320. Next, if the receiving address or the content code of the data is in agreement with that in the routine table 323, said data is sent in the direction of the transmission loop written thereon through interfaces 311 to 314 of directions other than the loop direction in which the data is received. When the data is sent to the terminal and to the loop direction, the data is erased from the receiving buffer 316.
When the data in the sending buffer 317 and the data in the receiving buffer 316 are to be sent for the first time, each on its own loop, the data processing and controlling unit sets the timer 318 for each of the data. If the data does not return even after a predetermined period of time has passed, the data processing and controlling unit sets the flag of an abnormal status register 319 corresponding to the loop into which the data is sent.
As the flag is set in the abnormal status register 319, the operational unit 315 broadcasts the abnormal condition on the loop in all directions via the interfaces 311 to 314. Upon receipt of the data which reports the abnormal condition, the data processing and controlling units broadcast the abnormal condition to the loop directions in which the data reporting the abnormal condition has not yet been received. The processor which has received the data that reports the abnormal condition converts the routine table 323 corresponding to the abnormal loop. For instance, a by-pass route may have been determined to cope with the abnormal condition that may take place in the loops, or if an abnormal condition develops somewhere, the data may be broadcast in all loop directions thereafter.
In the receiving function code table and address unit 320 there has been stored in advance a content code or an address of the data required by the terminal device 3005 that is connected to the data processing and controlling unit 300.
According to the system of the present invention, the defective portion may be separated to be checked. In this case, transmission of data is continued through other portions. Further, since the constituent units are all the same, maintenance can be easily carried out.
According to the present invention as described above, identical constituent units are successively connected to each other to form a plurality of looped structures. By changing the method of connection, a communication network can be constituted that can be easily changed into a one-dimensional, two-dimensional or three-dimensional structure. Further since it is based upon a looped network, transmission routing can be easily accomplished. Moreover, all of the constituent units have the same hardware and software, and no apparatus having particular functions such as a master apparatus is needed, thereby offering an advantage in regard to maintenance.