RU2665225C1 - Information handling unit - Google Patents

Abstract

FIELD: computer equipment.SUBSTANCE: invention relates to the field of computer technology, namely, to information systems. IHU includes machine-to-machine data communication controllers (MDCC). First computing machine (CM) is connected to the MDCC by a local bi-directional Link bus and local bus LM (BUS). Second CM is connected to the MDCC by a local bi-directional Link bus and local bus LM (BUS). Third CM is connected to the MDCC by a local bi-directional Link bus and local bus LM (BUS). Fourth CM is connected to the MDCC by a local bi-directional Link bus and local bus LM (BUS).EFFECT: technical result consists in increasing the performance by speeding up the data reconciliation functions that provide fault tolerance, as well as improving the reliability of inter-machine exchanges.1 cl, 3 dwg

Description

The invention relates to the field of computer technology, namely to information systems, and can be used in the construction of highly reliable computing and control systems designed to solve the problems of controlling and processing information in on-board systems (including spacecraft (SC)).

Known electronic computer (computer) [1], intended for use in control systems and data processing. The computer is a two-level multiprocessor computing system. Computers include a central processor, a graphics controller, one or more input / output devices, one or more programmable signal processors.

The disadvantage of this computer is that it does not have a redundant control processor capable of coordinated control of the calculation process, detect failures and failures, as well as perform autonomous reconfiguration of the system.

Known fail-safe computing system [2], designed to build highly reliable fail-safe on-board control systems.

The system contains four faces. Each face contains a dual-processor microprocessor system operating in parallel, where one central microprocessor (CPU) performs the main computational load, and the second, the input-output processor (I / O), prepares data for the CPU, responds to external interrupts, interacts with peripherals, and implements functions monitoring and diagnostic systems, dual-port random access memory, inter-face exchange equipment, a reconfigurator containing configuration registers, operating mode control registers and Registers sign test and functional serviceability faces logical hardware reliably disconnect the secondary power supply when the output facets of the system, via a distributed majorization control signals from all active faces, cold start circuit and an adapter faces. The system comprises a cycle forming device for synchronizing, in each cycle of the cyclogram, the start of operation of all faces of the system working asynchronously during the cycle.

The disadvantage of this system is the low performance in inter-faceted exchanges, a software method for majorizing data obtained as a result of inter-faceted exchanges, which together leads to a large time cost for providing the failure and fault tolerance functions of the system as a whole.

The closest technical solution to the claimed one is a fault-tolerant computing system with hardware-software implementation of fault tolerance and dynamic reconfiguration functions designed to solve the problems of controlling highly reliable airborne systems [3], which is selected as a prototype.

A fault-tolerant computing system with hardware-software implementation of fault tolerance and dynamic reconfiguration functions contains the first, second, third and fourth computers (VMs) connected by the first, second, third and fourth serial data buses, the first, second, third and fourth secondary power sources ( VIP), the first, second, third and fourth controllers of the inter-machine exchange (KMMO), the first, second, third and fourth controllers of a configuration management (KUK). All VMs in the system receive the same data on serial buses from external subscribers, then these data are coordinated through KMMO controllers between VMs of the working configuration to obtain guaranteed identical copies of data in these VMs. Based on the results of the coordination, faulty VMs can be identified, the type and location of the fault can be determined, and, if necessary, the system is reconfigured using the KUK controller. Data processing is performed on working VMs, then the processing results are again coordinated between the VMs and issued to serial subscribers via serial buses.

The disadvantage of this system is the low bandwidth of the inter-machine communication channels between the individual CMMOs, the need for software control of synchronization and majorization for each data word in the inter-machine communication, as well as the implementation of each of these channels in the form of a common serial interface bus. This together leads to a low rate of inter-machine exchanges and a decrease in the performance of the system as a whole. Also, this channel implementation has a low reliability of inter-machine exchanges, since failure of channel elements of one CMMO can lead to the impossibility of exchanging this channel between all CMMOs.

The technical result is to increase performance by speeding up the performance of data matching functions providing failure and fault tolerance in the information processing unit, as well as increasing the reliability of inter-machine exchanges by introducing additional channels of inter-machine exchanges.

