JP3338059B2 - Computer network system with transfer identification and arbitration mechanism - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION 【Technical field】

The present invention relates to a network connecting a large number of computers. The network according to the present invention triangulates a point-to-point channel and a bus-type broadcast channel.
Construct a topology that combines structures. This is a network system in which input / output devices connected to each computer are distributed as an arbitration mechanism that eliminates contention for transfer data without centrally managing one unit. According to the method of the present invention, a multi-CPU (central process
ing unit), LAN (local aria network), WAN (wide ari)
a network).

[Background Art]

The input / output device to the network is IMP (interface mess
age processor), packet switching node (packet switch)
node), relay system (intermediate system), data switching exchange (data switching exchange), router (route
r), bridge, ring interface (rin
g interface), repeater, ... in various terms. Point-to-point channels require dedicated transmission paths between computers. Therefore, when many computers are connected, laying the transmission path becomes complicated. Also, since many transmission paths are connected to one computer, it was necessary to decide which transmission path to select. An example is PCT / JP87 / 00787 [Data transfer system with transfer identification circuit]. The broadcast channel does not require a dedicated transmission line between computers, and the network is completed by connecting to the transmission line after laying the transmission line. However, since many computers are connected to one transmission line, there are times when transfer data competition occurs. As a method of avoiding transfer data contention, a common name for Ethernet (IEEE (Institute of
Electrical and Electronic Engineers (802.3) 1- Sustained CAMA / CD (carrier sense multiple acce)
In ss with collision detection-LAN), when a data collision occurs, the computer waits for transmission for a random time. This is fine if the traffic is small. However, with a large amount of traffic, a computer that wants to transmit first may be overtaken by a computer that wants to transmit later. Due to this problem, IEEE 802.4 token bus was used in machine control. The IEEE 802.4 Token Bus and IEEE 802.5 Token Ring are methods to eliminate transfer data contention by controlling input / output devices connected to each computer and distributing them without centralized management by a server unit. PCT / US87 / 02388 [I / O network for computer systems] is an example of arbitration performed by a microcomputer in an I / O device. In any case, in order to expand the network, it had to be used with other methods. To avoid contention for transfer data in a very short distance network, it was common to provide a switching device or arbiter device. PCT / FR81 / 00029 [rapid message transmission method between computers] is an example of having a switching device. As an example with an arbiter device, P
CT / DE85 / 00525 [multiprocessor system] and PCT / US8
8/02955 [Method and apparatus for interconnecting buses in a multi-bus computer system]. The logic synthesis of digital signals is based on the use of logic gate elements. However, as a method of performing logic synthesis without using a logic gate, wired logic (wi
red-logic). Wired logic is connected to the output of open collectors directly with a pull-up resistor.
e) and wired-OR, in which the outputs of open emitters are directly connected by pull-down resistors.

[Object of the invention]

Even if you remove a computer or add a computer via an I / O device without changing any I / O devices or connected computers to the network, messages will be sent between all the computers. Providing a transferable network eliminates the need for a post computer or host server.

[Definition of terms used in this specification]

1. Input / output devices to the network are called IMPs, packet switching nodes, relay systems, data switches, routers, bridges, repeaters, etc., and no unified concept has been established. However, transfer message destination control devices are widely separated from server units. In this specification, distributed input / output devices to a network are collectively referred to as a “MARS (Message Access and Repeat Server) unit”. 2. A computer is defined as meeting the following conditions. A. There are CPU and CPU bus. (It may be singular or plural.) B. There are RAM and I / O devices connected to the CPU bus, and the CPU starts control of the memory and I / O devices by programs in the ROM. C. MARS units are independent devices, not I / O or computers. By the above definition, an electric typewriter is not a computer. However, the printer controlled by the internal CPU is a computer, and the printer controlled by the external CPU is an I / O device. The MARS unit is a device outside the concept of a computer whether it is controlled by its own CPU or by an external CPU. 3. This specification is to disclose a method for a computer network, and is described as positive logic except for a transmission path.

