WO1996031968A1 - Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement - Google Patents

Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement Download PDF

Info

Publication number
WO1996031968A1
WO1996031968A1 PCT/JP1995/000653 JP9500653W WO9631968A1 WO 1996031968 A1 WO1996031968 A1 WO 1996031968A1 JP 9500653 W JP9500653 W JP 9500653W WO 9631968 A1 WO9631968 A1 WO 9631968A1
Authority
WO
WIPO (PCT)
Prior art keywords
unit
runway
message
terminal
bus
Prior art date
Application number
PCT/JP1995/000653
Other languages
English (en)
Japanese (ja)
Inventor
Yuji Nagai
Original Assignee
Mars Technology Institute Co., Ltd.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Mars Technology Institute Co., Ltd. filed Critical Mars Technology Institute Co., Ltd.
Priority to JP53016996A priority Critical patent/JP3338059B2/ja
Priority to PCT/JP1995/000653 priority patent/WO1996031968A1/fr
Priority to AU20857/95A priority patent/AU2085795A/en
Publication of WO1996031968A1 publication Critical patent/WO1996031968A1/fr

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00—Data switching networks
    • H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
    • H04L12/44—Star or tree networks

Definitions

  • the present invention is applied to a network connecting a large number of computers.
  • the network according to the present invention is a combination of a point-to-point channel and a bus-type broadcast channel in a tree structure. Build with a topology.
  • multi-CPU centra I processing unit
  • LAN locaria network
  • WAN wide area network
  • the network / ⁇ human output resynchronization is performed using an IP (interface message processor; ⁇ '3 3 3 3 ⁇ 3) (packet snitch node>, medium ⁇ , (intermed iate system.), Data exchange (da ta SK it ch ng exchange).) 1 ⁇ (router). , Br router router router router router ⁇ ⁇ r inter r r inter inter inter inter er), ... is called by such a term.
  • IP interface message processor
  • a point-to-point channel requires a dedicated runway between the computers, which complicates the installation of the runway when many computers are connected. Since many runways are connected to a computer, it was necessary to decide which runway to select. For example, a data transfer system with a PCTJP8700 S ].
  • the broadcast channel does not require a dedicated runway between computers, and the network is completed by connecting to the runway after laying the runway.
  • the IEEE 802.4 token bus and the IEEE 802.5 token ring are a method of eliminating transfer data transfer without using centralized management by a server unit.
  • PC TZU S 87 0 2 3 88 Example of arbitration performed by a micro computer in an I / O device] In any case, in order to expand the network, it had to be used in conjunction with other methods.
  • PCT / FR 8 1 0 0 0 2 9 [Computation «Fast message transmission method] can be used.
  • arbiter equipment PC TZD E 85 050 0 5 2 5 [Multiprocessor system] and PCT / US 88/0 295 5 5 [Method and device for interconnecting buses in a multi-pass computer system]
  • the method of logically synthesizing digital signals is based on the use of processing gate elements. However, there is a method to synthesize Ift without using logic gates (wire-logic ⁇ ni red-logic).
  • wired logic There are two types of wired logic: an AND tie (AND-tie) that connects the outputs of open collectors directly with a pull-up resistor, and a wired OR (iMred-OR) that directly connects the outputs of open emitters with a pull-down resistor.
  • AND-tie AND-tie
  • IMred-OR wired OR
  • the input / output devices to the network are called IMPs, packet switching nodes, relay systems, data exchanges, routers, bridges, repeaters, etc., and a unified concept has been established in various terms. No. However, the forward message destination device is now widely referred to as the servant.
  • the MARS unit is not an I / O or a computer but an independent machine S.
  • an ⁇ ⁇ ⁇ dynamic typewriter is not a computer.
  • the printer controlled by the internal CPU is a computer
  • the printer controlled by the external CPU is an IZO machine
  • the MAR Sunit can be controlled by its own internal CPU, or it can be controlled by an external CPU to be a group of computers outside the concept of a computer.
  • Fig. 3 is explained, the outline of the transmission, transfer and reception protocols in the LAN of the present invention is described in a concrete example, and it helps to draw the concept of the present invention.
  • Figure 1 shows the input / output appearance of the MARS (Message Access and Repeat Server) unit described in (Definition of terms used in this specification).
  • MARS Message Access and Repeat Server
  • Figure 2 shows an example of a LAN connected to a large number of MARS units in a network
  • Figure 3 shows a specific example of a frame format conforming to IEEE 802.3, which is a message in AN.
  • the MARS unit 1 will be explained.
  • the MARS unit 11 has two different transmission line terminals and one CPU bus connection terminal 12. Of the two transmission line terminals, one is an upper transmission line terminal 13, and the other is a lower transmission line terminal 14. To do.
  • Each transmission path terminal must be connected to a different transmission path bus of the broadcast channel or connected to a point-to-point channel transmission path.
  • the transmission line connected to the upper transmission line terminal 13 is the upper transmission line.
  • the transmission path connected to the lower transmission path terminal 14 is a lower transmission path.