The specified technical result is achieved by the fact that the first, second, third and fourth controllers of the inter-machine data exchange (CMOD) are additionally introduced into the information processing unit. Moreover, the first CMOD is connected by the first group of inputs and outputs to the first group of inputs and outputs of CMOD2, the second group of inputs and outputs with the first group of inputs and outputs of CMOD3 and the third group of inputs and outputs with the first group of inputs and outputs of CMOD4. The second CMOD is connected by the second group of inputs / outputs to the second group of inputs and outputs of the CMOD3 and the third group of inputs and outputs with the second group of the inputs and outputs of the CMOD4. The third CMOD is connected by a third group of inputs and outputs with a third group of inputs and outputs CMOD4. Moreover, the first VM is connected to CMOD1 with a local bi-directional Link interface bus and a local JIM bus (BUS). The second VM is connected to CMOD2 with a local bi-directional Link interface bus and a local LM bus (BUS). The third VM is connected to CMOD3 with a local bidirectional Link interface bus and local LM bus (BUS). The fourth VM is connected to CMOD3 with a local bi-directional Link interface bus and local LM bus (BUS).

In addition, the CMOD contains an output interface unit (SPM), a reception interface unit (USP), a control unit (UE), a data synchronization unit (USD), a data processing unit (UOD), the first, second and third data reception and output nodes ( UPVD-0, UPVD-1, UPVD-2), the first, second, and third buffers of output data (BUF-B0, BUF-V1, BUF-B2), the first second and third buffers of received data (BUF-P0, BUF-P1 , BUF-P2), a buffer of processed data (BUF-OD), a buffer of own data (BUF-SD), and the first group of inputs / outputs of the SPM is the first group of inputs and outputs of the CMOD, the first group in SPM outputs is connected to the first groups of inputs BUF-SD, BUF-B0, BUF-V1, BUF-B2, the second group of outputs SPM is connected to the first group of inputs UU, and the first group of inputs SPM is connected to the first group of outputs UU, the first group of outputs The BUF-SD is connected to the first group of inputs of the UFD, the second group of outputs of the BUF-SD is connected to the second group of inputs of the UF, and the second group of inputs of BUF-SD is connected to the second group of outputs of the UF, and the first group of outputs of BUF-V0 is connected to the first group of inputs of the air intake -0, the second group of outputs BUF-V0 is connected to the third group UU inputs, and the second group of BUF-V0 inputs is connected to the third group of UU outputs, the first group of UPVD-0 inputs and outputs is the fourth CMOD input-output group, the first UPVD-0 output group is connected to the first group of BUF-P0 inputs, the second the group of outputs UPVD-0 is connected to the fourth group of inputs of the UU, and the second group of inputs of UPUD-0 is connected to the fourth group of the outputs of UU, the first group of outputs of BUF-PO is connected to the second group of inputs of UOD, the second group of outputs of BUF-PO is connected to the fifth group inputs UU, and the second group of inputs BU -PO is connected to the fifth group of UU outputs, the first group of BUF-V1 outputs connected to the first group of UPVD-1 inputs, the second group of BUF-V1 outputs connected to the seventh group of UU inputs, and the second group of BUF-B1 inputs connected to the seventh group of outputs UU, and the first group of inputs and outputs of UPVD-1 is the fifth group of inputs and outputs of KMOD, the first group of outputs of UPVD-1 is connected to the first group of inputs of BUF-P1, the second group of outputs of UPVD-1 is connected to the eighth group of inputs of UU, and the second group inputs UPVD-1 is connected to the eighth group of outputs UU, moreover, the first group of outputs BUF-P1 is connected to the third group of inputs of the UOF, the second group of outputs BUF-P1 is connected to the ninth group of inputs of the UU, and the second group of inputs BUF-P1 is connected to the ninth group of outputs of the UU, and the first group of outputs BUF-B2 is connected to the first group of inputs UPVD-2, the second group of outputs BUF-B2 is connected to the tenth group of inputs UU, and the second group of inputs BUF-B2 is connected to the tenth group of outputs UU, the first group of inputs and outputs UPVD-2 is the sixth group of inputs and outputs KMOD , the first group of outputs UPVD-2 it is connected to the first group of inputs of BUF-P2, the second group of outputs of UPVD-2 is connected to the eleventh group of inputs of the control unit, and the second group of inputs of UPVD-2 is connected to the eleventh group of outputs of the control unit, the first group of outputs of BUF-P2 is connected to the fourth group of inputs of the control unit, the second group of outputs BUF-P2 is connected to the twelfth group of inputs UU, and the second group of inputs BUF-P2 is connected to the twelfth group of outputs UU, the first group of outputs UOD connected to the first group of inputs BUF-OD, the second group of outputs UOD connected to the sixth group of inputs UU, a p the first group of UFD inputs is connected to the sixth group of UF outputs, the first group of BUF-OD outputs connected to the first group of USP inputs, the second group of BUF-OD outputs connected to the thirteenth group of UU inputs, and the second group of BUF-OD inputs connected to the thirteenth group of outputs UE, and the first group of inputs and outputs of the USP is the second group of inputs and outputs of the CMOD, the first group of outputs of the USP is connected to the fourteenth group of inputs of the UM, and the second group of inputs of the USP is connected to the fourteenth group of outputs of the UM, and the first group of inputs and outputs s UU is the third group of inputs and outputs of the CMOD, the fifteenth group of outputs of the UU is connected to the first group of inputs of the DRC, and the fifteenth group of inputs of the UU is connected to the first group of outputs of the DRC.