SUMMARY OF THE INVENTION

Fig. 1, Fig. 2 and Fig. 3 are explained,
The outline of the transfer / reception protocol is described in a concrete example.
Helps to draw the concept of. FIG. 1 shows the MA described in [Definition of phrases used in this specification].
Insert the RS (Message Access and Repeat Server) unit
This is the output appearance. Fig. 2 shows the connection of many MARS units in the network.
Here is an example of a LAN. Fig. 3 shows messages in LAN, conforming to IEEE802.3
This is a specific example of the frame format to be used. The MARS unit in Fig. 1 will be described. * MARS unit 11 has two different transmission line terminals and one
It has a CPU bus connection terminal 12. One of the two transmission path terminals
Is the upper transmission line terminal 13 and the other is the lower transmission line terminal 14.
It is. * Each transmission line terminal is a different broadcast
Connected to a multi-channel transmission path bus or point-to-point
Connect to the transmission channel of the INT channel. * The transmission path connected to the upper transmission path terminal 13 is the upper transmission
Road. * The transmission path connected to the lower transmission path terminal 14 is the lower transmission
Road. * The names of the upper transmission line and the lower transmission line are
When viewed from the transmission path, the upper transmission path
Terminal or lower transmission path terminal. * The bus connected to CPU bus connection terminal 12 has memory 15
And a computer consisting of CPU16 are connected.
(Refer to the above [Definition of Words].) * Control of MARS Unit 11
It can be controlled by microprocessor or CPU16.
I don't know. The outline of the LAN of the present invention will be described with reference to FIG. * The network is completed by connecting the MARS unit and the transmission path.
Be completed. * All transmission paths are numbered, and a lowercase letter is added to the left of the transmission path.
Described in. * All MARS units are represented by rectangles, and inside the rectangles
The number is described in. * Memory 15 and CPU 16 connected to MARS unit are listed
Was omitted. * Fig. 2 is an example of a network, one digit used
The maximum number of was set to 7. * Transmission lines are divided into LAN transmission lines and WAN transmission lines.
It is. * The transmission line for LAN is indicated by a solid line with a width of 0.4 mm.
Is a broadcast channel and has many MARS units.
It is a bus type that connects to * The transmission line for WAN is indicated by a dashed line with a width of 0.4 mm.
Is a point-to-point channel, one MARS unit
Unit and one other MARS unit. (See Fig. 4
*) * All MARS units are placed on some layer.
Belong. * Only one MARS unit belongs to the highest layer.
And has the highest unit number in the LAN. (In Figure 2
* 7777) * The number of the LAN transmission line must have '0' in some digits.
There is only one. * MARS unit numbers do not have '0' in all digits. * The number of the WAN transmission line is the same as the MARS unit to be connected.
One number. * In one LAN transmission line, where in the LAN transmission line number
MARS unit with a number that replaces '0' with another number
Only the units are directly connected. Fig. 3 shows an example of a message.
Matt. * IEEE802.3 is preamble and frame start delimiter
Is added before the destination address and the checksum
Are added at the end. However, the present invention
It is omitted because it provides a method for constructing a work. * The message 31 is sent from the CPU 16 in FIG.
Message sent to the packet 11 and
This is the message flowing to the LAN. * Messages flowing on the transmission line are stored and transferred completely.
(Store-and-forword).
It is necessary to temporarily store the message. These messages are stored in the internal storage of MARS unit 11.
Even if you remember. It may be stored in the memory 15. The message 31 in FIG. 3 is the CPU 3167, and the network 31 in FIG.
3167 → (3107) → 3177 → (3077) → 3677 → (3607) → 36
An example in which transfer is performed in a chain is shown in the figure.
The concept of the file is explained. No message contention occurred due to the brief description
Suppose. ◎ Protocol of computer 3167 and MARS unit 3167
1. The CPU 3167 sends the message 31 to the MARS unit via the CPU bus.
Unit 3167, and the unit receives the message.
(Transfer in Fig. 1) 2. MARS unit 3167 receives the message and writes it.
Judge whether the address is correct or not. 3. If there is no error, it transforms itself into a “transfer unit”.
However, if there is an error, it is returned to CPU16. ◎ Protocol of transfer unit 3167: 1. Judgment of selection of transfer transmission path (upper / lower). Since the transfer unit 3167 has the layer number 2, its own number 31
Whether the second digit or more of the first 3657 of 67 and 31 match
to decide. Since it does not match “316x ≠ 365x”, select the upper transmission line
I do. 2. When the upper transmission path is selected;
Fixed. Since the transfer unit 3167 has the layer number 2, its own number 31
Whether the value matches by masking the second and lower digits of the first 3657 of 67 and 31
Judge whether or not. Since it does not match “31xx ≠ 36xx”, the maximum number '7
Send to number. That is, from the upper transmission path 3107,
Send it to the slot number 3177. Explanation: The upper transmission line of 3167 is 3107.
Units are 3117, 3147, 3167, 3177, and 3107
Change '0' to the maximum single digit number '7' used in the LAN
3177 can be determined. ◎ Protocol of MARS unit 3177: 1. MARS unit 3177 turns into a “receiving unit”
Receives message 31. 2. Judge whether it is a self-addressed or forwarded message. Judge based on the match between self-number 3177 and the leading 3657 of 31
It is. "Transfer unit" because it does not match "3177 ≠ 3657"
Transform into. ◎ Protocol of transfer unit 3177: 1. Judgment of selection of transfer transmission path (upper / lower). Since the transfer unit 3177 has the layer number 3, its own number 31
Whether the 3rd digit or more of the first 3657 of 77 and 31 match
to decide. Since it does not match “31xx ≠ 36xx”, select the upper transmission line
I do. 2. When the upper transmission path is selected;
Fixed. Since the transfer unit 3177 has the layer number 3, its own number 31
Masks the 3rd and lower digits of the first 3657 of 77 and 31 to match
Judge whether or not. Matched with “3xxx = 3xxx”. If they match, the layer number is 3 and the top 3657 of 31
The third digit 6 is adopted and transmitted to the upper transmission line No. 6. That is, from upper transmission path 3077 to unit number 3677
Send. Remarks: In the message received from other than CPU terminal,
If the message has no error
If it is received at the lower transmission line terminal,
Yes, if a signal is received at the upper transmission path terminal,
is there. ◎ Protocol of MARS unit 3677: 1. MARS unit 3677 receives message 31. 2. Judge whether it is a self-addressed or forwarded message. Judge based on the match between self-number 3177 and the leading 3657 of 31
It is. "Transfer unit" because it does not match "3677 ≠ 3657"
Transform into. ◎ Protocol of transfer unit 3677: 1. Judgment of selection of transfer transmission path (upper / lower). Since the transfer unit 3677 has the layer number 3, the leading 365 of the 31
For 7, check if the third digit or more matches. Selects the lower transmission line because it matches "36xx = 36xx"
It is. 2. When the lower transmission path is selected;
Fixed. Notice the second digit that is one less than the own layer number 3.31
Of the second digit of the first 3657 of the
Send it out. That is, from lower transmission path 3607 to unit number 3657
Send. ◎ Protocol of MARS unit 3657: 1. MARS unit 3657 receives message 31. 2. Judge whether it is a self-addressed or forwarded message. Judge by matching self-number 3657 with the first 3657 of 31
It is. It matched with “3657 = 3657”. Since there is a sign #, it is transformed into a "delivery unit". ◎ Protocol of delivery unit 3657: Interrupts CPU 3657 connected to MARS unit 3657
To inform or judge by polling
Flag. Message with frame format as shown in Fig. 3
Indicates that the source and destination MARS units exist on the LAN
If there is, the message is always transmitted through the transmission path.
Sent. If the inside of MARS unit 11 turns into a transfer unit
If there is a memory and a microprocessor,
Has no effect on the computer memory 15 and CPU 16
You can forward the message to. Therefore, it should be interpreted as [Definition of terms used in this specification].
Then, the computer is connected to the MARS unit for transfer.
You don't have to. 4 and 5, the transmission / transfer in the WAN of the present invention will be described.
The outline of the transmission / reception protocol is described in a concrete example, and
Help draw the concept of a network. Fig. 4 shows an example of the whole view of the network of the present invention.
LAN with many S units connected and WA with multiple LANs connected
Here is an example of N. Figure 5 shows the transfer message in the network.
3 is a specific example of a frame format conforming to 31 of 31.
It is. The network of FIG. 4 will be described. * Figures 41, 42, and 43 surrounded by fine broken lines are independent figures.
LAN similar to 2. * The transmission line for LAN is indicated by a solid line with a width of 0.4 mm, and the transmission line for WAN is
It is indicated by a dashed line with a width of 0.4 mm. * All LANs are WAN transmissions connected to the upper transmission path terminal.
There is only one route, which is the maximum unit number in the LAN. However, there are many WAN transmission lines connected to the lower transmission line terminals.
There are numbers. * Within one LAN, transmission lines and units with the same number
not exist. However, the same number may exist in another LAN.
Absent. For example, 2777 for LAN41 and 42, 3427 for LAN41 and 43, etc.
It is. * There is no WAN transmission path for units 1244 and 3424 on LAN41.
Was interpreted as not having been laid as unnecessary.
It is. Similarly, units 1267 and 3277 in LAN41 are used for LAN transmission.
It is interpreted that the road was not laid as unnecessary. * Lower-level transmission, such as units 2537 and 2477 on LAN42
Laying a WAN transmission line instead of a LAN transmission line at the transmission line terminal
You can. * Unit 3777 in LAN43 is a LAN
A transmission line for WAN was laid without a transmission line for transmission. Yo
In LAN43, 3777 is the maximum unit number. Similarly, in LAN42, 2777 is the maximum unit number. * MARS units connected to LAN transmission lines have been
Were connected in the same order, but LAN43 3377, 3677, 3477, 3177
You do not need to connect them in order. * WAN transmission line is a lower transmission line of one MARS unit.
Connect the terminal to the upper transmission line terminal of another MARS unit
This is the usual laying situation. For example, 42-2537 and 41-7777, 42-2546 and 43-3777,
is there. * As a special connection of WAN transmission line, lower transmission line terminal
Hot-Lin can be connected to each other.
Call it e). For example, 41-3423 and 43-3644, 43-3642 and 43-3426,
is there. * Both LAN transmission lines and normal WAN transmission lines have special connections.
Hotlines are all installed using the same transmission line.
It can be set. * The difference between the LAN transmission path and the WAN transmission path is the transmission path number.
And whether there is a '0' in the MARS unit
This is the difference between the initial settings. Figure 5 is a specific example of a message passing through the WAN transmission path.
is there. * Fig. 5 is a specific example consisting of three types.
It is. Message 51 for LAN41, Message 52 for LAN42, LAN4
3 uses message 53. These messages
Is connected to the WAN transmission line at the beginning of the message in Fig. 3.
MARS unit number is added. * Regardless of Fig. 3 and Fig. 5, when transferring within LAN,
The number written on the head address and the next frame
Judgment is made based on whether there is a mark on the In the network shown in FIG.
Transmission starts and a message is sent to the computer 3426 in LAN43.
Of the protocol of the present invention.
I do. ◎ LAN12 computer 1247 and unit 1247 prototype
Col: 1.41 CPU1247, MARS message via CPU bus
Send to the unit 1247, and the unit receives the message.
(Transfer in Fig. 1) Note: The message to be sent is 41 in Fig. 5. Message creation is the transfer route From, only the exit, destination, and sender are listed in order, and sent
Add a mark # to the original. 2. MARS unit 1247 receives the message and is described
Judge whether the address is correct or not. 3. If there is no error, it transforms itself into a “transfer unit”.
However, if there is an error, it is returned to CPU16. ◎ Protocol from transfer unit 1247 to 7777 of LAN41
Le: 1247 → (1207) → 1277 → (1077) → 1777 → (0777) →
Transfer with 7777 is performed in a chain. (Prototype using Fig. 2
◎ Protocol of MARS unit 7777 of LAN41: 1. Judge whether it is a self-addressed or forwarded message. MARS unit 7777 matches the start address 7777 of 51
Judge whether it is. (The protocol using Figure 2 and
Same) “7777 = 7777”. In addition, mark on the next frame
Whether it is a self-addressed or forwarded message depending on whether there is a message
do. 2. Is there a mark # next to the first 7777 in message 51?
Check if there is any. Since it is “7777 2546” and there is no mark,
Judge as a transfer message and send it to "WAN transmission unit"
Transform and send to WAN transmission path. ◎ Protocol of MARS unit 2537 on LAN 42: 1. Message 51 was changed to 52 because it was received from WAN transmission path.
Replace it. The change method is to replace the first 7777 of the message with the own number 2537.
Yes. Rotate to the position before # to change the address description position
I do. Add a mark # to the last rotated 2537. Change to message 52 and enter the sign #