  • the names of the upper transmission line and the lower transmission line are as viewed from the MARS unit. When viewed from the transmission line, they are connected to either the upper transmission line terminal or the lower transmission line terminal.
  • the computer connected to the CPU bus connection terminal 12 is connected to a computer consisting of a memory 15 and a CPU 16 (see the above [Definition of terms]).
  • the control of the MARS unit 11 may be controlled by the micro processor provided in the unit or by the CPU 16.
  • the outline of the LAN of the present invention will be described with reference to FIG. To do.
  • the network is completed by connecting the MARS unit and the runway.
  • All MARS units are represented by rectangles, and numbers are written in the rectangles. ⁇ ⁇ ⁇ The memory 15 and CPU 16 connected to the MARS unit are omitted from the illustration.
  • Figure 2 is an example of a network, where the single-digit * large number used is 7.
  • the runway is divided into a runway for LAN and a runway for WAN.
  • the transmission path for LAN is indicated by a solid line with a width of 0.4 M.
  • the transmission path for LAN is a broadcast channel and is a bus type that is connected to a large number of MARS units.
  • the WAN runway is a point-to-point channel, connected by one MAR Sunit and another MARS unit (see Fig. 4).
  • MARS unit to JR There is only one MARS unit to JR at the highest layer. It has the highest unit number in the LAN (7777 in Fig. 2).
  • the number of the runway for the LAN must have '0' in any digit, and there must be only one.
  • the number of the MARS unit does not have '0' in all digits.
  • the number of the WAN runway is the same as that of the connected MARS unit.
  • a LAN runway replace the '0' somewhere in the LAN runway number with another number
  • Figure 3 shows an example of a message, which is a frame format for LAN.
  • the antenna and the frame start delimiter are added before the destination address, and the checksum and the like are added after the destination address. It is omitted because it is provided.
  • the message 31 is a message transmitted from the CPU 16 in FIG. 1 to the MARS unit 11 and a message flowing to the LAN of the present invention.
  • Messages flowing on the runway need not be completely store-and-fornord, but messages must be temporarily stored.
  • the message 31 in Fig. 3 is sent to the network in Fig. 2 by the CPU 3167, and 367 ⁇ (3107) ⁇ 3177 ⁇ (3077) ⁇ 3677 ⁇ (3607)-* 3 657
  • the concept of the protocol of the present invention will be explained by showing an example in which the transfer is performed continuously, and it is assumed that no message combination occurs for the sake of general explanation.
  • the CPU 3167 sends the message 31 to the MARS unit 3167 via the CPU bus, and the unit receives the message. (Transfer in Fig. 1)
  • the MARS unit 3167 receives the message and judges whether the address described is correct or not.
  • Protocol of transfer unit 3 1 67
  • the transfer unit 3167 Since the transfer unit 3167 has layer number 2, it is determined whether or not the second digit or more of the first 3657 of the own numbers 3167 and 31 match.
  • transfer unit 3167 Since transfer unit 3167 has layer number 2, mask the first two digits of self-number 3167 with the first three digits 57 of 31 and determine whether they match.
  • MARS Unit 317 7 7 turns into “Reception Unit” and receives message 31.
  • the transfer unit 3177 Since the transfer unit 3177 has layer number 3, it is determined whether or not the third digit or more of the prefixes 3657 of the own numbers 3177 and 31 match.
  • transfer unit 3177 Since transfer unit 3177 has calendar number 3, mask the third digit of the first number 3567 of self-number 3177 and 31 and judge whether they match.
  • transmission is performed from the higher runway 3077 to the unit number 3677.
  • the MARS unit 36777 receives the message 31.
  • Judgment is made based on the match between the self number 3 1 7 7 and the head 3 6 5 7 of 31.
  • transfer unit 3677 Since transfer unit 3677 has layer number 3, it is determined whether or not the 3rd digit or more of the first 3657 of 31 matches.
  • MARS unit 3657 receives message 31.
  • Judgment is made based on the coincidence between the self number 36 57 and the beginning 365 7 of 31.
  • the message transfer can be performed without any influence on the memory 15 and the CPU 16 of the computer. Can do.
  • Fig. 4 shows an example of a panoramic view of the network of the present invention.
  • Figure 5 shows a transfer message in a network, and is an example of frame format conforming to 31 in Fig. 3.
  • AN «LAN runway is indicated by a solid line with a width of 0.4M
  • WAN runway is 0.4M in width. This is indicated by a broken line.
  • any LAN there is only one WAN runway connected to the upper runway terminal, which is the largest unit number of LA N ⁇ .
  • Units 1267 and 3277 on LAN41 are interpreted as not having the LAN runway unnecessary.
  • Unit 33777 in LAN 43 does not have a LAN runway as the upper runway, but has a WAN runway. This is the maximum unit number.
  • 277 7 is the maximum unit number.
  • the MARS unit connected to the LAN runway has been connected in the order of the number until now. It is connected to the tongue number as in LN43, such as 3377.3677, 3477, and 3177. There is no need to control.
  • a WAN runway connects the lower runway terminal of one MARS unit to the upper runway terminal of another MARS unit. 47777, 42-2546 and 43-3777.