The essence of the claimed invention is illustrated by the drawings shown in FIG. 1 - FIG. 3, where:

- in FIG. 1 shows the structure of an information processing unit;

- in FIG. 2 presents the cycle of the program with the introduction of CMOD,

where T1 is the cycle of the program, T2 is the target operation of the BOI, T3 is the synchronization and coordination of data between the VM through the CMMO, T4 is the synchronization and coordination of data between the VM through the CMMO and the CMOD;

- in FIG. 3 shows the structure of the inter-machine data controller.

These advantages of the claimed system over the prototype are achieved due to the fact that in the information processing unit containing the first 1, second 2, third 3 and fourth 4 VMs connected by the first 5, second 6, third 7 and fourth 8 serial data buses containing the first 9 , second 10, third 11 and fourth 12 VIPs, first 13, second 14, third 15 and fourth 16 KMMO, first 17, second 18, third 19 and fourth 20 KUK, the first 21 groups of outputs of which are connected to the first groups of inputs of the first, second , the third and fourth VM, the second 22 groups of inputs of which connected to the first group of inputs of the system, the second 23, third 24, fourth 25, fifth 26 groups of inputs which are connected to the first groups of inputs KMMO1 (13) and KUK1 (17), KMMO2 (14) and KUK2 (18), KMMO3 (15) and KUK3 (19), KMMO4 (16) and KUK4 (20), respectively, the second 27 groups of inputs of which are connected to the first groups of outputs of the VM (1, 2, 3, 4), the local bi-directional highway 28 of which is connected to local bi-directional highways KMMO (13, 14, 15, 16), the first 29 outputs of which are connected to the first inputs of the controllers KUK (17, 18, 19, 20), the second 30 groups of outputs of which are connected with the first groups of inputs of the VIP (9, 10, 11, 12), respectively, the groups of outputs 31 of which are connected to the third groups of inputs of the VM (1, 2, 3, 4), and the first 32 group of outputs of the first 13 CMMO is connected to the third groups of inputs of the second 14, third 15 and fourth 16 KMMO and first 17, second 18, third 19 and fourth KUK 20, the first 33 output group of the second 14 KMMO is connected to the fourth groups of inputs of the first 13, third 15 and the fourth KMMO 16 and the first 17, second 18, third 19 and fourth 20 KUK, the first 34 group of outputs of the third 15 KMMO is connected with the fifth groups the inputs of the first 13, second 14 and fourth 16 KMMO and the first 17, second 18, third 19 and fourth 20 KUK, the first 35 group of outputs of the fourth 16 KMMO is connected to six groups of inputs of the first 13, second 14 and third 15 KMMO and the first 17, second 18, the third 19 and the fourth 20 CUC, and the sixth 36 group of inputs of the system is connected to the second groups of inputs of the first 9, second 10, third 11 and fourth 12 VIPs, the first inputs of which are connected to the seventh 37 group of inputs of the system, the eighth 38 group of inputs of which is connected with second inputs of the first 9, second 10, tr this is 11 and the fourth 12 VIP, and the second 39 groups of outputs KMMO (13, 14, 15, 16) are connected to the fourth groups of inputs of the VM (1, 2, 3, 4) additionally introduced the first 21, second 22, third 23 and fourth 24 CMOD, two local buses of the LINK 40 interface, 41 of which are connected to the first, second, third and fourth VMs (1, 2, 3, 4), the local highway 28 of which is connected to the first, second, third and fourth CMOD (21, 22, 23, 24), and the first CMOD 21 is connected by a bidirectional Space Wire 42 interface to the second CMOD 22, the bi-directional Space Wire 45 interface of which is connected to the third CMOD 23, two the bidirectional SpaceWire 47 interface of which is connected to the fourth CMOD 24, the bidirectional SpaceWire 44 interface of which is connected to the first CMOD 21, the bidirectional SpaceWire 43 interface of which is connected to the third CMOD 23, the second CMOD 22 is connected by the bidirectional SpaceWire 46 interface to the fourth CMOD 24.