So I turned myself into a "transfer unit"
It is. ◎ Protocols from LAN42 transfer units 2537 to 2546
Le: 2537 → (2507) → 2547 → (2540) → 2546 Transfer is performed in a chain. (Protocol using Fig. 2
◎ Protocol of MARS unit 2546 of LAN42: 1. Judge whether it is a self-addressed or transfer message. MARS unit 2546 matches the start address 2546 of 52
Judge whether it is. (The protocol using Figure 2 and
Same) “2546 = 2546”. In addition, mark on the next frame
Depending on the presence or absence of #, whether it is addressed to itself or a forwarded message is judged again
Make a cut. 2. In message 52, there is a mark next to the first 2546
Find out how. "2546 3246" and no message
And transforms itself into a “WAN transmission unit” and transmits it over the WAN.
To the road. ◎ Protocol of MARS unit 3777 of LAN43: 1. Since message 52 was received from WAN transmission line, message 52 was changed to 53.
Change. The change method is to replace the first 2546 of the message with the own number 3777.
Yes. Rotate to the position before # to change the address description position
I do. Add a # to the last rotated 3777. Change to message 53 and enter the sign #
So I turned myself into a "transfer unit"
It is. ◎ Protocol for LAN43 transfer units 3777 to 3426
Le: 3777 → (3077) → 3477 → (3407) → 3427 → (3420) →
3426 and transfer are performed in a chain. (Prototype using Fig. 2
◎ Protocol of MARS unit 3426 of LAN43: 1. Judge whether it is a self-addressed or forwarded message. MARS unit 3426 matches 53 head address 3426
Judge whether it is. (The protocol using Figure 2 and
Same) “3426 = 3426”. In addition, mark on the next frame
Whether it is a self-addressed or forwarded message depending on whether there is a message
do. 2. In message 53, is there a mark # next to the top 3426?
Check if there is any. It is “3426 # 3777” and there is a mark.
Judge and transform into a "delivery unit". ◎ Protocol of delivery unit 3426 of LAN43: Interrupts CPU3426 connected to MARS unit 3426
To inform or judge by polling
Flag. As in message 51, the LAN exit and destination address are recorded.
If it is listed and there is a destination unit in FIG. 4, send
From the original computer to a computer on another LAN
, The message is forwarded through the described route
Is performed. Of course, send a message via the hotline
It is also possible. CPU3426 of LAN43 received message 53. Also
If you want to reply to CPU1247 on LAN41,
Just follow the reverse route. The reverse route to reply is And from the viewpoint of CPU3426 of LAN43, message 53
Rearrange and create messages as exit 3777, exit 2537, destination 1247, source 3426, and MARS unit 3426 of LAN43.
If you send it through the CPU bus, the reverse route sent
Follow. Figure 4 shows that there is no destination address or LAN or WAN transmission path
In the event of accidents such as disconnection, the message
Is not transferred. However, changes in messages 51, 52, 53
, It is possible to return from anywhere on the transfer route. ★ For example, a disconnection accident on the transmission line from 2547 to 2546 of LAN42
42; 2547 adds 2547 to message 52
"Exit 2537, destination 1247, # 2547, exit 2546, destination 342
6 ", and the MARS unit of LAN42 changes to a" transfer unit ".
If you do, you will be sent back correctly. ★ For example, the same applies to LAN43 from 3647 to 3644.
And re-arrange the message 53 as follows. “Exit 3777, Exit 2546, Destination 1247, # 3644, Destination 342
6 "★ For example, to the CPU bus terminal of the destination MARS unit 3426
If the computer is not connected, # 3644 above
To # 3426, "Exit 3777, Exit 2546, Destination 1247, # 3246, Destination 342
It can be returned as 6 ". The returned CPU1247 of LAN41 has the same disconnection judgment and destination.
3426, the computer is connected to the MARS unit 3426 of LAN43.
Data is not connected. Figure 6 shows an example of a theoretical LAN transmission line. Using FIG.
Then, a protocol that avoids message contention is explained.
In addition, between "WAN transmission unit" and "WAN reception unit"
Is point-to-point, and a dedicated transmission line is laid
Message conflict does not occur. Avoid message contention on LAN transmission path of the present invention
The solution was to use a dedicated transmission path. this
The protocol is briefly listed. 1. Determine whether the transmission line is being used. (RTE = read tender and RVC = read victor) 2. If not used, participate in the bid for securing the transmission path
It is. (WTE = write tender and Bus = messenger bus) 3. If you win the bid and secure the transmission path, declare victory and send
Become a unit. (WVC = write victory) 4. The winning sending unit sets the receiving unit address.
Broadcasting to other MARS units
It is. (WM = write message-address and WTG = write trigg
er) 5. Determine whether other MARS units should be transformed into receiving units.
Refuse. (RM = read message-address and RTG = read trigge
r) 6. The sending unit interacts with the receiving unit to determine if it exists.
It is. (Bus = Write and read messenger address-bus) 7.
Transfer the page. (WM = write message & data and RM = read message
& Data) 8. The transmission unit opens all transmission paths. (Null = low level) 9. The receiving unit is forcibly stopped receiving. (Null−RVC = read null−victory) FIG. 6 will be described, and specific description will be given along 1 to 9 above.
It is. FIG. 6 shows 2777,2377,2577,2177,2 in LAN42 of FIG.
677,2277,2477 connected transmission line 2077 as an example
Was. * 2677 [6 = 110] is given as a representative and indicates its meaning. 2677; Transmission line terminal of MARS unit number 2677 of LAN42. However, 2077 with '0' indicates transmission line 2077.
Yes. [6; One digit of '0' connected to transmission line 2077 replaced with '6'
number. Conversely, connecting to transmission line 2077 by describing [6]
Can be identified. = 110]; Indicates the bit state of '6' expressed in binary. In addition, it is used in LAN42 that the binary number is 3 digits.
The maximum number is [7 = 111] and can be represented by 3 digits in binary
Because. * 2777 is the lower transmission line terminal, 2177,2277,2377,2477,257
7,2677 are connected by the upper transmission line terminal,
Connected with a barter. (Refer to 13 and 14 in Fig. 9.)
Unit to the transmission line with an open collector inverter
Through the transmission terminal and the inverter from the transmission line 2077
It is divided into connected receiving terminals. * Transmission terminals WM2, WM1, WM0, WTG,
From WVC and WTE, pass each open collector inverter
And send signals to the transmission lines M2, M1, M0, TG, VC, and TE. * Signals appearing on transmission lines M2, M1, M0, TG, VC, TE
Receiving terminals RM2, RM2,
Obtained from RM1, RM0, RTG, RVC, RTE. * Relationship between one transmission line and the terminals of many MARS units
Is the structure of wired logic. * Widely used as logic synthesis in wired logic
Is connected by an AND tie, but one gate
Since two sets of inverters were used as elements, the MARS unit
From the point of view, it is a wired-OR connection. * M2, M1, M0, M2, M1, M0,
Laid three wires connected to .M2 to M0
And write it as Bus. * The three message buses are used for LAN42
The maximum number is '7', and when '7' is expressed in binary, it becomes 111.
Because three digits are required. Generally, in order to express the maximum number as the number of bus wires,
Lay more than the required number of bits. * WTG, RTG, and TG are connected via a computer data bus.
Signal trigger (trigger) to read data
It has the same meaning. However, in the present invention, when reading the Bus signal,
RTG is not always used. * WTE, RTE and TE use transmission line 2077 as two MARS units.
Used to indicate the status of bidding between dedicated transmission lines
Is performed. * WVC, RVC, and VC are the states when the winner is shown by bidding
Used to indicate status. MARS unit 2377 replaces MARS unit 2477, Fig. 3 or Fig. 5
When transmitting the message of
2677 gives an example of requesting transmission to 2277.
This will be described specifically. 1-1; Status of first transmission line: Bus = data RTG = H / L RVC = H
RTE = H [3] is transmitting to [4], so [3]
WTE and WVC are kept at H level (high level). Since the transmission path is wired-OR, all units are
Can recognize the RVC and RTE signals as H level. [5] and [6] wait until RTE and RVC go low.
ing. 1-2; End of transmission: Bus = 000 RTG = L RVC = H
→ L RTE = H → L As [3] has finished sending the message to [4],
Set own WTE and WVC to L level (low level)
Release all. 2-1: [5] participates in bidding: Bus = 101 RTG = L RVC = L
RTE = L → H [5] waits for RTE = L to participate in bidding
And immediately participate in the bidding as WTE ← H, Bus ← 101.
Add. 2-2; [6] participates in bidding: Bus = 110 RTG = L RVC = L
RTE = L ← H [6] waits for RTE = L to participate in the bidding
And immediately participate in the bidding as WTE ← H, Bus ← 110.
Add. There is a distance between [5] and [6], and the signal propagates
It takes time. 2-3: Signal propagates to transmission line: Bus = 111 RTG = L RVC = L
RTE = H Since the transmission path is wired OR, [5 = 101]
Since both of [6 = 110] flow to Bus, they are logically synthesized,
Bus = 111. 3-1; Victory judgment of [5]: Bus = 111 RTG = L RVC = L
RTE = H Check if [5 = 101] is the same as Bus = 111. Since it is not the same, if the 0th bit is examined, it is the same as [5 = xx1] = [Bus = xx1].
Yes. 3-2; Judgment of [6]: Bus = 111 RTG = L RVC = L
RTE = H Check if [6 = 110] is the same as Bus = 111. Since it is not the same, when examining the 0th bit, it is different from [6 = xx0] ＝ [Bus = xx1].
get off. WTE ← H, Bus ← 000 to secure the right to bid. 3-3; Signal propagates to transmission line: Bus = 101 RTG = L RVC = L
RTE = H [6 = 110] descends on bid, [5 = 101] remains
Bus = 101. [3 = 011] also participated in the bidding
Bus = 111, but the first and second bits
, And so on, for each transmission line propagation time,
Only one unit wins the game. 3-4: Victory declaration of [5]: Bus = 101 RTG = L RVC = L
RTE = H Check if [5 = 101] is the same as Bus = 101. Same, so WVC to declare victory to other units
← H. 4-1; Know the declaration of victory: Bus = 101 RTG = L RVC = L
→ H RTE = H For all units, Bus = 101 to [5] are winners
I know. [5] is the sender who won the bid, so do nothing
Wait for a while. [6] stepped down from bidding himself. If [1 = 001]
If was participating in the bidding, go down because RVC = H
It is made to be done. 4-2; Request transmission: Bus = 101 RTG = L RVC = H
RTE = H [5] requests transmission at [2 = 010].
Assuming that s ← 010, the trigger signal WTG ← H. 5-1; Judge whether it is addressed to yourself: Bus = 010 RTG = L → H RVC
= H RTE = H For units other than the source [5], the RTG after RVC = H =
Wait for L → H only once and read Bus. 5-2; Judgment of [2]: Bus = 010 RTG = L → H RVC
= H RTE = H [2 = 010] reads Bus = 010 and checks whether it is addressed to itself.
to decide. Since it is addressed to myself, give Bus ← 010 and my number and reply
do. 5-3; Other judgments: Bus = 010 RTG = L → H RVC
= H RTE = H Other than [2 = 010], read Bus = 010 and check if it is addressed to itself
Judge. Since it is not addressed to yourself, what happens after that until RVC becomes L level
Not sure. 6-1: Dialogue of [5]: Bus = 010 RTG = H RVC = H
RTE = H [5] is Bus ← 000 WTG ← L. 6-2; Judgment of [5]: Bus = ??? RTG = H → L RVC
= H RTE = H [5] reads Bus. I wanted the Bus to send from [5] to [2 = 010]
And [2 = 010] are synthesized by wired OR
Yes. Therefore, [5] is, despite Bus ← 000,
If Bus = 010, the destination unit [2 = 010] exists, and
If s = 000, it knows that the destination unit does not exist. 6-3; Dialog of [2]: Bus = 010 RTG = H → L RVC
= H RTE = H [2] considers its own memory margin after RTG = H → L
Then, release Bus as Bus ← 000 and change to “Receiver unit”.
I will. 7-1; Start transmission of [5]: Bus = 000 RTG = L RVC = H
RTE = H [5] detects that Bus = 000 is released and [2]
Knows that has entered the reception system, the message of Figure 3 and Figure 5
To the bus on the transmission line and use WTG as the trigger pulse
To start sending. 7-2; Start receiving [2]: Bus = data RTG = H / L RVC
= H RTE = H [2] is a “receiving unit”
Receive the flowing message in synchronization with RTG. 8-1; Transmission end of [5]: Bus = 000 RTG = L RVC = H
RTE = H [5] indicates that the transmission path is
Release all. Bus ← 000, WTG ← L, WVC ← L, WTE ← L. 8-2; Propagation of end: Bus = 000 RTG = L RVC = H
→ L RTE = H [6 = 110] went down by bid, but WTE ← H
is there. Transmission line is completely open with RTE = H due to wired OR
is not. 9-1; Forced termination of [2] reception: Bus = 000 RTG = L RVC = H
→ L RTE = H [2] detects RVC = H → L and “Receive unit”
It is released from. 9-2; [6] participates in rebiding: Bus = 000 RTG = L RVC = H
→ L RTE = H [6 = 110] went down for the first bidding,
He was waiting for the opportunity for a re-bidding while staying at H. The re-bidding is the same as the first bidding “2-2;”.
And participate in the bidding. If [1 = 001] participated in the first bidding
Since they were waiting for the opportunity to bid again with WTE ← H,
Participate in the second bidding. 9-3; Bid other than [6]: Bus = 000 RTG = L RVC = H
→ Except for L RTE = H [6], participate in the re-bid by RTE = H
Can not. The transmission of the message is completed in the order of 2377 → 2477, 2577 → 2277, 2677 → 2677 by the above protocol for the dedicated transmission path. Dedicated transmission of the present invention
State the features of the protocol to be used for
Ends. ☆ The priorities are determined uniquely, and when arranged from the highest,
[7], [3], [5], [1], [6], [2],
[4], in the order described in FIG. ☆ The number of memories belonging to the layer
It increases exponentially. ☆ Therefore, all connected to the network of the present invention
The computer wants to transfer at the same time and the bidding is started
Also, transfer from the upper transmission path to the lower transmission path has priority.
Thus, the network does not harden.