  • the lower runway terminals can be connected to each other.
  • the length of the transmission line K for LAN and the length of the runway for WAN are far from the initial setting to the MARS unit, depending on whether the runway number has '0' or not. This is a specific example of a message passing through the runway.
  • Figure 5 is a specific example consisting of three types of JB.
  • Message 5 is used in L AN 41
  • message 52 is used in L A N 42
  • message 53 is used in L A N 43.
  • the 1.41 CPU 1247 sends a message to the MARS unit 12447 via the CPU bus, and the unit receives the message. (Transfer in Fig. 1) And 41 in Fig. 5.
  • the message is created.
  • the transfer route is created.
  • the AR Sunit 7777 determines whether or not it matches the first address 7777 of 51. (Same as the protocol using Fig. 2)
  • the change method is to replace the first 7 7 7 7 of the message with the own number 2 5 3 7.
  • the MARS unit 2546 determines whether or not it matches the first address 2546 of 52. (Same as the protocol using Fig. 2)
  • message 52 check if there is a mark # next to the first 2 54 6. Since it is "2 546 34 2 6" and there is no mark, it is judged to be a transfer message, and it transforms itself into-("WAN transmission unit").
  • the MARS unit 3426 determines whether or not it matches the first address 3426 of 53. (Same as the protocol using Fig. 2)
  • Fig. 6 shows an example of a theoretical runway for a LAN.
  • Fig. 6 explains the protocol for avoiding message consolidation.
  • the "WAN transmission unit" and ("WAN The receiving unit J ISI is a point-to-point connection.
  • a dedicated runway is laid. No message consolidation occurs.
  • the method for avoiding message contention on the LAN runway of the present invention is a special-purpose runway.
  • the protocol was settled. The protocol is shown in ⁇ .
  • the winning sending unit broadcasts the address of the receiving unit to the other MARS units.
  • the other MARS unit decides whether to turn into a receiving unit.
  • the sending unit talks about the existence of the receiving unit.
  • Fig. 6 shows an example of the transmission path 2077 connected to 2777, 2377, 2577, 2177726777, 2277, and 2477 in the LAN 42 in Fig. 4.
  • C6 1-digit number that replaces '0' with '6', which was intimately connected to the driveway 2077.
  • the lower runway terminal 14 in Fig. 1 is 2777, and 2177, 2277, 2377, 2477.2577, and 2677 are the upper runway terminal 13 in Fig. 1.
  • the relationship between one runway and the terminals of many MARS units is a wired logic structure.
  • M2, M1, M0 are connected as three message buses as message buses on the runway. Collect VI 0 and write it as Bus.
  • the reason why the number of message buses is three is that the maximum number used for LAN 42 is '7', and '7' is expressed as a binary number of 111, which requires three digits.
  • WT G, R TG and TG have the same meaning as a synchronization signal (trigger) that reads data through the data bus of a computer.
  • R TG is not always used to extract the Bus signal.
  • WT E, R TE and TE are used to indicate the status of bidding to make runway 2077 between two MARS units;
  • WVC, RVC, and VC are used to indicate the situation when a winner is awarded by bidding.
  • [6] waits for RTE-L to participate in the bidding, and immediately participates in the bidding as WTE-H, Bus-110.
  • [5] is the sender who won the bid, so wait for a while without doing anything.
  • [5] is .Bus-000 WTG-L.
  • [6 110] landed himself in the bidding, but remained at WTE-H. Due to the wired door, the runway was not fully open at RTE-H.
  • [6 110] dropped off for the first bidding, but waited for the opportunity to bid again until WTE-H.
  • the re-bidding is the same as the first bidding "2-2:", and is entered as Bus-110.
  • the order of priority is determined uniquely, and if ordered from the highest, [7], [3], [5], [1]
  • the number of memories belonging to a layer increases exponentially as it goes to lower layers.
  • [Table 1] shows a method of using the MARS unit shown in Fig. 1 to construct the networks of the present invention shown in Figs.
  • one of the 'N's in the upper digit of the unit number is set to' 0 ', and the AN runway can be laid.
  • Ho ne connects the lower runway terminals of any two units
  • the LAN of the present invention is completed by writing appropriate numbers for 'HI' and 'N2' in [Table 1] and writing 'N' for social conventions.
  • Nl 4
  • 'F' is a hexadecimal representation commonly used in the computer world.
  • Nl 5
  • Nl 5
  • ' 1111 is a binary representation commonly used in the computer world.
  • 2.N2 defines the number of digits of the transmission K and MARS unit numbers.
  • N1XN2 calculates the number of bits required to represent the address excluding the mark # described in Figs. 3 and 5.
  • the layer number can be calculated from the unit number.
  • the layer number is calculated by adding 1 to the number of 'N' ffls used for the unit number.
  • the layer number is calculated by adding 1 to the number of digits with 'o' in the upper transmission path number.
  • the LAN created by the exceptional rules is connected to the hotline as a .WAN runway.
  • the advantage is that about 10% of computers can be connected.
  • Table 2 shows the maximum number of units that can be connected when the network of the present invention is constructed according to the method shown in Table 1.1. (30), (3 or. Is the maximum number of units that can be connected to one LAN transmission path.