The block diagram of the information processing unit is shown in FIG. one.

The information processing unit operates as follows.

The information processing unit includes three VMs of the working configuration and one VM in the cold reserve, the general scheme for countering failures and failures is as follows. Each VM of the working configuration solves the same target task of special software (STR), using the fault tolerance software (VET), the decision results are transferred to other VMs of the working configuration and receives from them their copies of the results of the decision, and then according to their own and received copies of the solution determines the correct result.

In a two-machine operating configuration, the correct solution is determined on the basis of two results and, if such a determination is impossible (there is no possibility of identifying a faulty VM using the results of the solution), then, depending on the available time resources, either a second solution of the STR problem is carried out, or a transition to an intermediate channel testing followed a second solution with a positive test result or a transition to a single-channel configuration based on a working channel of the system. If it is not possible to identify a faulty VM, a transition is made to a safe shutdown of the system.

In the process of all actions performed by the system through VET, continuous and end-to-end functional diagnostics is carried out, which detects the occurrence of malfunctions and identifies them by the place of occurrence and by type: failure, software failure or failure. When identifying a failure, its accounting and analysis of the consequences is carried out. If the consequences of the failure do not require the intervention of the STR program, then they are eliminated automatically. If support from the STR is necessary, then diagnostic information is generated, on the basis of which the STR makes the required decision. Failures that do not require the intervention of STRs include, for example, a failure in the inter-machine transmission of information, which is parried due to the input of information redundancy. A malfunction requiring support of the software application is, for example, a malfunction in the issuance of information to the subscriber.

Providing these program functions in the system requires a significant amount of time. Synchronization and coordination of data is performed on each cycle of operation of the STR in a strictly allocated time interval for these tasks within the cycle (see Fig. 2). During the synchronization and coordination functions, the information processing unit does not perform its target work. Increasing the speed of synchronization and data matching functions allows to reduce the execution time of these functions and increase the time allocated to the target operation of the system (see Fig. 2).

Improving performance when performing these functions is ensured by the introduction of CMOD.

The information transmitted in the information processing unit through the inter-machine channels is divided into two types:

1) short messages transmitted between VMs through KMMO for data exchange between VMs, synchronization, as well as system reconfiguration through QMS;

2) data arrays transmitted between VMs through CMOD with hardware synchronization.

Thus, in comparison with the prototype, the inter-machine exchange in the BOI is performed through the CMMO - for the transmission of short messages and through the CMOD - for the transfer of large data arrays.

Using KMMO for transferring large amounts of data between VMs leads to large time costs, since for transferring each word of data, it is necessary to perform a separate exchange on the inter-machine exchange bus, also on each exchange it is necessary to carry out program control of synchronization and majorization. The use of CMOD is optimal for the transfer of large amounts of data, since synchronization and majorization is performed in hardware for each word. Using CMOD to transmit short messages is also possible, but at the same time, it is necessary to spend time on each exchange for preparing data and setting up the entire chain of involved nodes for transmitting information.

Improving the reliability of inter-machine exchanges is achieved due to the fact that:

1) an additional mechanism for exchanging data between VMs through CMOD is introduced, i.e. inter-machine data exchange in the BOI is reserved;

2) the fully-connected topology of the links between the CMOD of all VMs in the BOI using serial channels of the SpaceWire interface was used. Failure of one of the SpaceWire channels will not lead to the impossibility of participation of this VM in the inter-machine exchange on other SpaceWire channels with the rest of the VMs.

CMOD is designed to coordinate data between VMs in BOIs. In addition to the transmission and reception of data, the CMOD synchronizes the received data and processes them in order to determine reliable data using the 2 out of 3 (majorization) voting function for each bit of data words received from three VMs.