DISCLOSURE OF THE INVENTION

In the following order, a method of constructing a "network system for computers with transfer identification and arbitration mechanism" will be disclosed.す る Disclose the rules for creating MARS unit numbers and transmission line numbers.図 Figures 7 and 8 are described as an example of the frame format, and Figure 9 is described as an example of the internal structure of the MARS unit.開 示 Disclose the MARS unit protocol. The second disclosed FIGS. 7 to 9 should be described in [Embodiments of the invention], but they will be described in [Disclosure of the invention] in order to clarify the protocol of the present invention. ◆ Disclose the rules for creating the network of the present invention. [Table 1] shows a method of constructing the network of the present invention of 41, 42, and 43 of FIGS. 2 and 4 using the MARS unit of FIG.

[Table 1] Network creation method. Write appropriate numbers for 'N1' and 'N2' in [Table 1]
The LAN of the present invention is completed by writing the numbers in the conventional wisdom. All devices and addresses that correspond one-to-one with the units created in [Table 1] are given the same numbers as the units. [Table 1] shows the numerical relationship, so substitution by the symmetry of numbers is allowed. For example, after completing the network, swap 'N' and '1'. Each item in [Table 1] will be described. 1.N1 represents the minimum number of buses in FIG.
And N1 = 3. 'N' is a number based on social wisdom. If N1 = 3,
'7'. If N1 = 4, it is '15' or 'F'. 'F' is
A hexadecimal representation commonly used in the computer world. When N1 = 5, it is '31' or '1F' or '11111'. '11111' is a binary representation commonly used in the computer world. 2. N2 defines the number of digits of the transmission path and MARS unit number. The necessary frames of the addresses described in FIGS. 3 and 5 are defined. Since N2 = 4, a 4-digit number was written in Fig. 3 and Fig. 5. If N2 = 5, a 5-digit number is written in FIGS. In addition, the mark # described in FIG. 3 or FIG.
The number of bits required to represent the address excluding is calculated. 3-6. A recursive formula widely used in the field of mathematics.
Described how to construct a network. Each item is
This was described in [Summary of the Invention]. 7-8. This is a necessary proviso when actually constructing the network of the present invention shown in FIG. 2 or FIG.
created. The layer number used in Fig. 2 can be calculated as follows if the network is created according to the rules in [Table 1].層 The layer number to which the own unit belongs is the number obtained by adding 1 to the number of 'N' of the unit number used for the corresponding unit. Transmission line connections other than those in [Table 1] are possible, and will be described as exceptions. As shown in 4. of [Table 1], "The LAN transmission line with the number where one of 'N' in the upper digit of the unit number is set to '0' is installed as the upper transmission line of the unit having the largest number. Can be done. " In this exception, [Table 1]
More units can be connected than the network created in. However, LAN transmission lines cannot be laid, but LANs can be connected by hotline connection. The hot-line connection increases the number of transfer paths between LANs. The exceptions are applied to each LAN in Fig. 2 and Fig. 4, and the connection diagram around the unit with the maximum number is shown in Fig. 5-A,
Fig. 5-B, Fig. 5-C, and Fig. 5-D illustrate the unit number, layer number, and transmission line number in each figure. * Fig. 5-A shows the connection near 7777 in Fig. 2. 7777 is the layer number 4. * Figure 5-B shows the connection near 7777 of LAN41, and 7777 is layer number 4. * Figure 5-C shows the connection near LAN 7777 of LAN42. * Figure 5-D shows the connection near 7777 of LAN43, and 3777 is the layer number 3. The layer number using the exception rule can be calculated as follows.層 The layer number to which the own unit belongs is the same number as the number of digits in which the upper transmission line number “0” of the corresponding unit exists, and 1 is added to the number of digits in which the lower transmission line number “0” exists. Is the number added. [Table 2] shows the maximum number of connectable units when the network of the present invention is constructed by the method of [Table 1].
(6), (7), ... (30), entered in the column of the number of units
(31) is the maximum number of units that can be connected to one LAN transmission line. “N1” and “N2” are the minimum number of buses and the number of digits of the unit address. “N1 × N2” is the number of binary digits required to represent one address. “The number of bits required for transfer” is the maximum value of the number of bits required to create the message shown in FIG. 3 or FIG. The calculation was performed excluding the number of bits used for the mark #. LAN and WAN1 are the LAN in Fig. 3 and the network in Fig. 5.
WAN2, WAN3, and WAN4 are the transmission paths for WAN at the bottom of FIG.
This shows a network in which LANs are connected sequentially. In the following [Disclosure of the Invention], N1 = 4 and N2 = 3 will be mainly explained.

