A distributed multi-core system synchronization mechanism programmable primitive information processing method, device and electronic device

By setting registers and performing registration, decrementing, and deregistration operations in the synchronization mechanism management unit, a hardware-based synchronization mechanism is implemented, which solves the problem of low synchronization efficiency between cores and improves the synchronization efficiency and data consistency of distributed systems.

CN118193198BActive Publication Date: 2024-11-19TSINGHUA UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410297989.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-15
Publication Date
2024-11-19
Estimated Expiration
2044-03-15

AI Technical Summary

Technical Problem

In many-core chips, the existing technology requires software to coordinate the data transmission and synchronization mechanisms between cores, which leads to low efficiency and affects the overall operating performance.

Method used

By setting registers and performing registration, decrement, and deregistration operations in the synchronization mechanism management unit, signaling messages are generated to achieve information synchronization between multiple computing cores. The hardware-based synchronization mechanism hides the synchronization signaling and improves synchronization efficiency.

Benefits of technology

It improves the synchronization efficiency between units in a distributed system, ensures data consistency in the computing core cluster, and enhances control efficiency and coverage.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118193198B_ABST
    Figure CN118193198B_ABST
Patent Text Reader

Abstract

The present disclosure relates to a method, device and electronic device for processing information of programmable primitives of synchronization mechanism of distributed multi-core system. The method is applied to at least one synchronization mechanism management unit connected to a computing core cluster, including: performing registration operation on multiple registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and writing a first value to each register; in response to the computing core indicated by any register completing the respective first command to be synchronized, performing a self-decrement operation on the first value in the register; in response to the value of any register reaching a preset second value, canceling the register; waiting for all the multiple registers in the synchronization mechanism management unit to be cancelled, and generating a signaling message. The embodiment of the present disclosure provides a method for implementing a synchronization mechanism in hardware, which can hide the synchronization signaling, improve the synchronization efficiency between the units of the distributed system, and achieve the consistency of the data of the distributed computing core cluster.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of computers, and in particular to a method, device and electronic device for processing programmable primitive information of a distributed multi-core system synchronization mechanism. Background Art

[0002] With the development of artificial intelligence technology, users have higher and higher requirements for chip processing capabilities. Due to the limited processing power of single-core chips, the use of multi-core chips is becoming more and more widespread. Distributing computing tasks to a combination of multiple cores for parallel execution can improve processing power. In a multi-core chip, it is necessary not only to execute the computing tasks assigned to each core, but also to transmit data between cores. The synchronization mechanism of intra-core computing tasks and inter-core communication tasks in a multi-core chip will affect the overall operating performance of the multi-core chip. In related technologies, the synchronization mechanism requires software to cooperate with hardware or to be implemented by software alone. Software implementation requires communication transmission, which is not efficient enough. Summary of the invention

[0003] In view of this, the present disclosure proposes a method, device and electronic device for processing programmable primitive information of a distributed many-core system synchronization mechanism.

[0004] According to one aspect of the present disclosure, there is provided an information processing method, which is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores, and the method comprises: the synchronization mechanism management unit performs a registration operation on multiple registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and writes a first value to each register; in response to the computing core indicated by any register completing its respective first command to be synchronized, performs a self-decrement operation on the first value in the register; in response to the value of any register reaching a preset second value, cancels the register; waits for all the multiple registers in the synchronization mechanism management unit to be canceled, and generates a signaling message, wherein the signaling message is used to represent the completion of the synchronization of multiple first commands.

[0005] In one possible implementation, the computing core cluster includes multiple computing cores located in different processors, each processor includes a synchronization mechanism management unit and multiple computing cores, wherein the synchronization mechanism management unit located in the same processor communicates with the computing cores through an internal bus, and the synchronization mechanism management unit located in different processors communicates with the computing cores through an external bus.

[0006] In one possible implementation, the synchronization mechanism management unit also includes at least one sequential container encapsulating a dynamic array, and the dynamic array of the sequential container is used to record the registration order of multiple registers. The method also includes: when there is a dependency relationship between multiple first commands to be synchronized, the sequential container records the registration order of multiple registers according to the dependency relationship between the multiple first commands to be synchronized; in response to the computing core indicated by the current register completing the first command to be synchronized, and the register whose registration order is before the register completes the cancellation, a self-decrement operation is performed on the first value in the register.

[0007] In a possible implementation, performing a decrement operation on the first value in the register includes: configuring an interrupt handling subroutine for each sequential container; in response to the sequential container receiving a vector feedback signal carrying a register identifier, triggering the interrupt handling subroutine to perform a decrement operation on the first value in the register indicated by the register identifier.

[0008] In a possible implementation, the method further includes: in response to any computing core receiving a signaling message returned by the synchronization mechanism management unit, the computing core executes a second command of the next timing.

[0009] In one possible implementation, waiting for multiple registers in a synchronization mechanism management unit to be completely deregistered and generating a signaling message includes: the synchronization mechanism management unit polls and checks the multiple registers in the synchronization mechanism management unit according to a preset period to obtain an inspection result; and generating a signaling message when the inspection result indicates that the multiple registers in the synchronization mechanism management unit are completely deregistered.

[0010] In one possible implementation, the first command to be synchronized includes multiple sub-commands, and in response to the computing core indicated by any register completing the respective first command to be synchronized, a decrement operation is performed on the first value in the register, including: in response to the computing core indicated by the register completing any sub-command, reading the current value from the register, the current value is less than or equal to the first value and greater than the second value; and performing a decrement operation on the current value in the register.