Each VM in the BOI is connected to its own CMOD, which receives data from this VM through one LINK type interface (Link-IN) and sends data to this VM through another LINK type interface (Link-OUT). At the same time, the CMOD of each VM is connected to the CMOD of the remaining VMs by separate interfaces of the SpaceWire type (SWO, SW1, SW2). Also, the CMOD of each VM for reading and writing its registers is connected to the local LM bus (BUS), on which the CMOD is a slave device similar to static memory.

According to the structural diagram presented in figure 3, CMOD contains:

1 - node data synchronization (DRC);

2 - control unit (UE);

3 - node interface issuance (SPM);

4 - buffer of own data of the FIFO type (BUF-SD);

5 - buffer of output data of the FIFO type of the SW0 interface (BUF-V0);

6 - node receiving and issuing data interface SW0 (UPVD-0);

7 - buffer of received data of the FIFO type of the SW0 interface (BUF-P0);

8 - buffer of output data of the FIFO type of the SW1 interface (BUF-V1);

9 - node receiving and issuing data interface SW1 (UPVD-1);

10 - buffer of received data of the FIFO type of the SW1 interface (BUF-P1);

11 - buffer of output data of the FIFO type of the SW2 interface (BUF-B2);

12 - node receiving and issuing data interface SW2 (UPVD-2);

13 - a buffer of received data of the FIFO type of the SW2 interface (BUF-P2);

14 - buffer of processed data of the FIFO type (BUF-OD);

15 - node pairing reception (USP);

16 - data processing unit (UOD);

17 - local bus LM (BUS);

18 - LINK type data input interface;

19 - LINK type data output interface;

20 is a first interface such as SpaceWire;

21 is a second interface such as SpaceWire;

22 is the third interface such as SpaceWire.

In order to perform data reconciliation between several VMs, each of the VMs participating in the exchange procedure must perform the following actions:

1. write in the register of CMOD the numbers of the VMs participating in the exchange procedure;

2. write in the register of the CMOD the number of words of data of the exchange procedure;

3. write to the CMOD registers the timeouts of receiving and issuing data words;

4. allow VM to receive data from the CMOD via the LINK type interface;

5. allow the VM to transmit data to the CMOD via an interface of type LINK;

6. write in the register of the CMOD a sign of the beginning of the exchange procedure;

7. expect in the register of CMOD a sign of the end of the exchange procedure;

8. read the signs of detected errors in the CMOD registers and reset them to prepare for the next exchange procedure.

After the sign of the beginning of the next exchange procedure is established, the sign of the end of the previous exchange procedure will be reset and all the CMOD nodes will begin to function as follows.

SPM via the Link-IN interface receives data from the VM intended for transmission to other VMs, and depending on the VM numbers involved in the exchange recorded earlier in the CMOD register, writes this data to certain buffers, such as BUF-SD, and / or BUF-V0, and / or BUF-B1, and / or BUF-B2.

If there is a generalized readiness signal BUF-SD, BUF-V0, BUF-V1, BUF-B2 for recording, which is issued to the SPM from the UE and is formed on the basis of the readiness signals for writing individual buffers, a record signal is issued from the SPM to the UU, according to which The UE generates write signals to individual buffers.

In this case, the BUF-SD is used to save data in the case when at least one of the VM numbers participating in the exchange coincides with its own VM number. And if at least one of the VM numbers participating in the exchange does not coincide with its own VM number, then the data is also recorded in the BUF-B0, and / or the BUF-B1, and / or the BUF-B2 in accordance with which of interfaces SW0, SW1, SW2 in the CMOD of this VM are used to connect it with the CMOD of other VMs participating in the exchange.

Each of the BUF-V0, BUF-V1, BUF-B2 gives the read -iness signal to the control unit, which is transmitted to UPVD-0, UPVD-1, UPVD-2, respectively, and in response to this signal, each of UPVD-0, UPVD -1, UPVD-2, if necessary, gives a read signal to the control unit, which is transmitted to BUF-V0, BUF-V1, BUF-B2, respectively. Each of the UPVD-0, UPVD-1, UPVD-2, through the SW0, SW1, SW2 interface, respectively, transmits to the CMOD another VM participating in the exchange procedure, data read from the BUF-V0, BUF-V1, BUF-B2, respectively , and receives data from the CMOD of this VM, which are recorded in the BUF-PO, BUF-P1, BUF-P2, respectively.