[Table 2] Maximum number of units that can be connected to the network. ◆ Specific examples will be described first to clarify the protocol of the present invention. FIGS. 3 and 5 show frame formats used in the network of the present invention. Many address groups as shown in FIG. 5 cannot be correctly transferred on the network of the present invention unless the "address message" and the "data field length" can be clearly separated. Three methods are shown as examples of clear separation. (1) Before describing the addresses in FIGS. 3 and 5, write the total number of addresses and the starting position of the token. This method belongs to the same category as writing "data field length" before the data part. (2) An end-mark is written after the address message shown in FIG. 3 or FIG. This method uses the end code (end-mar
k) and the bit state must be distinctly different from the address message. (3) Use more bits than necessary to represent one address, and also include the token used in FIGS. 3 and 5. Of the above three methods, (3) was adopted and [Table 2]
An example of the case where N1 = 4 and N2 = 3 is shown in [Table 3]. As shown in [Table 3], the end bit (end-mark)% other than the mark # and the non-passing route mark at the time of return are indicated by the token bit.
Can also be used.

[Table 3] Example of frame format of N1 = 4, N2 = 3, all bits = 16 FIG. 7 shows a case where the bits related to the address are determined as shown in [Table 3]. There is no essential difference between the frame of FIG. 7 and the frame of FIG. FIG. 7 shows a frame format during transfer of a route connecting six LANs, and is an example of the case of WAN4 in [Table 2]. The route is shown below, and is an example in the middle of [LAN3 entrance → LAN3 exit]. [1 source → 1 exit] ‥ → ‥ [2 entrance → exit] ‥ → ‥
[3 entrances → 3 exits] ‥ → ‥ [4 entrances → 4 exits] ‥ → ‥
[5 entrances → 5 exits] ‥ → ‥ [6 entrances → 6 destinations]

[Table 4] Method of error determination of address message Fig. 8 shows that any unit in LAN3 detects a failure,
Rearrange the address messages in FIG.
Here is an example of changing the end-mark). MARS that any unit in LAN3 relays and transmission lines
If a unit failure is found, [1 source ← 1 exit] ‥ ← [2 entrance ← 2 exit] ‥ ←
‥ [3 entrance ← 3 return-] will be returned. Therefore, 81 to 83 are determined so that FIG. 8 can be viewed simply as an address message in which the transfer destination is entered. 84 is the number of the MARS unit that has found a failure.
85-87 are 71-73 of the planned entry
.88, which is the entrance address of the LAN, is the destination address in FIG. The first frame of the address message described in [Table 4] to [Table 6] is 71 in FIG. 7 or 81 in FIG. 8, the second frame is 72 in FIG. 7 or 82,. It is 77 in 7 or 88 in FIG. The protocols described in [Table 5] and [Table 6] conformed to the widely used Neumann-type program description method. That is, i and j are variables. [Table 5] is an address change program when a message is received from the WAN transmission line. [Summary of the Invention]
Rotated as 51, 52, 53 in Fig. 5. [Table 6] shows an address change program for returning a message. This program uses the LAN shown in Fig. 2.
It can be applied to the case of returning only within.

[Table 5] Address change program when message is received from WAN transmission path

[Table 6] Address change program when returning a message Fig. 9 shows an example of the internal structure of the MARS unit. FIG.
Is explained. * 12, 13, and 14 in FIG. 9 are the same as 12, 13, and 14 in FIG. 1, and are the CPU bus connection terminal 12, the upper transmission line terminal 13, and the lower transmission line terminal 14, respectively. * The solid line with a width of 0.4 mm is called a data bus in general computers. * The dashed line with a width of 0.4 mm is called an address bus in general computers. * Control wires are indicated by broken lines with a width of 0.2 mm. * There are three memories 20, which are ordinary memories used for general computers. * The selectors 21, 22, and 23 are the memory 20 and the CPU 1 shown in FIG.
A device that connects six buses or three internal buses. * In the memory 20, write and read accesses from the four data buses are time-shared (time sharing system). * In a strict sense, the simultaneousness of memory access is three accesses of the number of memories. * The names and main responsibilities of the processor that controls the MARS unit are as follows. SPU06 (select processing unit); memory management MPU26 (main processing unit); message content processing UPU36 (upper side processing unit); communication on the upper transmission line LPU46 (lower side processing unit); communication on the lower transmission line * Microprogram 07 , 27,37,47 are places to write the control protocol programs to be given to the respective processing units 06,26,36,46. * Registers 28, 38, and 48 are places for temporarily storing data such as initial settings. * ALUs (arithmetic and logic unit) 29, 39, and 49 are responsible for executing calculations. * Bus converters 34, 35, which convert four digital signal groups to connect external connection terminals 13, 14 to the internal bus
44 and 45 are provided.

[Table 7] Memory switching status table in SPU06 The functions of the SPU06 communicating with the other processor units 16, 26, 36, and 46 to operate the selectors 21, 22, and 23 are shown in Tables 7 to
This is explained together in [10].

[Table 8] Protocol for SPU06 to generate ACK by write request

[Table 9] Protocol for SPU06 to generate ACK by read request The relationship between the memory 20 and the CPU16, MPU26, UPU36, LPU46 is the same as the concept of the conventional common memory type multiprocessor, but the construction method is different. The MARS unit is a construction method that transfers messages between the terminals 12, 13, 14 and the memory 20. The message is transferred from the computer to the memory 20 through the terminal 12 and the terminal 13
The message is transferred to the memory 20 of another MARS unit through or 14, and the message is transferred from the memory 20 to the computer through the terminal 12. [Table 7] shows the contents of a register indicating which PU the memory 20 is under control of during any given period. The hardware is designed to switch the selector according to the contents of the registers in [Table 7]. SPU06 is set when the other PU (16, 26, 36, 46) wants to write to any of the three memories 20 or when the other PU wants to read from memory 20 (Table 8). The register is switched according to [Table 9], and an ACK (acknowledge) signal is output. SPU06, when prompted to release memory 20, [Table 10]
Disconnect by.

[Table 10] Protocol when SPU06 is prompted to release memory ◆ Disclose the message transfer protocol in the MARS unit. The following protocol is described based on the description of the data flow system. * Protocols required for the present invention are [Table 11] to [Table 26].
This will be explained. * Words and idioms after the '§' symbol indicate related protocols. * The instruction cell of the data flow system starts processing according to the firing signal, and the result is passed to the next instruction cell. * The operation of the instruction cell is described as “operation” in each table. * Signals generated by the operation of the instruction cell are described as “firing”, “seeding”, and “ACK (acknowledge)” in the frames of each table. Note that “digestion” is the logical inversion of “seed fire”. * Combine “Seed Fire”, “Digestion” and “ACK” to create the next firing signal. * For example, 1 to 4 in [Table 11] and 4 to 6 in [Table 14] are C
This is the processing of PU16, and the "operations" are listed in order.
This represents sequential processing equivalent to the processing order of a Neumann computer. * For example, in [Table 11], it is divided into three groups, 1 to 4, 5 to 7, 8, because the processors are different. Thus, the instruction cells of different processors are stored in one [Table].
When describing in, the main processor was clarified. * For example, 1 to 5 in [Table 13] are instruction cells of the MPU26 processor. It is divided into four groups, 1,2,3,4-5 by ruled lines. The four groups show multiprocessors that can execute instruction cells independently of each other. However, instruction cells 4 and 5 in one group are sequential processes that execute 4 and then execute 5. * This is a hardware configuration in which the firing signals of all instruction cells in the MARS unit are set to 0 = L level when the power is turned on. Therefore, all instruction cells cannot start the operation by themselves. * In addition, before and after the power is turned on, WM2, WM1,.
VC and WTE are hardware configurations that are all at L level.

[Table 11] Protocol for receiving messages from CPU16 The MARS unit is prompted to transfer the message from the CPU 16 in [Table 11] to the memory 20, and the operation of the instruction cell is started.