[0011] According to another aspect of the present disclosure, an information processing device is provided, which is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores, and the device includes: a registration module, which is used for the synchronization mechanism management unit to perform a registration operation on multiple registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and write a first value to each register; a self-decrement module, which is used to perform a self-decrement operation on the first value in the register in response to the computing core indicated by any register completing its respective first command to be synchronized; a cancellation module, which is used to cancel the register in response to the value of any register reaching a preset second value; and a generation module, which is used to wait for all the multiple registers in the synchronization mechanism management unit to be cancelled and generate a signaling message, and the signaling message is used to characterize the synchronization of multiple first commands.

[0012] In one possible implementation, the computing core cluster includes multiple computing cores located in different processors, each processor includes a synchronization mechanism management unit and multiple computing cores, wherein the synchronization mechanism management unit located in the same processor communicates with the computing cores through an internal bus, and the synchronization mechanism management unit located in different processors communicates with the computing cores through an external bus.

[0013] In one possible implementation, the synchronization mechanism management unit also includes at least one sequential container encapsulating a dynamic array, and the dynamic array of the sequential container is used to record the registration order of multiple registers. The device also includes a storage module, which is used to: when there is a dependency relationship between multiple first commands to be synchronized, the sequential container records the registration order of multiple registers according to the dependency relationship between the multiple first commands to be synchronized; in response to the computing core indicated by the current register completing the first command to be synchronized, and the register whose registration order is before the register completes the cancellation, and performs a self-decrement operation on the first value in the register.

[0014] In this way, the storage module is used for dynamic array encapsulation of sequential containers, and the registration execution operation in the synchronization mechanism management unit implements the storage operation.

[0015] In a possible implementation, performing a decrement operation on the first value in the register includes: configuring an interrupt handling subroutine for each sequential container; in response to the sequential container receiving a vector feedback signal carrying a register identifier, triggering the interrupt handling subroutine to perform a decrement operation on the first value in the register indicated by the register identifier.

[0016] In a possible implementation, the device further includes a response module, configured to: in response to any computing core receiving a signaling message returned by the synchronization mechanism management unit, the computing core executes a second command of the next timing.

[0017] In a possible implementation, the generation module is used to: the synchronization mechanism management unit polls and checks multiple registers in the synchronization mechanism management unit according to a preset period to obtain an inspection result; when the inspection result indicates that all multiple registers in the synchronization mechanism management unit are cancelled, generate a signaling message.

[0018] In one possible implementation, the self-decrement module is used to: when the first command to be synchronized includes multiple sub-commands, in response to the computing core indicated by the register completing any sub-command, read a current value from the register, the current value is less than or equal to the first value and greater than the second value; and perform a subtraction operation on the current value in the register.

[0019] According to another aspect of the present disclosure, a many-core chip is provided, the many-core chip comprising a synchronization mechanism management unit and a plurality of computing cores, the synchronization mechanism management unit connecting each computing core, the synchronization mechanism management unit executing the information processing method as described above.

[0020] According to another aspect of the present disclosure, an electronic device is provided, wherein the electronic device includes at least one multi-core chip as described above.

[0021] According to another aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which computer program instructions are stored, wherein the computer program instructions implement the above method when executed by a processor.

[0022] According to another aspect of the present disclosure, a computer program product is provided, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above method.

[0023] The information processing method provided by the embodiment of the present disclosure is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores. The method includes: the synchronization mechanism management unit performs a registration operation on multiple registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and writes a first value to each register; in response to the computing core indicated by any register completing its respective first command to be synchronized, performs a self-decrement operation on the first value in the register; in response to the value of any register reaching a preset second value, cancels the register; waits for all the multiple registers in the synchronization mechanism management unit to be canceled, and generates a signaling message, and the signaling message is used to represent the completion of the synchronization of multiple first commands.

[0024] In this way, the embodiments of the present disclosure provide a method for implementing a synchronization mechanism in hardware, which can hide the synchronization signaling, improve the synchronization efficiency between the units of the distributed system (such as computing cores), facilitate subsequent and event message collaboration, improve control efficiency and coverage, and achieve data consistency in the distributed computing core cluster.

[0025] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

[0027] Figure 1 A schematic diagram of a computing core cluster according to an embodiment of the present disclosure is shown.

[0028] Figure 2 A schematic diagram showing an information processing method according to an embodiment of the present disclosure applied to a single synchronization mechanism management unit.

[0029] Figure 3 A schematic diagram showing an information processing method according to an embodiment of the present disclosure being applied to multiple synchronization mechanism management units.

[0030] Figure 4 A flowchart of an information processing method according to an embodiment of the present disclosure is shown.

[0031] Figure 5 A schematic diagram showing task synchronization of a computing core cluster according to an embodiment of the present disclosure is shown.

[0032] Figure 6 A schematic diagram showing a synchronization coordination mechanism of an information processing method according to an embodiment of the present disclosure.

[0033] Figure 7 A block diagram of an information processing device according to an embodiment of the present disclosure is shown.

[0034] Figure 8 A block diagram of an electronic device according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0035] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0036] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0037] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0038] Figure 1 A schematic diagram of a computing core cluster according to an embodiment of the present disclosure is shown, wherein the computing core cluster includes multiple computing cores located in different processors. Figure 1 As shown, the computing core cluster includes multiple processors, and each processor may include multiple computing cores. The processors may be connected via an external bus (e.g., a communication bus outside the processor chip), that is, the computing cores located in different processors communicate via the external bus; the computing cores inside the processor may be connected via an internal bus (e.g., a communication bus inside the processor chip). The communication bus may include a data bus, a control bus, and an address bus.