At the same time, each of the BUF-P0, BUF-P1, BUF-P2 generates a ready-to-write signal to the UU, which is transmitted to UPVD-0, UPVD-1, UPVD-2, respectively, and in response to this signal, each of UPVD-0 , UPVD-1, UPVD-2, if necessary, issues a recording signal to the UU, which is transmitted to the BUF-P0, BUF-P1, BUF-P2, respectively.

DRM, depending on the VM numbers involved in the exchange, recorded earlier in the CMOD register, reads from certain buffers, such as BUF-SD, and / or BUF-P0, and / or BUF-P1, and / or BUF-P2, data to be processed in order to determine reliable data, and records reliable data in the BUF-OD.

If there is a generalized readiness signal BUF-SD, BUF-P0, BUF-P1, BUF-P2 for reading, which is issued to the UOF from the UE and is formed on the basis of the readiness signals for reading individual buffers, a read signal is issued from the UOF to the UU, according to which The UE generates read signals into separate buffers.

Then, using the 2 out of 3 (majorization) voting function, for each bit of the original data words received from three VMs, it receives a valid data word and, if necessary, generates signals in the UM indicating that the valid data word does not coincide with the original data words, according to which UU signs of detected errors are established.

And then, if there is a signal that the BUF-OD is ready for recording, which is transmitted to the UOF through the UE, a recording signal is issued from the UOD to the UU that is broadcast to the UUF-OD.

USP reads from the BUF-OD reliable data and transmits them to the VM via the Link-OUT interface.

If there is a readiness signal for the BUF-OD for reading, which is transmitted to the USP through the UE, from the USP to the UU, a read signal is issued, which is transmitted to the BUF-OD.

The DRC counts the time during which there is no generalized readiness signal BUF-SD, BUF-V0, BUF-V1, BUF-B2 for recording from the control unit, but at least one of the readiness signals for recording individual buffers is present. When this time coincides with the timeout for the output of data words, signals are generated that indicate a violation of the time for issuing data words from individual buffers and arriving at the control unit, which are used to establish the signs of detected errors, which subsequently mask the signals coming from these buffers and the signals going to these buffers.

Similarly, the DRC counts the time during which there is no generalized BUF-SD, BUF-P0, BUF-P1, BUF-P2 readiness signal from the UU, but at least one of the readiness signals for reading individual buffers is present. If this time coincides with the timeout for the output of data words, signals are generated that indicate a violation of the time for receiving data words in separate buffers and arriving at the control unit, by which the signs of detected errors are established, which subsequently mask the signals coming from these buffers, and the signals going to these buffers.

The UM contains the control registers and the state of the CMOD and transmits the necessary control and status signals between the other nodes of the CMOD, while the data is transmitted directly between the other nodes of the CMOD.

After the number of words of the data of the exchange procedure is read from the BUF-OD, the termination sign will be set and the sign of the beginning of the exchange procedure will be reset.

To perform data transfer and coordinated reception of data between VMs, it is necessary for all VMs to configure the CMOD registers for the selected operating mode, enable the reception and output of data via LINK-type interfaces, start the data exchange mechanism in the CMOD and wait for its operation to complete. At the same time, the data from this VM, depending on the settings of the registers of the CMOD nodes, will be transferred to certain VMs, and the data from them will be coordinated and accepted into this VM.

The VM to which data will be transmitted and from which data will be received, as a rule, but not necessarily, includes this VM, in which for these purposes the CMF uses BUF-SD without the need to implement another interface, such as SpaceWire, which is closed to itself .

Despite the fact that the 2 out of 3 (majorization) voting function is always performed for three arguments, the CMOD allows exchange between two VMs. For this, it is necessary to specify the data of the first VM for two arguments, and to specify the data of the second VM for the third argument. In this case, if the data do not match, the CMOD will record the corresponding sign, and the result of the 2 out of 3 voting function (majorization) will always be determined by the data of the first VM.

Also, in the case of two VMs, it is possible for all three arguments to specify data from another VM. Moreover, none of the signs of mismatch will be recorded, and the result of the voting function 2 of 3 (majorization) will always be equal to the data of another VM. That is, in this mode, the CMOD can be used as a simple controller for exchanging data between two VMs.