[Table 12] Initial setting protocol of MARS unit The message that the CPU 16 first transfers to the memory 20 is MA
Initial setting data of the RS unit, which is transferred to each processor according to the protocol shown in [Table 12]. The terminals 12, 13, 14 to be transferred are determined by the protocol shown in [Table 13]. Note that WAN

[Table 13] Protocol for MPU 26 to determine transmission path selection Except for the reception from the transmission line, the MPU 26 reads out the data in a time sharing manner and executes it according to the protocol shown in [Table 13], even during the transfer from the terminals 12, 13, 14 to the memory 20. If the terminal to be transferred is 12, transfer it to the computer using the protocol in [Table 14].

[Table 14] Protocol for passing messages to CPU 16 Send. The program for increasing the value of the address bus in [Table 11-7] is also executed from [Table 14]. If the terminal to be transferred is 13 and it is a LAN transmission path, use the protocol in [Table 15]. If the terminal to be transferred is 14 and it is a LAN transmission path, use the protocol in [Table 16].

[Table 15] Protocol for UPU36 to transfer messages to LAN transmission path If a LAN transmission line is connected to the terminal to be transferred, one LAN
A bid is made between MARS units connected to the transmission line.
The protocol for participating in a bid and declaring a victory is [Table 17].
. Even if you participate in a bid and secure a transmission path, you cannot transfer without a MARS unit at the transfer destination. Destination
The protocol for confirming the existence of the MARS unit is shown in [Table 18].
. If [Table 17] and [Table 18] are part of [Table 15], UPU
36 programs

[Table 16] Protocol for LPU 46 to transfer messages to LAN transmission path

[Table 17] Protocol to participate in bid for securing LAN transmission path and confirm victory If it is a part of [Table 16], it is an LPU46 program.

[Table 18] Protocol for confirming existence of destination unit If the transmission line failed or the destination unit did not exist, it was shown in [Summary of the Invention] that the message could be returned to the source MARS unit. The failure of the LAN transmission line is determined by [Table 17-3_7], and the failure or unconnection of the MARS unit connected to the LAN transmission line is determined by [Table 18-1_9]. Also in the WAN transmission line, [Table 23-3_1] judges that the transmission line is faulty. Regardless of the failure, the protocol that returns the message is [Table 19]. After changing the frame format address of the message to be returned in [Table 19-3_1], it is assumed that the message has been received, and the processing for determining the transfer destination in [Table 13-4] is executed. The protocol for reception from the LAN transmission path is [Table 20]. If the terminal of the WAN transmission line that is not a hotline is the upper transmission line terminal 13, the transmission is performed according to the protocol of [Table 21] and the reception is performed according to the protocol of [Table 24]. If the terminal of the WAN transmission line that is not a hotline is the lower transmission line terminal 14, the transmission is performed according to the protocol of [Table 22] and the reception is performed according to the protocol of [Table 25].

Table 19: Protocol for MPU 26 to change address message for return

[Table 20] Protocol received by broadcasting from LAN transmission path

[Table 21] Protocol for sending messages to WAN transmission line on upper terminal side

[Table 22] Protocol for sending messages to WAN transmission line on lower terminal side When the MARS unit connected to the LAN transmission line competes for a bid, the highest priority unit is [7 = 111] connected to the lower transmission line terminal 14 as described with reference to FIG. Although it is point-to-point, there is a possibility of message contention if it is laid with the wired OR shown in Fig. 6, which is the same as the LAN transmission line.
Give priority to the MARS unit connected to. In the case of a hotline WAN transmission line, both MARS units use the lower transmission line terminal 13 (see FIG. 4). Therefore, as described in the remarks column of [Table 22] and [Table 25], one MARS unit is regarded as being connected to the upper transmission line terminal 13, and the priority order in the case of contention is mutually determined in advance. Keep it.

[Table 23] Protocol following [Table 21] or [Table 22] The continuation of [Table 21] and [Table 22] is shown in [Table 23]. The continuation of [Table 24] and [Table 25] is shown in [Table 26]. Since the WAN transmission line was the same as the LAN transmission line in Fig. 6, the protocol was explained on the assumption that message contention would occur. However, if one-way WAN transmission lines were laid mutually, No conflicts occur.

[Table 24] Protocol for receiving messages from WAN transmission line on upper terminal side

[Table 25] Protocol for receiving messages from WAN transmission line on lower terminal side

[Table 26] Protocol following [Table 25] or [Table 26] Usually, a transition status table is shown to make the program clearer. However, in the embodiment having four processors according to the present invention, since the description is made using the instruction cell of the data flow system, FIG. 10 shows a transition table which facilitates understanding. In Fig. 10-A, the data flow of the initial setting and [Table] used are described. Figure 10-B shows the flow of the received data and the [Table] used. Fig. 10-C shows the flow of data sent and the table used.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Usually, only one optimal embodiment will be described as an example of the invention. However, as shown in FIG. 9, there were a plurality of processors, and in [Disclosure of the Invention], the protocol was explained based on the description of the data flow system. Although there are many computer architectures such as Neumann computer, data flow computer, fuzzy computer, neural network, etc., it is no exaggeration to say that no computer can execute [disclosure of the invention] as it is. Therefore, as an embodiment of the present invention, an embodiment that realizes the protocol of the present invention in the following cases will be described.例 Example of implementation with one sequential processing type processor and ALU.例 Example of implementation using 3 to 4 sequential processors and ALU. (4) An example in which an instruction cell including all operations is realized by an electronic circuit. ◆ Embodiment in which the protocol of the present invention is realized by one sequential processing type processor. Hardware is as follows. * A three-state buffer gate is provided at the terminal 12 in FIG. 9, and the I / O shown in [Definition of words used in this specification] is included, including bus conversions 34, 35, 44, and 45. * Combine the solid lines with a width of 0.4 mm shown in Fig. 9 in parallel as one data bus. * Combine the 0.4 mm wide parallel dashed lines in Fig. 9 into one address bus that defines the memory 20. * MPU26, UPU36, LPU46, ALU29,39,49 of Fig. 9 is one μP
U (micro processing unit). Memory 20, selectors 21, 22, 23, registers 28, 38, 48, microprograms 27, 3
Collect 7,47 in one μ memory. With the above hardware, the control wires described with a width of 0.2 mm in FIG. 9 consist of μPU, μmemory, I / O,
And terminal 12. In other words, the configuration is the same as that of a microcomputer using the 8080, 6800, and Z80 chips that are widely used as general CPUs (or MPUs). A method for realizing the protocol of the present invention with a microcomputer will be described. * Create an address map that divides all addresses of μ memory into the following five blocks. 1. Block that stores programs. 2. A block that stores many transition flags. 3. Block corresponding to the memory 20 connected to the selector 21 in FIG. 4. Block corresponding to the memory 20 connected to the selector 22 in FIG. 5. Block corresponding to the memory 20 connected to the selector 23 in FIG. * In the protocol of the present invention, "ignition", "open flame", "ACK" ...
Are all treated as transition flags, and are described in the above “Block that stores many transition flags”. * Divide the program into one main loop program and many subroutine programs. * Complete the "operation" of the instruction cell of the protocol of the present invention in one subroutine. Also, in the program of one subroutine, change the contents of the above flag group corresponding to "ignition", "seeding", etc. given to the next instruction cell. * In the end, subroutines can be created for a number of transition flags and a series of "actions". If there is one "fire", the transition flag and the subroutine correspond one to one. However,
Since there are cases where multiple ignition signals such as "seed fire" and "digestion" occur, the number of subroutines is smaller than the number of transition flags. * One main routine reads all the transition flags sequentially and determines whether or not it should be a firing signal. If it is an ignition signal, call a subroutine. After reading all the transition flags, reread the transition flags from the beginning. * "Τ" set in [Table 12] is very short, so it may be processed by the number of system clocks. (Table 12)
The "τ1" set in must be the exact time unified in one transmission line, but if processed with the number of system clocks, it cannot be received from other transmission lines. Therefore, the following method is used. * The timer interrupts every certain time earlier than “τ1”. The interrupt routine described below is created without reading the transition flag that creates the firing signal for the portion waiting for “τ1” time in the main routine. ☆ Read the transition flag and read the next transition flag if it is not the firing signal. After reading the last transition flag, exit from the interrupt routine. ☆ If it is an ignition signal, increase the number of passages by 1
If the time has not reached "1", the next transition flag is read. ☆ When the time reaches “τ1”, set the number of passes to '0' and call the corresponding subroutine. By doing the above, one sequential processor and AL
This can be achieved with U. The advantages and disadvantages of this case are described. Advantages: Since the hardware design of the MARS unit is simple, the cost is low. Disadvantages: Information transmission speed is slow because “τ1” cannot be shortened. If there is no problem if the transmission speed of information is further reduced, it is also possible to perform all the above processing by interruption for a fixed time. In this case, according to the protocol of the present invention, the terminals 13 and 14 of FIG. 1 become I / O, the memory 15 becomes μ memory, the CPU 16 becomes μPC, and the MARS unit 11
Only data buses, address buses, and wires that control I / O. The methods described below increase the information transmission speed but increase the cost. ◆ An embodiment implemented with 3 to 4 sequential processors and ALU. When the hardware is configured as shown in FIG. 9, there are four sequential processors. SPU06 is a processor that is in charge of memory management.
6 is a processor that is responsible for processing the contents of the message. Even if SPU06 and MPU26 are used as one processor, parallel processing of messages is possible, and the total number of processors becomes three. The method of porting the protocol of the present invention to a program is as follows.
This is the same as “◇ Embodiment for realizing the protocol of the present invention with one sequential processing type processor”. One main and many subroutine programs are described in 27, 37, 47, and the internal transition flags are 28, 37. Store in 38,48 registers or separate memory. ◆ An embodiment in which the instruction cell, including all operations, is realized by an electronic circuit. The first patent to demonstrate the concept of a data flow processor is PCT / US79 / 00453 [a processor for dynamic programming]. This basic data flow processor is a concept that operates an ALU installed outside the processor according to transition data information. PCT / JP94 / 00595 [Gates connecting digital logic circuits], [Industrial applicability and advantages], describes how to create a data flow computer. Figure 11 is a partial modification of [Figure 14] of PCT / JP94 / 00595. The method of realizing the protocol of the present invention by an electronic circuit is described. * 94 in Fig. 11 is a "junction gate", which is disclosed in the above-mentioned patent. * As a specific example, an embodiment that realizes the protocol shown in [Table 20-4] is given. 1. Create [Table 8-3] or [Table 8-4] of SPU06 with the sequential processing circuit of 91. 2. Create the “operations” in [Table 20-3_1] to [Table 20-3-5] using 92 sequential processing circuits. 3. Create “operations” in [Table 20-4_1] using 93 sequential processing circuits. 4. The pulse signal of "ACK = seed fire = [Table 20-3]" from 91 is input to the junction gate 94, and "Digest = [Table 20-3]" from 92.
_4] ", the junction gate 94 is cleared, and the output of the junction gate 94 becomes the" ignition "level signal of 93. The same circuit is used even if the pulse signal is used as the level signal. * The instruction cell responsible for the protocol of the present invention may not be a sequential process but may be simply a "combinational logic circuit". In this case, even if the D-FFs in FIG.
It is disclosed in the above-mentioned patent that it can be realized because there is a clock φ in 4. * In the same way, create the “operations” of all the programs and protocols in [Table 4] to [Table 26] using a sequential processing circuit or a combinational circuit, and set the respective “ignition”, “seeding”, and “ACK”
If an ignition signal is generated by a junction gate that inputs, all programs and protocols can be replaced with electronic circuits.

