Equipment switching method and device, equipment and storage medium

Abstract

The invention discloses an equipment switching method and device, equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: acquiring a state parameter of production equipment which is executing a training task, wherein the state parameter is used for describing an operation state of the production equipment; and if the state parameter meets the troubleshooting condition, switching from the production equipment to the first standby equipment to execute the training task. In the execution process of the training task, the production equipment which may fail is determined. After the production equipment which possibly breaks down is determined, the standby equipment is adopted to replace the production equipment to execute the training task, and the training task does not need to be interrupted. Therefore, the problems that the switching time is long and the execution efficiency of the training task is not high because the standby equipment is switched after the production equipment breaks down and the training task is interrupted are avoided, the switching time of the equipment for executing the training task is shortened, and the execution efficiency of the training task is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to device switching methods, apparatus, devices, and storage media. Background Technology

[0002] With the continuous development of large-scale model technology, the size and number of parameters of various large models are increasing, leading to a larger training scale. The training task of large models is usually performed in parallel, that is, the training task is broken down into multiple parts, and multiple devices or computing nodes execute the computations simultaneously, thereby achieving rapid task completion. However, frequent equipment failures will constantly cause interruptions to the training task, which means significant wasted resources.

[0003] In related technologies, by deploying backup equipment, the training task is backed up at various backup points (also known as checkpoints) during its execution. If the production equipment executing the training task fails and interrupts the training task, a backup device is selected to replace the failed production equipment, and the training task is restarted from the backup point closest to the interruption point to continue its execution.

[0004] However, restarting the training task from the backup point is a time-consuming process, resulting in low efficiency. Furthermore, as the scale of the training task increases, its efficiency further decreases. Summary of the Invention

[0005] This application provides a device switching method, apparatus, device, and storage medium, which reduces the device switching time for performing training tasks and improves the efficiency of training task execution.

[0006] To achieve the above objectives, the embodiments of this application adopt the following technical solutions:

[0007] Firstly, a device switching method is provided. The device switching method provided in this application includes:

[0008] Obtain the status parameters of the production equipment currently executing the training task. These status parameters describe the operating status of the production equipment. If the status parameters meet the troubleshooting conditions, switch from the production equipment to the first standby equipment to execute the training task. The troubleshooting conditions indicate the probability of equipment failure, and the first standby equipment is a standby device in an idle state.

[0009] The solution provided in this application acquires the status parameters of the production equipment during the execution of the training task. Upon determining that a production equipment malfunction may occur, a backup device is used to replace it and perform the training task, eliminating the need for training task interruption. This avoids the long switching time and low training task execution efficiency that occurs when the training task is interrupted due to a production equipment malfunction. By shortening the switching time of the equipment performing the training task, the efficiency of the training task execution is improved.

[0010] One possible implementation is that, if the status parameters meet the check conditions, the system switches from the production equipment to the first standby equipment to execute the training task. Specifically, if the status parameters meet the check conditions, after completing the checkpoint backup operation for the training task, the system switches from the production equipment to the first standby equipment to execute the training task. It can be seen that the equipment switchover occurs after the checkpoint backup operation is completed, at which point the backup data is up-to-date. This avoids the problem of repeated training between the most recent checkpoint and the interruption point caused by restarting the training task from the most recent checkpoint, thus improving the execution efficiency of the training task.

[0011] Another possible implementation involves switching from the production equipment to the first standby equipment to execute the training task. Specifically, the production equipment stops executing the training task, and the first standby equipment uses the backup data of the training task to continue executing the training task. This backup data is the latest version. Since the backup data of the training task fully reflects the training task parameters, the first standby equipment can continue executing the training task using the backup data, thus improving the execution efficiency of the training task.

[0012] Another possible implementation involves switching from the production equipment to the first standby equipment after the checkpoint backup operation of the training task is completed. Specifically, after the checkpoint backup operation is completed, backup data is obtained. The production equipment stops executing the training task, and the first standby equipment uses the backup data to execute the training task. In this implementation, the backup data is generated after a potential production equipment failure is identified; therefore, this backup data provides the most accurate representation of the training task's execution status. Using this backup data after equipment switching ensures a seamless transition for the standby equipment, preventing duplicate execution of the training task and improving its efficiency.