The introduction of one CMOD for each VM in the BOI increases the speed, since it allows you to free some of the computing resources of the VM from the task of matching data between the VMs and use these resources to solve the target task, and also reduces the total time spent matching the data between the VMs account hardware implementation of the exchange procedure. The presence in CMOD of several-word data buffers between all nodes of the data path and the possibility of programming timeout times over a wide range also improves performance, since it reduces the requirements for the synchronization of the computing process in different VMs and allows smoothing and, as a whole, increasing the rate of coordination data using CMOD as compared to word-by-word coordination in CMMO. In this case, the highest performance is achieved by using the direct memory access mechanism in the VM when receiving and issuing data via unidirectional LINK-type interfaces.

The presence of separate bi-directional SpaceWire-type interfaces for connecting CMOD of different VMs to each other, compared with the common buses between the transmitter and receivers of the CMMO, can improve the reliability of inter-machine data exchanges in the BOI, since if one of the SpaceWire-type interfaces between the CMOD of the two VMs fails, only the opportunity their simultaneous inclusion in the working configuration.

Information sources

1. RF patent No. 2344472 "Electronic computing machine."

2. RF patent No. 2439674 "Method of forming a fault-tolerant computing system and fault-tolerant computing system."

3. RF patent No. 2455681 "Failsafe computing system with hardware-software implementation of the functions of fault tolerance and dynamic reconfiguration."

Claims (2)