[Industrial applicability]

As described in [Object of the invention], the present invention provides for the transfer of messages between any computer regardless of whether the computer is removed or the computer is added via an input / output device to the network. It is a network to be realized. However, the transmission lines in Fig. 6 are connected by wired OR. Theoretically, an unlimited length of transmission line can be laid, but the transmission line shown in Fig. 6 can be used in actual industries only within a few tens of centimeters. Therefore, a method of laying a long-distance transmission line without breaking the wired OR, which is the basis of the network theory of the present invention, will be described. Figure 12 shows a method of actually laying a transmission line in industry. Figure 12 is explained. * Since all six transmission lines in Fig. 6 are the same, any one was taken out. * Although there are seven MARS units in Fig. 6, two arbitrary units were taken out. * The two connected NOT gates described in Fig. 6 and Fig. 9 are shown separated in 95 in Fig. 12. * If two BUFF gates in 96 are replaced with wires and three AND gates in 97 are replaced with wires, it will be exactly the same as the wired OR in Fig. 6. Therefore, it is confirmed that the number of transmission lines in Fig. 12 is one. * If the ‥ on the left side 95 is L level and the ‥ on the right side 95 is L level, the transmission lines 98 and 99 become H level, and R ‥ of 95 can detect L level. * If ‥ on the left side 95 is H level and ‥ on the right side 95 is L level, the transmission line 98 is L level, 99 is H level, and R ‥ 95 can be detected as H level. * Assuming that the left side 95 is L level and the right side 95 is H level, the transmission line 98 is at H level and 99 is at L level. * If the left-hand side 95 is set to the H level and the right-hand side 95 is set to the H level, the transmission lines 98 and 99 are set to the L level, and the R level of 95 can be detected as the H level. In the above description, it can be confirmed that the wired OR similar to that of FIG. 6 is established in the case of two MARS units in FIG. Similarly, even if many MARS units are connected, wired OR is established if 96 and 97 are installed. The distance between 95 and 96 is D1, the distance between 96 and 97 is D2, the distance between 97
This is described as D3. * D1 is an open collector single wire connection.
If it is less than 10cm, the signal may not be transmitted correctly. * If a dedicated transmission / reception IC is used for 96 and 97, D2 and D3 will transmit signals correctly for about 100m. * There are two sets of one-way transmission lines between 97s. If Figure 6
If Fig. 12 is applied to Fig. 12, there are two sets of transmission lines created with six one-way traffic. Conversion of one-way parallel signals to serial signals is possible with existing technology. * If the signal is converted to a serial signal, it is possible to use the existing transmission line such as a coaxial cable or optical cable. As described above, in the case of industrial use, the components 96 and 97 in FIG. 12 are changed according to the distance of D1, D2, and D3, serial conversion is added, and QAM (quadrature amplitude modur
ation). Using the method of the present invention, the method of constructing a multi-CPU system, the features of a LAN system, the features of a WAN connection, and the method and features of creating a centralized network server are described. This paper describes how to build a multi-CPU system. * Since a loosely coupled multi-CPU system is for communication between short-range processors, it is laid on the transmission line in Fig. 6 without using Fig. 12, and a network is created in Fig. 2. * The common memory type multi-CPU system, etc. describes the transmission path securing bidding protocol (Claim 3) in [Table 17] on multiple arbiter devices and is constructed as follows. ☆ The CPU prompts its arbiter device to pass through the gate of the common bus. ☆ When the arbiter device is prompted, it can pass through the gate of the common bus when winning the transmission line
Inform ☆ The CPU can read and write to the common memory because the common bus is in an exclusive state. ☆ When the CPU finishes reading and writing using the common memory, it prompts its arbiter device to release it so that other arbiter devices can win the bid for securing the transmission path. The features of 41, 42, 43 in Fig. 4 and the LAN system in Fig. 2 are listed. ★ Message communication on public telephones can be called from any telephone in the world to any telephone.
A common use is to call a short distance or a specific person. The same applies to the network of the present invention, in which the data is transferred immediately when the distance is short, and the number of transfers increases when the distance is long. The LAN created by the method of the present invention is a realistic method. ★ When the MARS unit starts receiving in [Table 13-3] and [Table 21-5], it is possible to activate the protocol to start transmission immediately. Conditions that must be able to store data, such as the packet method, have been removed. That is, data transfer is performed in a continuous manner. Conversely, if there is an inconvenience in the continuous method, it automatically changes to the packet method. ★ The biggest disadvantage of the continuous method is when computers connected to all MARS units wish to transfer messages at the same time. However, as described with reference to FIG. 6, the first priority of the bid is the unit of the upper layer, and the message from the upper layer is given the highest priority due to the degree of freedom of the memory described in [Table 8]. Delivered reliably to lower-layer computers. This means that even if there is a lot of traffic,
This means that no loss of forwarded messages will occur. * The three methods described in [Embodiments of the invention] differ in the message transfer rate in each transmission path. However, they can be connected within one LAN. ☆ Because, at most, one LAN transmission line [Table 1]
Connect only 'N' MARS units described in. ☆ Therefore, it is not necessary to set “τ1” for the upper transmission line in [Table 12-6] and “τ1” for the lower transmission line in [Table 12-8] to the same value. Also, it is possible to set different numerical values from CPU16. ☆ When a message received from an arbitrary transmission path is transmitted to another transmission path, it is temporarily stored in the memory 20, so no inconvenience occurs even if the transfer rate is different. ★ To obtain at most 'N' consistency means that high-level technology is not required in laying transmission lines. * If the computer is defined in paragraph 2 of [Definition of phrases used in this specification], only one MARS unit must be compatible, and message transfer is performed by using a transmission path selection protocol, conflict With a protocol that eliminates, you can transfer without the help of a computer. ★ If the transmission line is faulty or the destination unit is not connected, it is returned to the source computer following the reverse route of the transfer. ★ Returning a transmission line failure to the sending computer means that the network transmission line failure can be immediately determined. The features of the WAN connection in Fig. 4 are described. ★ The WAN protocol for connecting the LANs [Table 21] to [Table 26] is not a method of transferring messages in a continuous manner.
I chose the packet method. ★ The reason why the WAN transmission path is set to the packet method is to consider that the laying person and the user of the transmission path are separated as the area becomes wider. In other words, it indicates that it is possible to calculate the usage fee of the transmission path. ★ The method of calculating the usage fee using a computer connected to the MARS unit is to send a firing signal to start [Table 14] in [Table 26-2] while receiving messages continuously. You can get the charge calculation data. * Hot-Line, one of the WAN transmission lines shown in Fig. 4, can be changed only by changing the address message indicating the route in Fig. 7.
Can be used to transfer messages between any computers. * If a Hot-Line is laid, there will be multiple transfer routes, and if a certain transmission line fails, another emergency transmission line can be secured. This paper describes how to create a centralized network server. ★ Put the transmission line and MARS unit in one case. ★ Do not connect a CPU like [Table 11] or [Table 12] to all units. ★ As described in the LAN feature, in order to ensure consistency between one MARS unit and one computer, the lower layer units and computers have been changed from [Table 12] and [Table 13] to protocols such as RS232C. You may. ★ By disconnecting the lower transmission line terminal of the lowest unit, if N1 = 4 and N2 = 3, the MARS unit will be 2,95
With five, a centralized network server that can connect 2,744 computers is completed. The centralized network server created by the above method has the following features. * One server is the same as in Fig. 2, with one WAN transmission line that can be connected at the upper transmission line terminal and 2,744 transmission lines that can be connected at the lower transmission line terminal. Therefore, one centralized network server is designated as 42 in FIG. 4, and 2,744 centralized network servers of 41 and 43 can be connected. * When two servers are connected as shown in Fig. 4, the number of computers that can be connected is the total number of servers (1 + 2744) x the number of computers that can be connected to one server (2744) = 7,532,280. ★ Communication between computers is not only RS232C, but also RS449, S100
Bus, GB-IB, MULTBUS, EUROBUS, Centronics type, ... One CPU at the bottom of the centralized network server
In order to ensure consistency with the connection terminal 12, if the MARS unit is replaced with only [Table 12] and [Table 13] and dedicated terminals are provided,
Standards can be unified even with existing different communication formats. * Computers that fall under the definition of [phrases used in this specification] are not limited to office equipment such as printers and floppy disks, but include coolers, audio equipment, industrial machinery, etc. There are many types. But M
Just by matching one of the ARS units, everything can be connected and control and information can be obtained from any desktop computer. ★ Of course, distributing MARS units throughout the building or in a limited area without storing the server in a single housing,
This is easy from the explanation of Fig. 12.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04L 12/44 H04L 12/28