[0039] In a possible implementation, the computing core includes a processing component and a storage component. The processing component may include a dendrite unit, an axon unit, a cell body unit, and a routing unit. The storage component may include a plurality of storage units.

[0040] In one possible implementation, multiple processors can be integrated into a brain-like computing chip, which is a neuromorphic circuit that integrates storage and computing. It uses the processing mode of the brain as a reference and simulates the transmission and processing of information by neurons in the brain to improve processing efficiency and reduce power consumption. Each processor may include multiple computing cores, which can independently process different tasks or process the same task in parallel to improve processing efficiency. Information between computing cores can be transmitted through routing units within the computing cores.

[0041] A processing component and a storage component may be provided within the computing core. The processing component may include a dendrite unit, an axon unit, a cell body unit, and a routing unit. The processing component may simulate the information processing mode of neurons in the brain, wherein the dendrite unit is used to receive signals, the axon unit is used to send spike signals, the cell body unit is used for integrated transformation of signals, and the routing unit is used to transmit information with other computing cores. The processing component within the computing core may read and write to multiple storage units of the storage component to interact with data with the storage component within the computing core, and may respectively undertake respective data processing tasks and / or data transmission tasks to obtain data processing results, or communicate with other computing cores. Among them, communicating with other computing cores includes communicating with other computing cores within the processor, and communicating with computing cores within other processors.

[0042] In scenarios where computing tasks are relatively large, the computing tasks will be assigned to a combination of multiple cores for parallel execution. In order to improve data processing capabilities and achieve consistency in data in a distributed computing core cluster, a synchronization mechanism management unit can be set in the computing core cluster. The synchronization mechanism management unit can be connected to any computing core in the computing core cluster to protect the synchronization and consistency of the computing core cluster system, effectively control or separate unauthorized resources or computer programs in the computing core cluster system, and facilitate the system architecture of automatic control network programs.

[0043] Figure 2 FIG. 1 is a schematic diagram showing an information processing method according to an embodiment of the present disclosure applied to a single synchronization mechanism management unit. Figure 2 As shown, for the synchronization mechanism management unit and the computing core cluster on the same processor chip, the synchronization mechanism management unit and the computing core can communicate through the internal bus. For example, a data channel and a message channel can be formed between the synchronization mechanism management unit and the computing core through the internal bus of the processor chip. In this way, the synchronization mechanism management unit can send computing tasks to the computing core cluster through the data channel, and the computing core cluster can return the computing results of the computing tasks to the synchronization mechanism management unit through the data channel. The synchronization mechanism management unit can receive the computing results from the host control machine (e.g. Figure 2 The business controller and business processing accelerator in the system may be composed of one or more computing cores) to receive control instructions.

[0044] like Figure 2 As shown, for the synchronization mechanism management unit and the computing core cluster in different processor chips, the synchronization mechanism management unit and the computing core can communicate through an external bus. For example, the communication between the synchronization mechanism management unit and the off-chip computing core can be achieved with the help of a signaling encoding and decoding module, a signaling channel routing table, and a signaling channel extended routing.

[0045] Optionally, the synchronization mechanism management unit is used to manage the sequentiality or synchronization of operations between the computing cores in the computing core cluster. For example, the synchronization mechanism management unit can be used to barrier the execution order of the computing cores in the computing core cluster so that one or several computing cores obtain the execution results before one or several other computing cores start the operation, just like a fence, so that after the operations of all the computing cores to be synchronized before the synchronization primitive are completed, the instructions after the synchronization primitive can start to execute.

[0046] Among them, the synchronization mechanism management unit can be an electronic circuit composed of general analog components, digital circuit components, etc., and can be appropriately deformed according to the actual application scenario. The embodiments of the present disclosure do not limit the specific circuit structure of the synchronization mechanism management unit.

[0047] Optionally, the synchronization mechanism management unit may be in the macroinstruction processor ( Figure 2 The macroinstruction processor may be used to parse and execute macroinstructions. For example, the macroinstruction processor may perform replacement operations according to predefined rules to expand macroinstructions into corresponding codes or instruction sequences. For another example, when a program is compiled or run, the macroinstruction processor may automatically process macroinstructions, thereby completing batch processing and optimization of codes. In a specific application process, the bus commands received by the macroinstruction processor may include: macroinstruction execution start, macroinstruction execution pause (halt), macroinstruction block sleep (sleep), macroinstruction block resume execution, macroinstruction execution wake-up, and macroinstruction execution stop, which is not limited in the embodiments of the present disclosure.

[0048] Figure 3 FIG. 1 is a schematic diagram showing an information processing method according to an embodiment of the present disclosure applied to multiple synchronization mechanism management units. Figure 3 As shown, the computing core cluster includes multiple computing cores located in different processors. Multiple synchronization mechanism management units can be set in the computing core cluster. Each synchronization mechanism management unit can synchronize any multiple computing cores in the computing core cluster. If the synchronization mechanism management unit and the multiple computing cores to be synchronized are located in the same processor, they can communicate via an internal bus; if the synchronization mechanism management unit and the multiple computing cores to be synchronized are located in different processors, they can use a signaling channel encoding and decoding module, a signaling channel routing table, and a signaling channel extended routing to achieve external bus communication between processor chips.

[0049] It should be noted that although Figure 1 to Figure 3 As an example, a computing core cluster with a synchronization mechanism management unit is introduced, but those skilled in the art will appreciate that the present disclosure should not be limited thereto. In fact, the user can flexibly set the location and number of synchronization mechanism management units in the computing core cluster according to personal preferences and / or actual application scenarios.