1. The information processing unit (BOI), containing the first, second, third and fourth computers (VMs) connected by the first, second, third and fourth serial data buses, the first, second, third and fourth secondary power sources (VIP), the first , the second, third and fourth inter-machine exchange controllers (KMMO), the first, second, third and fourth configuration management controllers (KUK), the first groups of outputs of which are connected to the first groups of inputs of the first, second, third and fourth VMs, the second groups of inputs of which connected to the first group of inputs of the system, the second, third, fourth, fifth groups of inputs of which are connected to the first groups of inputs KMMO1 and KUK1, KMMO2 and KUK2, KMMO3 and KUK3, KMMO4 and KUK4, respectively, the second groups of inputs of which are connected to the first groups of outputs of the VM, the local bi-directional highway of which is connected to the local bi-directional highways of KMMO, the first outputs of which are connected to the first inputs of the KUK controllers, the second output groups of which are connected to the first groups of VIP inputs, respectively, the output groups in which are connected to the third groups of inputs of the VM, and the first group of outputs of the first CMMO is connected to the third groups of inputs of the second, third and fourth CMMO and the first, second, third and fourth CMC, the first group of outputs of the second CMMO is connected to the fourth groups of inputs of the first, third and the fourth CMMO and the first, second, third and fourth CMC, the first group of outputs of the third CMMO is connected to the fifth groups of inputs of the first, second and fourth CMMO and the first, second, third and fourth CMC, the first group of outputs the fourth CMMO is connected to the sixth groups of inputs of the first, second and third CMMOs and the first, second, third and fourth CCCs, and the sixth group of inputs of the system is connected to the second groups of inputs of the first, second, third and fourth VIPs, the first inputs of which are connected to the seventh group of inputs system, the eighth group of inputs of which is connected to the second inputs of the first, second, third and fourth VIPs, and the second groups of outputs of the CMMO are connected to the fourth groups of inputs of the VM, characterized in that, in order to increase the speed The mechanisms of information coordination of data between VM backup channels, the first, second, third and fourth inter-machine data exchange controllers (CMOD) are additionally introduced into the BOI, the first CMOD connected to the first group of inputs and outputs with the first group of inputs and outputs of CMOD2, the second group of inputs outputs with the first group of inputs and outputs KMOD3 and the third group of inputs and outputs with the first group of inputs and outputs KMOD4, the second CMOD is connected by the second group of inputs and outputs with the second group of inputs and outputs KMOD3 and the third group I-inputs with a second group of inputs and outputs KMOD4, the third CMOD is connected by a third group of inputs and outputs with a third group of inputs and outputs KMOD4, the first VM connected to CMOD1 local bi-directional interface bus Link and the local bus LM (BUS), the second VM is connected with CMOD2 a local bidirectional Link interface bus and a local LM bus (BUS), the third VM is connected to CMOD3 a local bi-directional Link interface bus and a local LM bus (BUS), the fourth VM is connected to CMOD3 a local bi-directional Link interface bus and a local bus No LM (BUS). 2. BOIs according to claim 1, characterized in that the CMOD comprises an issuance interface unit (UMC), a reception interface unit (USP), a control unit (UU), a data synchronization unit (USD), a data processing unit (UOD), the first, the second and third nodes of data reception and output (UPVD-0, UPVD-1, UPVD-2), the first, second and third buffers of the output data (BUF-B0, BUF-V1, BUF-B2), the first second and third buffers received data (BUF-P0, BUF-P1, BUF-P2), a buffer of processed data (BUF-OD), a buffer of own data (BUF-SD), and the first group of inputs / outputs of the SPM is the first group of inputs and outputs in CMOD, the first group of SPM outputs is connected to the first groups of inputs BUF-SD, BUF-B0, BUF-V1, BUF-B2, the second group of SPM outputs is connected to the first group of UU inputs, and the first group of SPM inputs is connected to the first group of UU outputs moreover, the first group of outputs of the BUF-SD is connected to the first group of inputs of the UOF, the second group of outputs of the BUF-SD is connected to the second group of inputs of the UU, and the second group of inputs of BUF-SD is connected to the second group of outputs of the UU, and the first group of outputs of BUF-B0 is connected with the first group of inputs UPVD-0, the second group of outputs BUF-B0 it is connected to the third group of inputs of the control unit, and the second group of inputs of the BUF-B0 is connected to the third group of outputs of the control unit, the first group of inputs and outputs of the UPVD-0 is the fourth group of inputs and outputs of the CMOD, the first group of outputs of the UPVD-0 is connected to the first group of inputs of the BUF -P0, the second group of outputs UPVD-0 is connected to the fourth group of inputs of the control unit, and the second group of inputs of UPVD-0 is connected to the fourth group of outputs of the control unit, the first group of outputs BUF-P0 is connected to the second group of inputs of the control unit, the second group of outputs BUF-P0 connected to the fifth group of inputs UU, and T the first group of inputs BUF-P0 is connected to the fifth group of outputs of the control unit, the first group of outputs of the BUF-V1 connected to the first group of inputs of the UPVD-1, the second group of outputs of the BUF-B1 connected to the seventh group of inputs of the control unit, and the second group of inputs of BUF-B1 is connected with the seventh group of outputs UU, and the first group of inputs and outputs UPVD-1 is the fifth group of inputs and outputs KMOD, the first group of outputs UPVD-1 is connected to the first group of inputs BUF-P1, the second group of outputs UPVD-1 is connected to the eighth group of inputs UU , and the second group of inputs UPVD-1 is connected to the eighth the first group of outputs UU, and the first group of outputs BUF-P1 is connected to the third group of inputs UOD, the second group of outputs BUF-P1 is connected to the ninth group of inputs UU, and the second group of inputs BUF-P1 is connected to the ninth group of outputs UU, and the first group of outputs BUF-B2 is connected to the first group of inputs of UPVD-2, the second group of outputs BUF-B2 is connected to the tenth group of inputs of UU, and the second group of inputs BUF-B2 is connected to the tenth group of outputs of UU, the first group of inputs and outputs of UPVD-2 is the sixth CMOD I / O group, the first group the outputs of UPVD-2 is connected to the first group of inputs of BUF-P2, the second group of outputs of UPVD-2 is connected to the eleventh group of inputs of UU, and the second group of inputs of UPVD-2 is connected to the eleventh group of outputs of UU, the first group of outputs of BUF-P2 is connected to the fourth group of inputs UOD, the second group of outputs BUF-P2 is connected to the twelfth group of inputs UU, and the second group of inputs BUF-P2 is connected to the twelfth group of outputs UU, the first group of outputs UOD connected to the first group of inputs BUF-OD, the second group of outputs UOD connected with the sixth a group of UU inputs, and the fifth group of UOD inputs is connected to the sixth group of UU outputs, the first group of BUF-OD outputs connected to the first group of USP inputs, the second group of BUF-OD outputs connected to the thirteenth group of UU inputs, and the second group of BUF-OD inputs connected to the thirteenth group of AC outputs, the first group of inputs and outputs of the USP is the second group of inputs and outputs of the CMOD, the first group of outputs of the USP is connected to the fourteenth group of inputs of the AC, and the second group of inputs of the USP is connected to the fourteenth group of outputs of the AC I group of inputs and outputs of the control unit is the third group of inputs and outputs of the CMOD, the fifteenth group of outputs of the control unit is connected to the first group of inputs of the control panel, and the fifteenth group of inputs of control unit is connected to the first group of control panel outputs.

Images