Claims (5)

(57) [Claims] 1. A unit for forming a network for a digital computer and a transmission line A. For a unit: A-1. One connection terminal for a digital computer (12) and one connection terminal for an upper transmission line (13). ) And one lower transmission line terminal (14). A-2. A number of units are assigned different numbers. A-3. A single digit has a different number. Units are grouped into one unit group, B. A bus-type transmission line for connecting units is laid, and C. One bus-type transmission line is connected to many units existing in one unit group as conditions. C-1. Connecting the lower transmission line terminal of one unit and the upper transmission line terminals of many units having other unit numbers to one bus type transmission line, or C-2. One bus type transmission On the road By connecting the higher-level transmission line terminals of all the units to form a LAN in which the unit groups are connected by a bus-type transmission line, one of the transmission line terminals (13, 14) is not connected to the bus-type transmission line D. Hierarchical network with multiple transfer routes by connecting the lower transmission line terminals of two units in one constructed LAN to each other with a dedicated transmission line on a one-to-one basis Or E. A wide-area network with a transmission line connecting multiple LANs by connecting one unit in the constructed LAN and one unit in the other constructed LAN with a dedicated one-to-one transmission line. . 2. In a message received from an arbitrary terminal of a unit having a self-number having three terminals, A. If the head address and the self-number are the same, select a digital computer connection terminal; B. For digits whose head number is different from the self-number and whose maximum number used in the network is greater than the number (layer number) obtained by adding 1 to the number used consecutively to the self-number Compare and collate, and if they are the same, select the lower transmission line terminal, and change the number corresponding to the value obtained by subtracting the layer number from the address listed at the beginning to the layer number obtained by subtracting the layer number from the own number by one. Transfer the number replaced with the corresponding digit as the transfer destination unit number. C. The address described at the beginning and the own number are different, and. Compare and match the digits above the layer number, and if they are different, select the upper transmission line terminal. C-1. If the numbers are the same, the digit of the digit corresponding to the layer number from the first address is
C-2. If the numbers are different, replace the maximum number with the digit corresponding to the layer number in the own number. By transferring the number replaced with a number as the transfer destination unit number, and comparing and comparing the message head address and the unit own number, only one terminal is selected from the three transmission line terminals, and further selected. Unit that has a protocol that can identify the transfer destination unit connected to the specified terminal. 3. The network for a digital computer according to claim 1, wherein the network for a digital computer includes the following elements: connected to one bus-type transmission line composed of wired ORs and assigned different numbers. A plurality of units: A. Wait until the TE is released, B. Wait until the VC is released, C. Put the TE in a use state by setting WTE to H level, E. Sends the signal of the own unit number to the transmission line (WM2 to WM0), D. Waits for the signal to propagate through the entire transmission line, and E. If VC is in use, sets the data bus transmission line to L level. Repeat from B. F. If the signal appearing on the data bus transmission line is different from the own unit number, F-1. Compare the signal appearing on the data bus transmission line with the own unit number for 1 bit. If they are different, set the data bus transmission line to L level, wait until VC is in use, repeat from B. F-2.1 Compare the signal that appeared on the data bus transmission line with the unit number for one bit If they are the same, compare one different bit and repeat from D. G. If the signal appearing on the data bus transmission line is the same as its own unit number, G-1. G-2. Wait for the signal to propagate through the entire transmission path by sending a signal indicating that the VC is in use. By detecting the signal appearing on the transmission path and by putting the signal on the transmission path, G-2. If the desired unit participates until it wins the competitive bid with another unit by the method of ABCDEFF-1.F-2., And knows that it has won the competitive bid by the method of G., Win a competitive bid to another unit in the method of. That other units open the data bus transmission line in the manner described in E. above at the time of G-2., And from a plurality of units using the bus-type transmission line. A unit that has an arbitration mechanism that selects only one unit. 4. The unit according to claim 2, wherein each of the units connected to the lowermost transmission line constituting the network does not have a protocol relating to a lower transmission line terminal. 5. A closed digital computer network, comprising: A-1. An address is assigned to a computer or a unit having a transfer function which is a component of the network; A-2. The complete destination of the entire message frame from any source component of the network to the destination component, depending on the destination address, and the number of existing closed networks that satisfy the conditions of A-1. And A-2. Are connected to each other to transfer data. B. A message frame transmitted from an arbitrary transmission source of the first network is represented by a first network exit address, a second network exit address,. Network exit address, nth network destination address, marked first
C. the source address of the network, the length of the data field, the data, in order, C. internal transfer to the egress component of the first network according to A-2 above, D. the egress of the closed existing network A one-to-one dedicated transmission line is laid between the component and the entrance component of the next closed existing network. E. For the message frame received by the exit component, E-1. If the address of the element and the element address are the same and the next second address is marked, it shall be a message frame addressed to itself. E-2. First address and configuration If the element addresses are the same and the next second address is not marked, the message frame is transferred to the one-to-one transmission line laid in D. above. F. Ingress component of the network connected to the dedicated transmission line used in E-2 above only when the frame meets the conditions of E-2 above. F-1. In the address frame of the received message frame, the address from the second address to the unmarked address is shifted by one frame to the left to change the address description position, and F-2. After the processing of F-1., The last address frame not marked is subjected to the processing of describing the marked entry component address of F. The contents of the address frame part of the received message frame are processed. G. Create a new message frame for internal transfer on the closed existing network to which it belongs, and H. Second existing network Shall transfer the G. message frame internally to its own egress component according to A-2 above, and I. repeat the above D. through H. methods. I-1. Third through n. I-2. The entry component of the existing network changes the contents of the address frame part. I-2. In the networks up to the (n-1) th, it is forwarded internally to its own exit component. A-2. Introduces a message frame to its own destination component according to A-2., And applies a series of methods B. to J. to send from any source that is a component of a closed existing network. The message frame described in paragraph B. above is sent to multiple closed existing networks by a series of addresses and marks at the top of the message frame. It changes each time it passes through a dedicated transmission line laid one-to-one, passes through a closed existing network that is responsible for relaying, and is delivered to any existing network destination component. The component that received the message frame described in B. or G. above performs one of a series of methods that can transfer the data after the description of the source address of the described message frame without changing it. As an act, the exit component is E-1.E-
Method 2 in which the inlet component performs F-1.F-2.