[0050] Figure 4 A flowchart of an information processing method according to an embodiment of the present disclosure is shown. Figure 4 As shown, the method is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores. The method includes:

[0051] In S11, the synchronization mechanism management unit performs a registration operation on a plurality of registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and writes a first value into each register;

[0052] Exemplarily, the synchronization primitive (also called blocking primitive) can be a preconfigured instruction, or a preconfigured program segment consisting of several instructions, such as creation, deletion, waiting (blocking\timeout blocking\non-blocking), release, setting, broadcasting, query and other instructions. The synchronization primitive is used to indicate the execution order of synchronization tasks between computing cores in a computing core cluster, so that the access of some computing cores to shared resources (such as shared resources stored by one or more computing cores) is safe and orderly.

[0053] Exemplarily, the synchronization mechanism management unit may include multiple registers, and the registers may be sequential logic circuits. For example, the registers may be composed of triggers and logic gate circuits with storage functions. The embodiments of the present disclosure do not limit the specific circuit structure of the registers.

[0054] In S12, in response to the computing core indicated by any register completing the respective first command to be synchronized, a decrement operation is performed on the first value in the register;

[0055] In S13, in response to the value of any register reaching a preset second value, canceling the register;

[0056] In S14, the multiple registers in the synchronization mechanism management unit are waited to be all cancelled, and a signaling message is generated, wherein the signaling message is used to indicate that the synchronization of the multiple first commands is completed.

[0057] For example, suppose that in a computing core cluster, there are N (N>1) computing cores (e.g., computing core 1 to computing core N) that jointly perform tasks, such as multi-layer convolution operations on the data to be processed. Since the amount of data to be processed is relatively large, the data to be processed can be split, and each computing core can perform convolution operations on part of the split data. In this process, since the convolution operation of each layer depends on the convolution operation result of the previous layer, the convolution operation of each layer can be synchronized, and the convolution operation of the same layer of the split data corresponding to each computing core is used as the first command to be synchronized.

[0058] For example, in step S11, when N computing cores perform the convolution operation of the first layer, the synchronization mechanism management unit may perform a registration operation on N registers (e.g., registers Fence1 to FenceN) set in the synchronization mechanism management unit according to the acquired synchronization primitive, bind a mapping relationship between register Fence1 and computing core 1, and write a first value (e.g., 1) to register Fence1; bind a mapping relationship between register Fence2 and computing core 2, and write a first value (e.g., 1) to register Fence2; and so on, bind a mapping relationship between register FenceN and computing core N, and write a first value (e.g., 1) to register FenceN;

[0059] In step S11, the synchronization mechanism management unit completes the registration according to the triggering requirements of the synchronization primitive, and writes a first value (for example, 1) to registers Fence1 to FenceN. It can be seen that in step S12, in response to the calculation core indicated by any register in registers Fence1 to FenceN completing the corresponding first command to be synchronized, such as the first layer convolution operation of the split data, the first value in the register (for example, 1) is decremented; and in S13, in response to the value of any register in registers Fence1 to FenceN reaching a preset second value (for example, 0), the register is cancelled;

[0060] For example, if computing core 1 completes the first layer of convolution operation of the corresponding split data, a subtraction operation is performed on the first value (e.g., 1) in register Fence1, the value stored in register Fence1 becomes the second value (e.g., 0), and register Fence1 is cancelled;

[0061] If the computing core 2 completes the first layer convolution operation of the corresponding split data, a subtraction operation is performed on the first value (for example, 1) in the register Fence2, and the value stored in the register Fence2 becomes the second value (for example, 0), and the register Fence2 is cancelled;

[0062] By analogy, if the computing core N completes the first layer of convolution operation of the corresponding split data, a subtraction operation is performed on the first value (for example, 1) in the register FenceN, the value stored in the register FenceN becomes the second value (for example, 0), and the register FenceN is cancelled.

[0063] In step S14, if there is a register containing a first value (for example, 1) among registers Fence1 to FenceN, the permission of the computing core corresponding to the register with the second value to execute other commands or tasks is locked, so that it enters a waiting state until all registers Fence1 to FenceN in the synchronization mechanism management unit are cancelled. A signaling message can be generated to indicate that computing cores 1 to N have obtained the first layer convolution operation results of the corresponding split data, thereby realizing the synchronization of multiple first commands.

[0064] Optionally, the synchronization mechanism management unit polls and checks multiple registers in the synchronization mechanism management unit according to a preset period to obtain a check result; when the check result indicates that all multiple registers in the synchronization mechanism management unit are cancelled, a signaling message is generated.

[0065] For example, a logic circuit module that can perform polling checks on registers can be set in the synchronization mechanism management unit. The synchronization mechanism management unit uses the logic circuit module to poll and check the first values ​​stored in registers Fence1 to FenceN according to a preset period. If there is a register with a first value (for example, 1) in registers Fence1 to FenceN, the permission of the computing core corresponding to the register with the second value to execute other commands or tasks is locked, so that it enters a waiting state until all registers Fence1 to FenceN in the synchronization mechanism management unit are cancelled. A signaling message can be generated to indicate that computing cores 1 to N have obtained the first layer convolution operation results of the corresponding split data, thereby realizing the synchronization of multiple first commands.

[0066] The preset period may be any value, such as a clock period of a computing core, and the embodiments of the present disclosure do not limit the specific value of the preset period.

[0067] Through polling checks at preset periods, the synchronization mechanism management unit can promptly detect changes in register values, and when all registers are deregistered, it can quickly generate signaling messages. This timely response helps reduce waiting time and improve the overall operating efficiency of the computing core cluster.

[0068] It should be understood that in actual applications, the embodiments of the present disclosure do not specifically limit the number of registers in the synchronization mechanism management unit and the correspondence between each register and the computing core.