  • N1 and N2 are the minimum number of .Bus and the number of digits of the unit address.
  • Nl X 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 and Fig. 5 / The number of bits used for the mark # Calculated except for
  • LAN ⁇ is the network of LA ⁇ and 05 in Fig. 3.
  • BAN4 is the network in which the LAN of Fig. 3 is connected to the WAN transmission path at the bottom of Fig. 5 in order.
  • FIG. 5 The frame format used in the network of the present invention in FIGS. 3 and 5 is shown in FIG. 5, and a large number of addresses are composed of an “address message” and a “data field length”. If they cannot be clearly separated, they cannot be correctly transferred on the network of the present invention.
  • the end bit (endiark), ⁇ , and the return route mark $ at the time of return can be used in addition to the # mark by using the token bit.
  • Example of frame format of all bits 16 token address parity bit 3rd digit-2nd digit 1st digit bit
  • Figure 7 shows the frame format during the transfer of a route with six LANS » which is an example of AN4 in [Table 2].
  • the route is shown below, and is an example of [L A N 3 Inlet-L A N 3 Out].
  • Fig. 8 shows an example in which an arbitrary unit in LAN 3 detects the failure, rearranges the address messages in Fig. 7, and changes the mark ⁇ token and end-nark).
  • Figure 8 the destination is simply filled in. Decide on 81-83 so that you can view it as an address message.
  • 84 is the MARS unit that found the camp.
  • Numbers 85 to 87 are the lower entry addresses of the LANs 71 and 73 that were to be entered in Figure 7 from now on.
  • 88 is the destination address in Figure 7 and Table 4 ”
  • the first frame of the address message described in [Table 6] is 71 in Fig. 7 or 8 in Fig. 8, and the second frame is 72 in Fig. 7 or 82 in Fig. 8, and the final frame is It is 77 in Fig. 7 or 88 in H8.
  • [Table 5] is an address change program when a message or message is received from the WAN transmission route. .
  • 'Table' This is an address and change program for returning a message / sage. This is also applicable to the case of return only within the LAN as shown in Fig. 2.
  • the change of the message frame is completed by the operation of 1 to 4.
  • a solid line with a width of 0.4 BBI is called a data bus in a general computer.
  • a dashed line with a width of 0.4 ⁇ is called an address bus in a general computer.
  • the troll wires are shown.
  • I-selectors 21, 22.23 are .20 and the CPU 16 bus shown in Fig. 1 or three internal parts. It is a connection to connect the bus.
  • Memory 20 is a time-divisional access method that uses four data buses (tine shar ins systeo)-»K
  • the memory access concurrency in a tight sense is .3 memory accesses. 3 ⁇ 45
  • SPU 06 select processing unit: Memory management
  • MPU 26 (aain processing unit): Message content processing
  • a and U ar ithmetic and logic unit 29, 39, 49.
  • bus converters ffi34, 35, 44 and 45 for converting four digital signal groups.
  • the selectors 21 to 23 are determined.
  • the selectors 21 to 23 are connected to the memory 20 connected to the determined selector and the data bus corresponding to the processors 16, 26, 36, and 46.
  • processor 5 (1 »1" 6 enab 1 e) or RE (read enable) signal is output from processor 16, 26, 36, 46. Connect the applicable bus to the applicable memory 20 .
  • ⁇ '01, 10, 11 'written in each of the four W's in the status table is less than one each.
  • the degree of freedom of memory is the amount that is not made accessible in [Table 7] of the three memories 20 in Fig. 9.
  • W WM. Wl W WM. Wl
  • WL are given to specify the setting memory .
  • Operation According to the set WC, WM, WU and WL.
  • ACKSPU06 sends a readable signal to MPU26.
  • MPU 26 requests read from memory 20 for itself
  • RU.RL is given to specify the constant memory.
  • Operation Write the value in the corresponding R column in the R column for MPU.
  • ACKSPU06 sends an operation completion signal to MPU26.
  • MPU 26 requests read from memory 20 for another processor
  • ACKSPU06 sends a readable signal to the determined processor
  • SPU 06 sends an operation complete signal to MPU 26.
  • the B8 section of the memory 20 and the CPU 16, MP U26, UP U36, and LPU 46 is the same as the concept of the conventional common memory type multiprocessor, but the rebuilding method is different.
  • the MAR Sunit is a method of transferring messages in the context of the memories 12, 13, 14 and the memory 20.
  • the message is transferred from the computer to the memory 20 through the terminal 12 and is transferred through the terminal 13 or 14.
  • the message is transferred to the memory 20 of another MARS unit, and the message is transferred from the memory 20 to the computer through the terminal 12.
  • Table 7 shows the contents of the registers indicating which PUs the memory 20 is under at any given time.
  • the hardware is designed to switch the selector according to the contents of the register in [Table 7].
  • the SPU 06 is used when another PU (16, 26, 36, 46) wants to write to any of the three memories 20, or when another PU wants to read from the memory 20. 8) or [Table 9] to switch registers, and issue an ACK (acknowledge) signal.
  • SP U06 disconnects when prompted to release memory 20 by [Table 10].
  • MPU 26 prompts release of writing to memory 20
  • Alossesor 16.26, 36, 46 prompts release of reading from memory 20