[0013] Another possible implementation involves the production equipment ceasing its training task, while the first standby device continues the training task using backup data. Specifically, this can be achieved by disconnecting the production equipment from other devices, and then connecting the first standby device to other devices that communicated and / or exchanged data with the production equipment during the training task. The production equipment's data is then loaded into the first standby device. The first standby device then continues the training task using the backup data. By only disconnecting other devices that were connected to the production equipment during this training task, and by only connecting the first standby device to other devices that communicated and / or exchanged data with it during this training task, rather than disconnecting all connections to the production equipment, this partial connection method allows for a faster switchover from the production equipment to the first standby device, resulting in higher equipment switchover efficiency.

[0014] Another possible implementation involves using the operating parameters of the equipment when a malfunction occurs as the screening condition. The status parameters must meet the screening condition, specifically by ensuring the status parameters reach the operating parameters. This provides a method to determine the probability of a production device malfunctioning. By setting the threshold value of the malfunctioning status parameters as the operating parameters, the production device is identified as potentially malfunctioning when the status parameters reach the operating parameters. By setting operating parameters and determining the probability of a malfunction based on whether the production device's status parameters reach the operating parameters, this screening method is relatively simple and direct. After obtaining the status parameters, the probability of a malfunction can be directly determined, improving the efficiency of determining the probability of a malfunction and thus improving the execution efficiency of the training task.

[0015] Another possible implementation involves screening conditions including a status score greater than or equal to a first threshold; and status parameters meeting the screening conditions. Specifically, this can be achieved by scoring the production equipment based on its status parameters to obtain a status score; and ensuring the status score is greater than or equal to the first threshold. This provides another method for determining the probability of a production equipment malfunction. By scoring the production equipment, potentially malfunctioning equipment is identified based on the scores. This method clearly and explicitly displays the status distribution of the production equipment and allows for one-click selection of equipment with a probability of malfunction based on its status score, improving the efficiency of determining the probability of a malfunction and thus improving the execution efficiency of the training task.

[0016] Another possible implementation is that the first backup device belongs to a backup device pool. The device switching method provided in this application further includes: obtaining the status parameters of the backup devices in the backup device pool; and maintaining the backup devices in the backup device pool based on the status parameters. By obtaining the status parameters of the backup devices and maintaining backup devices with abnormal status parameters, the backup devices in the backup device pool can always be in a normal operating state, so that they can be called upon to replace the generating device at any time.

[0017] Another possible implementation involves maintaining the backup devices in the backup device pool based on their status parameters. This can be specifically achieved through one or more of the following operations: resetting and repairing the backup devices based on their status parameters; issuing alarms and warnings for the backup devices based on their status parameters; and conducting performance evaluation tests on the backup devices based on their status parameters.

[0018] Secondly, a device switching apparatus is provided, the apparatus comprising:

[0019] The acquisition module is used to acquire the status parameters of the production equipment that is performing training tasks. The status parameters are used to describe the operating status of the production equipment.

[0020] The switching module is used to switch from the production equipment to the first standby equipment to perform the training task if the status parameters meet the troubleshooting conditions. The troubleshooting conditions indicate the probability of equipment failure, and the first standby equipment is the standby equipment in an idle state.

[0021] One possible implementation involves a switching module, specifically used to: stop the production equipment from executing the training task, and use the backup equipment to continue executing the training task using backup data.

[0022] Another possible implementation is a switching module, specifically used to: if the status parameters meet the troubleshooting conditions, after completing the checkpoint backup operation of the training task, switch from the production equipment to the first standby equipment to execute the training task.

[0023] Another possible implementation involves a switching module comprising a backup unit and a switching unit. The backup unit is specifically used to obtain backup data after the checkpoint backup operation of the training task is completed. The switching unit is specifically used to ensure that when the production equipment stops executing the training task, the first standby equipment uses the backup data to continue executing the training task.