[0069] The method of hardware-implemented synchronization mechanism provided by steps S11 to S14 can hide the synchronization signaling (it is implemented by the underlying hardware synchronization mechanism management unit that is not user-oriented, rather than being exposed to the user for configuration or use through software or user interface), improve the synchronization efficiency between the units of the distributed system (such as computing cores), facilitate subsequent and event message collaboration, improve control efficiency and coverage, and achieve data consistency of the distributed computing core cluster.

[0070] It should be understood that the first value above is the initial value written to the register by the synchronization mechanism management unit according to the acquired synchronization primitive, and the second value is the target value used to determine whether to perform a cancellation operation on the register. The embodiments of the present disclosure do not limit the specific values ​​of the first value and the second value.

[0071] In one possible implementation, the first command to be synchronized includes multiple sub-commands, and step S12 may include: in response to the computing core indicated by the register completing any sub-command, reading a current value from the register, the current value being less than or equal to the first value and greater than the second value; and performing a subtraction operation on the current value in the register.

[0072] For example, the first command to be synchronized may include K (K>1) subcommands, and the K subcommands may be executed by different computing cores. For example, if the K subcommands are respectively executed by computing cores 1 to K, and there is no dependency relationship between the K subcommands, the synchronization mechanism management unit may register a register in the synchronization mechanism management unit according to the acquired synchronization primitive, bind the register to the computing cores 1 to K, and write a first value K to the register;

[0073] Whenever the synchronization mechanism management unit detects that a computing core among computing cores 1 to K has completed a subcommand, it can perform a subtraction operation on the current value in the register until all K subcommands are executed. Only then will the value in the register reach the second value, thereby canceling the register.

[0074] In this way, the synchronization mechanism management unit can wait for multiple sub-commands of multiple computing cores corresponding to each register to be executed completely, and then generate a signaling message. In this way, each register can provide synchronization services for multiple computing cores, which is conducive to improving the scalability and flexibility of the information processing method.

[0075] The following is an exemplary description of the information processing method of the present disclosure.

[0076] In a possible implementation, the synchronization mechanism management unit further includes at least one sequential container encapsulating a dynamic array, the dynamic array of the sequential container is used to record the registration order of multiple registers, and the method further includes: in the case where multiple first commands to be synchronized have dependencies, the sequential container records the registration order of multiple registers according to the dependencies of the multiple first commands to be synchronized; in response to the computing core indicated by the current register completing the first command to be synchronized, and the register whose registration order is before the register completes the cancellation, and the first value in the register is decremented. Among them, the synchronization mechanism can be understood as a dynamically configurable sequential container, and the length of the dynamic array in the sequential container (such as the number of elements in the dynamic array) can represent the registered register.

[0077] Exemplarily, it is assumed that 8 registers are set in the synchronization mechanism management unit, for example, register Fence0 to register Fence7. If the synchronization primitive received by the synchronization mechanism management unit is to synchronize the first command 1 to be executed by computing core 1, the first command 2 to be executed by computing core 2, and the first command 3 to be executed by computing core 3, and the first command 3 to be executed by computing core 3 depends on the result of the first command 2 to be executed by computing core 2, and the first command 2 to be executed by computing core 2 depends on the result of the first command 1 to be executed by computing core 1.

[0078] In this case, the synchronization mechanism management unit can select any three registers from register Fence0 to register Fence7 for registration, for example, register Fence1 to register Fence3 can be selected, and according to the dependency relationship between first command 1 to first command 3, register Fence1 can be first bound to computing core 1 for mapping, and a first value (for example, 1) can be written to register Fence1; then register Fence2 can be bound to computing core 2 for mapping, and a first value (for example, 1) can be written to register Fence2; then register Fence3 can be bound to computing core 3 for mapping, and a first value (for example, 1) can be written to register Fence3.

[0079] At the same time, the sequence container of the synchronization mechanism management unit will record the registration order of registers Fence1 to Fence3. For example, in a dynamically configurable sequence container, the dynamic array can be configured as an array containing 3 elements. The first element can be set to the address index of register Fence1, the second element can be set to the address index of register Fence2, and the third element can be set to the address index of register Fence3. The registration order of the registers is recorded by sorting the elements in the dynamic array.

[0080] In this case, the first command 1 executed by computing core 1 does not depend on the commands of other computing cores (for example, the first command 2 of computing core 2, or the first command 3 of computing core 3). The synchronization mechanism management unit can instruct computing core 1 to execute the first command 1, and in response to computing core 1 completing the first command 1, perform a self-decrement operation on the first value in register Fence1 to obtain the second value, cancel register Fence1, and delete the first element from the dynamic array. At this time, the address index recorded in register Fence2 becomes the first position.

[0081] The first command 2 executed by computing core 2 depends on the first command 1 of computing core 1. If the address index of register Fence1 is recorded as the first one, it means that the first command 1 in computing core 1 has not been executed yet, and the first command 2 in computing core 2 needs to wait for the first command 1 in computing core 1 to be executed; if the address index of register Fence2 is recorded as the first one, it means that the first command 1 in computing core 1 has been executed. The synchronization mechanism management unit can instruct computing core 2 to execute the first command 2, and in response to computing core 2 completing the first command 2, perform a self-decrement operation on the first value in register Fence2 to obtain the second value, cancel register Fence2, and delete the first element from the dynamic array. At this time, the address index of register Fence3 is recorded as the first one.