  • Fire extinguishing is the logical inversion of seed fire.
  • the four groups show multiprocessors that can execute instruction cells independently of each other. However, the four and five instruction cells in a single group execute four
  • the hardware configuration is such that WM 2, WM 1, ⁇ , W VC, and W TE in Fig. 6 are all at the L level before and after turning on the power.
  • the MARS unit is prompted to transfer the message from CPU 16 in [Table 11] to memory 20, and the operation of the instruction cell is started.
  • CPU IB is initially transferred to the memory 20 messages, be default setting data MA RS Yuni' bets.
  • protocol of Table 12 forwards to each processor.
  • the MPU 26 reads in a time-sharing manner and executes according to the protocol shown in [Table 13], even during transfer from the terminals 12, 13, 14 to the memory 20. If the power terminal is 12, it is transferred to a computer using the protocol in [Table 14].
  • the firing MPU 26 is prompted to forward the message described in the memory 20 associated with the selector described in the R column for the MPU in [Table 7] to the CP LM6.
  • ⁇ Table 13-4 _2 Table 13-5 _ 1 Operation
  • the memory type is ROM, it is set in ⁇ Table 12-2
  • Ignition SPU 06 is notified from CPU 16 that reception has been completed. ⁇ Table 14-6
  • Operation CPU 16 is prompted to receive a message from SPU 06.
  • the killing work CPU 16 is synchronized with the RE signal generated by itself,
  • the CPU 16 outputs a reception completion signal to the SPU 06. ⁇ Table 14-3 Send.
  • the firing MPU 26 prompts the UPU 36 to transfer the message described in the memory 20 specified in the R shelf for the MPU in [Table 7].
  • [Table 17] and [Table 18] are UPU36 programs if they are part of [Table 15].
  • Ignition MP U26 prompts LP U46 to transfer the message described in memory 20 specified by Rffil for the MPU in [Table 7].
  • ⁇ Table 13-5 _4 Set operation RL to RL ⁇ Requested [Table 9-2] and finished.
  • Ignition I and PU 46 determine the transfer destination unit number. ⁇ Table 16-1_1 Operation! About the address of the first frame of the message.
  • Ignition LPU 46 starts sending messages to the RAN driveway.
  • ⁇ Table 18-2 _2 Reads the top of the operation message ⁇ Figure 9; 46 ----- -20
  • Behavior Bus Performs '0', and after time “te”, performs WTG L,
  • Bus 1-digit destination. Destination unit is ⁇ page 18 6-2
  • step 2 Since it exists, the "ignition" signal in step 2 is output and the operation ends.
  • the failure of the LAN driveway is determined by [Table 17-3 _7], and the mate or unconnected position of the MARS unit connected to the LAN driveway is determined by [Table 18-1 _9]. Also, in the WAN driveway, it is judged that the driveway is »» in [Table 23-3_1].
  • the transmission follows the protocol shown in Table 21 and the reception follows the protocol shown in Table 24.
  • the transmission follows the protocol in [Table 22] and the reception follows the protocol in [Table 25].
  • Ignition MPU 26 is prompted to return the message.
  • the permission signal from ACKSP U06 is also used as the "seeding" signal in 4.
  • Operation Address bus for memory 20 is set to 0. ⁇ Figure 9; 36, 46 --- 20 Operation Only once when a change is detected in any RTG
  • Ignition MPU26 prompts PPU36 to transfer the message described in the memory 20 specified by R «8 for MPU in Table 7J. ⁇ Table 13-5 _:
  • Ignition LP U 36 gets read permission from SP U06. ⁇ Table 22-1 _1
  • one MARS unit is considered to be connected to the upper transmission line terminal 13 as described in Remarks 1 in [Table 22] and [Table 25]. Program.
  • the protocol for the WAN was set to be the same as the LAN runway in Fig. 6, and the protocol was explained assuming that message competition would occur. No message conflicts occur.
  • Ignition i UPU 36 is set to “LAN transmission line” by default.
  • transition status table is provided for a clearer understanding of the program. However, in the embodiment having four processors of the present invention, the transition status table is described in the instruction cell of the data flow system. Figure 10 shows the transition table that helps.
  • Figure 10-A shows the flow of the default data and the table used.
  • 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
  • FIG. 9 there are a plurality of processors, and [disclosure of the invention] conforms to the description of the data flow system. And explained the protocol.
  • a three-state buffer is provided at terminal 12 in Fig. 9, and I / O 0 shown in [Definition of words used in this specification] is included, including bus conversion 34, 35.44.45. 9. Make one solid line written in parallel with a width of 0.4 an of 9 into one, and use it as a data bus. * Make one dashed line written in parallel with a width of 0.4> »in Fig. 9 to define memory 20 Address bus.
  • a block containing a number of migration flags 1.
  • Block corresponding to memory 20 connected to selector 23 in FIG. 1 Block corresponding to memory 20 connected to selector 23 in FIG.
  • the transition flag and sub-rating correspond one-to-one. However, since there are cases where multiple fire signals such as "seed fire” and "fire extinguishment” occur, the transition flag is used.
  • the number of subroutines is smaller than the number of subroutines.
  • One main routine reads all transition flags sequentially and determines whether or not it should be a firing signal. If it is a firing signal, calls a subroutine. After reading all the transition flags. Rereads the transition flags from the beginning.