[0024] Another possible implementation involves a switching unit, specifically used to: disconnect the production equipment from other equipment, and link the first backup equipment to other equipment, which refers to devices that communicate and / or exchange data with the production equipment during the execution of the training task. The equipment data from the production equipment is then loaded into the first backup equipment. The first backup equipment uses the backup data to continue executing the training task.

[0025] Another possible implementation involves checking the operating parameters of the equipment when a malfunction occurs; the switching module also includes a troubleshooting unit. Specifically, the troubleshooting unit is used to determine whether the status parameters have reached the operating parameters.

[0026] Another possible implementation involves screening conditions including a status score greater than or equal to a first threshold; the screening unit is specifically used to: score the production equipment based on status parameters to obtain a status score for the production equipment. The status score of the production equipment is greater than or equal to the first threshold.

[0027] Another possible implementation is that the first backup device belongs to a backup device pool; the acquisition module is specifically used to: acquire the status parameters of the backup devices in the backup device pool. The device switching device provided in this application also includes: a maintenance module. The maintenance module is specifically used to: maintain the backup devices in the backup device pool according to the status parameters of the backup devices.

[0028] Another possible implementation involves a maintenance module, specifically used for: resetting and repairing the backup device based on its status parameters; issuing alarms and warnings for the backup device based on its status parameters; and performing performance evaluation tests on the backup device based on its status parameters.

[0029] Thirdly, a computer device is provided, the computer device comprising: a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the device switching method as described above.

[0030] Fourthly, a computer-readable storage medium is provided, wherein at least one computer program is stored therein, the at least one computer program being loaded and executed by a processor to implement the device switching method as described above.

[0031] Fifthly, a computer program product is provided, the computer program product comprising a computer program stored in a computer-readable storage medium; the computer program is read from and executed by a processor of a computer device from the computer-readable storage medium, causing the computer device to perform the device switching method as described above.

[0032] The solutions provided in the third to fifth aspects above are used to implement the methods provided in the first aspect or any possible implementation method above, and their specific implementations will not be described in detail.

[0033] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other. Attached Figure Description

[0034] Figure 1 This is a flowchart of an industry-standard device switching method provided by an exemplary embodiment;

[0035] Figure 2 This is a schematic diagram of the architecture of a computer system provided in an exemplary embodiment of this application;

[0036] Figure 3 This is a schematic diagram of the architecture of another computer system provided in an exemplary embodiment of this application;

[0037] Figure 4 This is a flowchart of a device switching method provided in an exemplary embodiment of this application;

[0038] Figure 5 This is a flowchart of a device switching method provided in an exemplary embodiment of this application;

[0039] Figure 6 This is a schematic diagram of a device switching method provided in an exemplary embodiment of this application;

[0040] Figure 7 This is a flowchart of a device switching method provided in an exemplary embodiment of this application;

[0041] Figure 8 This is a block diagram of a device switching apparatus provided in an exemplary embodiment of this application;

[0042] Figure 9 This is a schematic diagram of the structure of a computer device provided in an exemplary embodiment of this application. Detailed Implementation

[0043] In the embodiments of this application, in order to clearly describe the technical solutions of the embodiments of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different. The technical features described by "first" and "second" have no sequential or size order.

[0044] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0045] In the embodiments of this application, at least one can also be described as one or more, and multiple can be two, three, four or more, and this application does not impose any restrictions.

[0046] Furthermore, the network architecture and scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0047] To facilitate understanding, the following explanations are provided for several terms used in this application.

[0048] Artificial intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence in order to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0049] The training task, in machine learning or deep learning, refers to training a model using algorithms and large amounts of data so that the model can learn from the data and make predictions or decisions. The training task is a crucial step in realizing artificial intelligence, determining the model's performance in practical applications. With the widespread application of artificial intelligence technology, training tasks play a vital role in various industries and fields, including finance, healthcare, and education. By continuously optimizing the training process, the accuracy and efficiency of the model can be improved, thereby driving innovation and development in related industries. To improve training speed and efficiency, training tasks are typically computed in parallel, meaning that multiple computing units or devices process data simultaneously, thus shortening the time required for the training task.