[0082] The first command 3 executed by computing core 3 depends on the first command 1 in computing core 1 and the first command 2 in computing core 2. If the address index of register Fence1 is recorded as the first, it means that the first command 1 in computing core 1 and the first command 2 in computing core 2 have not been executed yet, and the first command 3 in computing core 3 needs to wait for the first command 1 in computing core 1 and the first command 2 in computing core 2 to be executed.

[0083] If the address index of register Fence2 is recorded as the first, it means that the first command 1 in computing core 1 has been executed, but the first command 2 in computing core 2 has not been executed. The first command 3 in computing core 3 needs to wait for the first command 2 in computing core 2 to be executed.

[0084] If the address index of register Fence3 is recorded as the first one, it means that the first command 1 in computing core 1 and the first command 2 in computing core 2 have been executed. The synchronization mechanism management unit can instruct computing core 3 to execute the first command 3, and in response to computing core 3 completing the first command 3, perform a self-decrement operation on the first value in register Fence3 to obtain the second value, cancel register Fence3, and delete the first element from the dynamic array. At this time, the dynamic array is empty, indicating that the synchronization is completed. The synchronization mechanism management unit can generate a signaling message to notify computing cores 1 to 3, and computing cores 1 to 3 can perform other tasks.

[0085] In this way, it can be used to ensure that homogeneous tasks in the distributed computing core cluster are executed sequentially, thereby ensuring the consistency of the data related to the upper-level application tasks. Since the synchronization mechanism is implemented by the synchronization mechanism management unit as hardware, the synchronization signaling can be hidden, improving the synchronization efficiency between the units of the distributed system (such as computing cores); and the synchronization mechanism can work in conjunction with event messages, which can improve control efficiency and coverage and achieve distributed cluster data consistency.

[0086] In a possible implementation, performing a decrement operation on the first value in the register includes: configuring an interrupt handling subroutine for each sequential container; in response to the sequential container receiving a vector feedback signal carrying a register identifier, triggering the interrupt handling subroutine to perform a decrement operation on the first value in the register indicated by the register identifier.

[0087] Configuring an interrupt handling subroutine for each sequential container means that each container can independently process the vector feedback signal it receives, allowing synchronization operations to be performed in parallel between multiple containers, thereby greatly improving the efficiency of synchronization. In addition, by configuring an independent interrupt handling subroutine for each sequential container, the system can easily add or remove sequential containers as needed without making complex modifications to the entire synchronization mechanism, which has better flexibility and scalability and can adapt to changing task synchronization requirements.

[0088] In the example, after the synchronization mechanism management unit completes the registration, when the register maintenance unit of the synchronization mechanism management unit receives the vector feedback signal, it triggers the register maintenance interrupt processing subroutine of the synchronization mechanism management unit (wherein, a corresponding interrupt processing subroutine can be configured for each sequential container in the program), and the instructions in the interrupt processing subroutine complete the operation of subtracting 1 from the value of the corresponding register. The program itself waits according to the value of the register.

[0089] If there is only one interrupt handling subroutine in the synchronization mechanism management unit, the result will be that the entire program needs to wait for the vector feedback signal of the interrupt handling subroutine before continuing to execute.

[0090] In a possible implementation, in response to each computing core to be synchronized receiving a signaling message returned by the synchronization mechanism management unit, indicating that the synchronization is complete, each computing core can end the waiting state and execute the second command of the next timing. The second command can be a read-write command, a computing command, a control command, a transmission command, etc. The embodiment of the present disclosure does not limit the specific function of the second command.

[0091] In a possible implementation, multiple sequence containers may use the same register, or multiple registers may be used to implement more complex program control, which is not limited in the embodiments of the present disclosure.

[0092] Figure 5 A schematic diagram showing task synchronization of a computing core cluster according to an embodiment of the present disclosure is shown as follows: Figure 5 As shown, in a certain scenario, the first step is that computing core B submits a piece of data resources to computing core A, and the second step is that computing core B commands computing core A to use the data resources for image rendering. From the first step to the second step, computing core B needs to confirm that the data resources are completely submitted to computing core A.

[0093] In this case, the synchronization mechanism management unit introduced in steps S11 to S14 above can be used to add a synchronization primitive to the computing core A between the first step and the second step to confirm that the data resources have been submitted before issuing the rendering instruction. For example, if the computing core B wants to confirm that all previous computing commands have been executed, it can insert a synchronization primitive after the computing command and wait for the computing core A to return the signaling message of the synchronization primitive, so that it can confirm that all previous computing commands have been executed.

[0094] For example, the synchronization mechanism management unit may be used to register or unregister two registers corresponding to computing core A and computing core B, thereby achieving synchronization between computing core A and computing core B tasks.

[0095] like Figure 5 As shown, in the order of the timeline, computing core B sends a computing command, a synchronization primitive is inserted after the computing command of computing core B, and computing core A executes the computing command; computing core B waits for a signaling message, and computing core A continues to execute the computing command; computing core A adds a synchronization primitive after confirming that the computing command has been executed; computing core B confirms the signaling message returned by the synchronization primitive, and the system can execute other commands.

[0096] pass Figure 5 It can be seen that the information processing method of the embodiment of the present disclosure realizes the synchronization of multiple computing cores in a computing core cluster. This synchronization mechanism will confirm that all previous commands have been executed, insert a synchronization primitive after the command, and wait for the computing core to return the signaling message of the synchronization primitive, then it can be confirmed that all previous commands have been executed, thereby completing the synchronization feature.

[0097] Figure 6 A schematic diagram showing a synchronization coordination mechanism of an information processing method according to an embodiment of the present disclosure is shown as follows: Figure 6In the synchronization diagram shown, each task can correspond to a computing core in the computing core cluster for cluster and higher-level control, wherein each computing core can have four working states, including, for example, computing state, data sending state, data receiving state, and waiting state (for example, the task in the corresponding computing core is in waiting, stopping, or sleeping state).