  • the hardware design of the MA FtS unit is simple, and the cost is low.
  • the protocol of the present invention is as follows: the terminals 13 and 14 in FIG. 1 become IZO.
  • the memory 15 becomes; u memory, the CPU 16 becomes .alpha.PC, and the MARS unit 11 Only the control of the data bus, address bus, and I / O is controlled. The method described below increases the information transmission speed but increases the utility.
  • the SPU 06 is a processor that is in charge of memory management
  • the MPU 26 is a processor that is in charge of processing the contents of messages.
  • This SPU 06 and MP are used as one processor.
  • the parallel processing of messages is also possible, and the total number of processors is three.
  • the method of porting the protocol of the present invention to an algebra is described in "01 serial processing type processor.
  • One main and many subroutine programs are described in 27, 37, 47, and the internal transition flag is set to 28, 38, 48 register or another S. In the memory. ⁇ Realize the instruction cell, including all operations, in an electronic circuit.
  • PCT US 79Z 004 53 Processor for dynamic meter B method.
  • This basic data flow processor is a processor based on transition data. This is the concept of moving the ALU installed outside the
  • PCT / JP 94/00 5 9 5 The method of creating a data flow computer is described in the [Industrial availability and advantages] section of [Gates connecting digital logic circuits].
  • PC TZ JP 94 Z 0 595 5 [ Figure 14] was borrowed and partially modified. A method for realizing the protocol of the present invention in an electronic circuit is described.
  • Reference numeral 94 in FIG. 11 is a "junction gate", which is disclosed in the aforementioned patent.
  • the junction gate 94 is cleared, and the output of the junction gate 94 is used as the level signal of the ignition j of 93. Even if the pulse signal is a level signal, the circuit is the same.
  • 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 aforementioned patent that it can be realized because of the clock.
  • operation j of all the programs and protocols in [Table 4] to [Table 26] is created by a sequential processing circuit or a combinational circuit, and the respective “ignition”, “seeding j”, and “ACK” are input. If an ignition signal is created at the junction gate, the aramograms and protocols are all replaced with electronic circuits.
  • the present invention provides a method for removing a computer from any computer, connecting a computer via an input / output device E to a network, and connecting a computer to any computer. It is a network that realizes message transfer.
  • the runway in Fig. 6 is connected by wired OR. Theoretically, the runway of unlimited length can be laid, but the runway in Fig. 6 can be used in actual industry. Is limited to less than 1 Ocm.
  • 98 is the H level.
  • 99 is the L level. It can be detected as the H level with the R... of 95. If the W... of the left measurement 95 is the H level, and the W ⁇ 'of the right 95 is the H level, , Runway
  • D1 is an open-collector single-wire connection, and if it is not within a few 10 cm, the signal may not be transmitted correctly.
  • the common memory type multi-CPU system, etc. describes the protocol (claim 3) for securing a transmission path in [Table 17] for multiple arbiter seismic units.
  • the CPU is constructed as follows. Encourages the device to pass through the common bus gate. ⁇ ⁇ When prompted, the arbiter system S allows the CPU to pass through the common bus gate when it wins the driveway bidding. Inform .
  • the CPU can read and write to the common memory because the common base is monopolized. (4) When the CPU finishes reading and writing using the common memory, the CPU urges its arbiter device to open so that other arbiter devices can win the bid for securing the runway. List the features of the LAN system.
  • ⁇ Message communication in public speaking can be called from any speaker in the world to any wrapper, but it is common practice to call a short distance or a specific party. The same applies to the network of the invention.
  • the data is transmitted immediately when the distance is short, and the number of transmissions increases when the distance is long.
  • the LAN created by the method of the present invention is a realistic method.
  • Embodiment J of the invention differ in the message speed on each driveway, but can be connected within one LAN.
  • the reason why the WAN runway is packaged is to take into account that the runner and the user of the runway are separated according to the wide area. In other words, the charge for the runway is calculated. It indicates that it is possible to do so.
  • the method of calculating the meal charge on a computer connected to the MARS unit is shown in Table 26-2 J. If a firing signal is issued to start Table 14 in J, a message will be sent. It is possible to obtain charge calculation data while receiving a series of messages.
  • the bottom unit and computer are as described in the LAN feature. To ensure consistency with one MARS unit and only one computer. You may change to a protocol such as C.
  • Nl 4.
  • One server is the same as in Fig. 2, with one V7A N runway that can be connected to the upper runway terminal and 2,744 runways that can be connected to the lower runway terminal.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Small-Scale Networks (AREA)
  • Computer And Data Communications (AREA)

Abstract

Le réseau selon la présente invention a une topologie dans laquelle des canaux de point à point et des canaux d'émission du type bus sont combinés ensemble pour former une structure aborescente. Le dispositif entrée/sortie connecté au réseau local est appelé Serveur Accès et Répétitions Messages (SMAR) (conformément au petit pavé représenté dans la figure). Les nombres de référence (41), (42), (43) indiquent différents réseaux locaux. Pas plus d'un ordinateur ne peut être connecté à une unité SMAR. Les adresses des unités SMAR connectées à une ligne de transmission du type à un bus dans le réseau local satisfont à la condition qu'un certain numéro à un chiffre soit remplacé. Quand l'ordinateur a fourni le message aux unités SMAR, un itinéraire d'acheminement est attribué automatiquement depuis l'adresse de destination sans l'aide de l'ordinateur et l'envoi aussi bien que le retour des messages s'effectuent de manière séquentielle, par commutation de paquets quand l'utilisation d'un système continu pose des problèmes. Quand une adresse est composée de 4 bits x 3 chiffres = 12 bits, 2744 jeux de (41) et (43) et 8.108.731 ordinateurs peuvent être connectés même dans deux couches de réseaux de zones locales, conformément au dessin.
PCT/JP1995/000653 1995-04-04 1995-04-04 Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement WO1996031968A1 (fr)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP53016996A JP3338059B2 (ja) 1995-04-04 1995-04-04 転送識別と調停機構を有するコンピュータ用ネットワークシステム
PCT/JP1995/000653 WO1996031968A1 (fr) 1995-04-04 1995-04-04 Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement
AU20857/95A AU2085795A (en) 1995-04-04 1995-04-04 Computer network system having functions of transport discri mination and arbitration

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP1995/000653 WO1996031968A1 (fr) 1995-04-04 1995-04-04 Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement

Publications (1)