[0098] There are two tasks in this synchronization sequence, namely task 0 and task 1, where task 0 is deployed to three working modules (such as computing core 0, computing core 1 and computing core 2), and task 1 is deployed to another three working modules (such as computing core 3, computing core 4 and computing core 5).

[0099] In the example, for task 0, it means that computing core 0 and computing core 2 send data to computing core 1 respectively (see Figure 6 Send 0 and Send 1 in Task 0), when computing core 0 finishes sending data, it receives the signaling message of computing core 1 (see Figure 6 After computing core 1 receives data from computing core 0 and computing core 2 (see Figure 6 Receive 0 and receive 1 in task 0) and send signaling messages (see Figure 6 Signaling message B) is sent to computing core 2, after which computing core 1 and computing core 2 enter the computing state respectively.

[0100] In the example, for task 1, computing core 3 sends data to computing core 4 (see Figure 6 When the computing core 3 has finished sending the data, it waits to receive the signaling message of the computing core 4 (see Figure 6 The computing core 4 receives the data (see Figure 6 After receiving 0 in Task 1, the calculation is started immediately and the data is sent (see Figure 6 Task 1 sends data to computing core 5, which receives the data (see Figure 6 The receiving part 1 of task 1) and waits for the signaling message of computing core 4 (see Figure 6 After the signaling message D), computing core 4 and computing core 5 start computing respectively.

[0101] according to Figure 6 It can be seen that the information processing method of the embodiment of the present disclosure can be deployed within a single task, and a very flexible synchronization mechanism can be implemented for communication sending and receiving.

[0102] Figure 7 A block diagram of an information processing device according to an embodiment of the present disclosure is shown as follows: Figure 7As shown, the device is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores. The device includes:

[0103] A registration module 71, configured for the synchronization mechanism management unit to perform a registration operation on a plurality of registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and write a first value into each register;

[0104] A decrement module 72, configured to perform a decrement operation on a first value in any register in response to the computing cores indicated by any register completing their respective first commands to be synchronized;

[0105] A cancellation module 73, configured to cancel the register in response to the value of any register reaching a preset second value;

[0106] The generating module 74 is used to wait for all the registers in the synchronization mechanism management unit to be cancelled, and to generate a signaling message, wherein the signaling message is used to represent the synchronization of the plurality of first commands.

[0107] In one possible implementation, the computing core cluster includes multiple computing cores located in different processors, each processor includes a synchronization mechanism management unit and multiple computing cores, wherein the synchronization mechanism management unit located in the same processor communicates with the computing cores through an internal bus, and the synchronization mechanism management unit located in different processors communicates with the computing cores through an external bus.

[0108] In a possible implementation, the synchronization mechanism management unit further includes at least one sequence container encapsulating a dynamic array, and the dynamic array of the sequence container is used to record the registration order of multiple registers, such as Figure 7 As shown in the dotted part, the device may also include a storage module 75, which is used to: when there is a dependency relationship between multiple first commands to be synchronized, the sequence container records the registration order of multiple registers according to the dependency relationship between the multiple first commands to be synchronized; in response to the computing core indicated by the current register completing the first command to be synchronized, and the register whose registration order is before the register is cancelled, and a self-decrement operation is performed on the first value in the register.

[0109] In this way, the storage module 75 is used for dynamic array encapsulation of sequential containers, and implements storage operations on the registration execution operations in the synchronization mechanism management unit.

[0110] In a possible implementation, performing a decrement operation on the first value in the register includes: configuring an interrupt handling subroutine for each sequential container; in response to the sequential container receiving a vector feedback signal carrying a register identifier, triggering the interrupt handling subroutine to perform a decrement operation on the first value in the register indicated by the register identifier.

[0111] In a possible implementation, the device further includes a response module ( Figure 7 (not shown), used for: in response to any computing core receiving a signaling message returned by the synchronization mechanism management unit, the computing core executes the second command of the next timing.

[0112] In a possible implementation, the generation module 74 is used to: the synchronization mechanism management unit polls and checks multiple registers in the synchronization mechanism management unit according to a preset period to obtain an inspection result; when the inspection result indicates that all multiple registers in the synchronization mechanism management unit are cancelled, generate a signaling message.

[0113] In one possible implementation, the self-decrement module 72 is used to: when the first command to be synchronized includes multiple sub-commands, in response to the computing core indicated by the register completing any sub-command, read the current value from the register, the current value is less than or equal to the first value and greater than the second value; and perform a subtraction operation on the current value in the register.

[0114] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0115] The embodiment of the present disclosure further provides a many-core chip, which includes a synchronization mechanism management unit and multiple computing cores. The synchronization mechanism management unit connects each computing core and executes the information processing method as described above.

[0116] In a possible implementation, an embodiment of the present disclosure further provides an electronic device, wherein the electronic device includes at least one multi-core chip as described above. The electronic device may be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc.

[0117] The embodiment of the present disclosure also provides a computer-readable storage medium on which computer program instructions are stored, and the computer program instructions implement the above method when executed by a processor. The computer-readable storage medium can be a volatile or non-volatile computer-readable storage medium.

[0118] The embodiments of the present disclosure also provide a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above method.