Publication Number Publication Date
WO1996031968A1 true WO1996031968A1 (fr) 1996-10-10

Family

ID=14125827

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP1995/000653 WO1996031968A1 (fr) 1995-04-04 1995-04-04 Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement

Country Status (3)

Country Link
JP (1) JP3338059B2 (fr)
AU (1) AU2085795A (fr)
WO (1) WO1996031968A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003092221A1 (fr) * 2002-04-25 2003-11-06 Mars Technology Institute Co.,Ltd Mecanisme de securite de reseau
WO2007010606A1 (fr) * 2005-07-21 2007-01-25 Mars Technology Institute Co., Ltd. Systeme reseau utilisant une adresse a trois etats

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6190549A (ja) * 1984-10-04 1986-05-08 ジーメンス・アクチエンゲゼルシヤフト 樹枝構成のデータ通信網を有するデータ伝送装置
JPS62232233A (ja) * 1986-04-01 1987-10-12 Ricoh Co Ltd ロ−カルエリア・ネツトワ−ク制御方式
JPS6395751A (ja) * 1986-10-13 1988-04-26 Hitachi Ltd デ−タ伝送システム

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6190549A (ja) * 1984-10-04 1986-05-08 ジーメンス・アクチエンゲゼルシヤフト 樹枝構成のデータ通信網を有するデータ伝送装置
JPS62232233A (ja) * 1986-04-01 1987-10-12 Ricoh Co Ltd ロ−カルエリア・ネツトワ−ク制御方式
JPS6395751A (ja) * 1986-10-13 1988-04-26 Hitachi Ltd デ−タ伝送システム

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003092221A1 (fr) * 2002-04-25 2003-11-06 Mars Technology Institute Co.,Ltd Mecanisme de securite de reseau
WO2007010606A1 (fr) * 2005-07-21 2007-01-25 Mars Technology Institute Co., Ltd. Systeme reseau utilisant une adresse a trois etats

Also Published As

Publication number Publication date
JP3338059B2 (ja) 2002-10-28
AU2085795A (en) 1996-10-23

Similar Documents

Publication Publication Date Title
JP4808514B2 (ja) 大域的非同期ネットワークオンチップシステムにおける通信ノードアーキテクチャ
TWI259373B (en) Method, apparatus, and system for adjusting link width
JP4638216B2 (ja) オンチップバス
CN101593159B (zh) 使用关键度信息来路由高速缓存一致性通信
US8402197B2 (en) Multi-port system and method for routing a data element within an interconnection fabric
US7165131B2 (en) Separating transactions into different virtual channels
US8774199B2 (en) Fibre channel arbitrated loop bufferless switch circuitry to increase bandwidth without significant increase in cost
TWI473012B (zh) 具分散內嵌式交換之多處理運算技術
US20080205421A1 (en) Fibre channel arbitrated loop bufferless switch circuitry to increase bandwidth without significant increase in cost
CN108243103A (zh) 克洛斯织物网络中分配路由协议信息的装置、系统和方法
US20070053350A1 (en) Buffering data packets according to multiple flow control schemes
WO1999039479A1 (fr) Arbitrage reparti dans un bus a duplex integral
JP2008041099A (ja) データ処理装置のための相互接続論理
US20080310458A1 (en) Electronic Device and Method of Communication Resource Allocation
JP2006229992A (ja) 非サイクル有向グラフで接続された構成要素間の通信方法
JPH08505722A (ja) 非サイクル有向グラフに関わる固有アドレス割当て、ノード自己識別及びトポロジーマッピングの方法及び装置
WO1999039461A1 (fr) Arbitrage anticipe dans un bus en duplex integral
JP4759135B2 (ja) デジタル通信ネットワークのための分散スイッチと接続制御構成および方法
JPS6262697A (ja) デ−タ送信制御方式
JPH0787466B2 (ja) 通信端末の挿入制御方法
JPH0629986A (ja) ハイブリッドローカルエリアネットワークおよびデータメッセージ送信方法
US4872163A (en) Contention mechanism for communication networks
WO1996031968A1 (fr) Systeme de reseau d'ordinateurs ayant des fonctions de discrimination et d'arbitrage dans l'acheminement
US20080263248A1 (en) Multi-drop extension for a communication protocol
CN103023608B (zh) 对通信控制器用的时隙表进行编码的系统和方法

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AM AU BB BG BR BY CA CN CZ EE FI GE HU IS JP KG KR KZ LK LR LT LV MD MG MN MX NO NZ PL RO RU SG SI SK TJ TT UA US UZ VN

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): KE MW SD SZ UG AT BE CH DE DK ES FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN ML MR NE SN TD TG

DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase

Ref country code: CA