[0119] Figure 8 1 shows a block diagram of an electronic device according to an embodiment of the present disclosure. For example, the electronic device 1900 may be provided as a server or a terminal device. Figure 8 , the electronic device 1900 includes a processing component 1922, which further includes one or more processors, each processor including a synchronization mechanism management unit and multiple computing cores, and a memory resource represented by a memory 1932 for storing instructions that can be executed by the processing component 1922, such as an application. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0120] The electronic device 1900 may also include a power supply component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input / output interface 1958 (I / O interface). The electronic device 1900 may operate based on an operating system stored in the memory 1932, such as Windows Server 2003. TM , Mac OS X TM , Unix TM ,Linux TM , FreeBSD TM or similar.

[0121] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions, which can be executed by the processing component 1922 of the device 1900 to perform the above method.

[0122] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0123] Computer readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device. Computer readable storage medium can be, for example, (but not limited to) an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (non-exhaustive list) of computer readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, for example, a punch card or a convex structure in a groove on which instructions are stored, and any suitable combination thereof. The computer readable storage medium used here is not interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated by a waveguide or other transmission medium (for example, a light pulse by an optical fiber cable), or an electrical signal transmitted by a wire.

[0124] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0125] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0126] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0127] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0128] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0129] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0130] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An information processing method, characterized in that: The method is applied to at least one synchronization mechanism management unit connected to a computing core cluster, the synchronization mechanism management unit being used to implement information synchronization between multiple computing cores, and the method comprises: The synchronization mechanism management unit performs a registration operation on a plurality of registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and writes a first value into each register; In response to the computing cores indicated by any register completing their respective first commands to be synchronized, performing a self-decrement operation on the first value in the register; In response to a value of any register reaching a preset second value, canceling the register; Waiting for all the multiple registers in the synchronization mechanism management unit to be cancelled, and generating a signaling message, wherein the signaling message is used to indicate that the synchronization of the multiple first commands is completed; The synchronization mechanism management unit further includes at least one sequence container encapsulating a dynamic array, the dynamic array of the sequence container is used to record the registration order of multiple registers, and the method further includes: In the case where a plurality of first commands to be synchronized have dependencies, the sequence container records the registration order of the plurality of registers according to the dependencies of the plurality of first commands to be synchronized; In response to the computing core indicated by the current register completing the first command to be synchronized, and the register before the register in the registration order completing the cancellation, a self-decrement operation is performed on the first value in the register; The step of performing a decrement operation on the first value in the register includes: Configure an interrupt handling subroutine for each sequential container; In response to the sequential container receiving a vector feedback signal carrying a register identifier, the interrupt handling subroutine is triggered to perform a decrement operation on a first value in a register indicated by the register identifier.

2. The method according to claim 1, characterized in that: The computing core cluster includes multiple computing cores located in different processors, each processor includes a synchronization mechanism management unit and multiple computing cores, wherein the synchronization mechanism management unit located in the same processor communicates with the computing cores through an internal bus, and the synchronization mechanism management unit located in different processors communicates with the computing cores through an external bus.

3. The method according to claim 1 or 2, characterized in that: The method further includes: in response to any computing core receiving a signaling message returned by the synchronization mechanism management unit, the computing core executes a second command of the next timing.

4. The method according to claim 1 or 2, characterized in that: Wait for multiple registers in the synchronization mechanism management unit to be fully deregistered and generate signaling messages, including: The synchronization mechanism management unit polls and checks a plurality of registers in the synchronization mechanism management unit according to a preset period to obtain a check result; In the case where the checking result indicates that all the multiple registers in the synchronization mechanism management unit are cancelled, a signaling message is generated.

5. The method according to claim 1 or 2, characterized in that: The first command to be synchronized includes a plurality of sub-commands, and in response to the computing core indicated by any register completing the first command to be synchronized, a self-decrement operation is performed on the first value in the register, including: In response to the computing core indicated by the register completing any subcommand, reading a current value from the register, the current value being less than or equal to the first value and greater than the second value; Decrement the current value in the register by one.

6. An information processing device, characterized in that: The device is applied to at least one synchronization mechanism management unit connected to a computing core cluster, and the synchronization mechanism management unit is used to realize information synchronization between multiple computing cores. The device includes: A registration module, used for the synchronization mechanism management unit to perform a registration operation on a plurality of registers set in the synchronization mechanism management unit according to the acquired synchronization primitive, and write a first value into each register; A self-decrement module, configured to execute a self-decrement operation on a first value in any register in response to the computing core indicated by any register completing the respective first command to be synchronized; A cancellation module, configured to cancel the register in response to the value of any register reaching a preset second value; A generating module, used for waiting for all the multiple registers in the synchronization mechanism management unit to be cancelled, and generating a signaling message, wherein the signaling message is used for representing the synchronization of the multiple first commands; The synchronization mechanism management unit further includes at least one sequence container encapsulating a dynamic array, the dynamic array of the sequence container is used to record the registration order of multiple registers, and the device further includes a storage module for: In the case where a plurality of first commands to be synchronized have dependencies, the sequence container records the registration order of the plurality of registers according to the dependencies of the plurality of first commands to be synchronized; In response to the computing core indicated by the current register completing the first command to be synchronized, and the register before the register in the registration order completing the cancellation, a self-decrement operation is performed on the first value in the register; The step of performing a decrement operation on the first value in the register includes: Configure an interrupt handling subroutine for each sequential container; In response to the sequential container receiving a vector feedback signal carrying a register identifier, the interrupt handling subroutine is triggered to perform a decrement operation on a first value in a register indicated by the register identifier.

7. A many-core chip, characterized in that: The many-core chip comprises a synchronization mechanism management unit and a plurality of computing cores, wherein the synchronization mechanism management unit is connected to each computing core, and the synchronization mechanism management unit executes the information processing method according to any one of claims 1 to 5.

8. An electronic device, characterized in that: The method comprises at least one many-core chip as claimed in claim 7.