[0050] It should be noted that all information (including but not limited to equipment information, personal information of the target, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in this application are authorized by the target or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the status parameters of the production equipment and the status parameters of the standby equipment described in the embodiments of this application are all authorized by the target or fully authorized by all parties, and the collection, use, and processing of the status parameters comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0051] The following is a brief explanation of the industry-standard device switching methods.

[0052] For example, such as Figure 1 The flowchart shown illustrates a standard industry-standard device switching method. The process of this standard method includes:

[0053] 1. Checkpoint Backup Operation. During the training tasks performed on the production equipment, checkpoint backup operations are performed periodically. Checkpoint backup typically refers to saving the current state at a specific checkpoint during the training task. In case of errors or the need to restart, the system can quickly restore to the most recent checkpoint and continue calculations based on the backup data saved at that checkpoint, thereby reducing data loss and improving system robustness. 2. Production Equipment Failure. During the training tasks performed on the production equipment, a failure occurs. 3. Failure Detection. After the production equipment failure, the operations and maintenance system detects the failure. 4. Training Task Stoppage. After the operations and maintenance system detects the production equipment failure, it interrupts the training task. 5. Notification to the Scheduling Platform. The operations and maintenance system notifies the scheduling platform to interrupt the training task and call up a backup device. 6. Stop Calling the Failed Device and Add a Backup Device. The scheduling platform interrupts the training task and stops calling the failed device. After stopping the call to the failed device, a backup device is selected to replace it. 7. Training Task Initialization. After determining the backup device to replace the failed device, the training task is initialized. 8. Establishing a Connection. After initializing the training task, the standby device is fully linked using the hierarchical cluster computing library (HCCL), meaning the standby device is linked to the same nodes or devices as the failed device. 9. Binary Compilation. After linking the standby device, the source code is compiled into an executable file for the standby device. 10. Load Backup Data. After binary compilation, the backup data stored at the most recent checkpoint is loaded. 11. Continue Training Task. After loading the backup data stored at the most recent checkpoint, the standby device continues the training task based on the backup data.

[0054] In industry-standard equipment switchover methods, equipment replacement is only implemented after production equipment malfunctions and training tasks are interrupted. Taking action after a production equipment failure may result in the inability to switch over due to the lack of prior deployment of backup equipment. Secondly, if multiple production devices fail, insufficient backup equipment and personnel resources may be needed for switching, paralyzing training tasks. Furthermore, switching over equipment after a production failure requires establishing a full HCCL link between the backup device and other devices performing training tasks, which is time-consuming. Moreover, loading backup data from the most recent checkpoint to continue training necessitates re-running the training tasks between the most recent checkpoint and the failure point, leading to prolonged training interruptions and reduced efficiency. This efficiency decreases further as the scale of training tasks increases.

[0055] Based on this, this application provides a device switching method. By acquiring the status parameters of the production equipment during the execution of the training task, and determining that a production equipment malfunction may occur, a first backup device is used to replace the production equipment to perform the training task, without interrupting the training task. This avoids the long switching time and low execution efficiency of the training task caused by the interruption of the training task due to a production equipment malfunction. It shortens the switching time of the equipment performing the training task and improves the execution efficiency of the training task.

[0056] The solutions provided by the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0057] The solution provided in this application can be applied to Figure 2 An example of a computer system. (e.g.) Figure 2 The illustrated computer system architecture diagram includes a server 200. Alternatively, the computer system includes a terminal 100 and a server 200. The device switching method provided in this embodiment can be executed by the terminal 100, the server 200, or both the terminal 100 and the server 200; this embodiment does not limit the execution of this method.

[0058] Terminal 100 can be an electronic device such as a mobile phone, tablet computer, in-vehicle terminal (vehicle infotainment system), wearable device, personal computer (PC), or vehicle terminal. A client application for the target application can be installed and run on terminal 100. This target application can be an application referencing the device switching method, or it can be other applications providing device switching functionality; this application does not limit the specific application. Furthermore, this application does not limit the form of the target application, including but not limited to applications (Apps), mini-programs, etc., installed on terminal 100, and it can also be in web page form.

[0059] Server 200 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services such as cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing services such as big data. Server 200 can be a backend server for the aforementioned target application, used to provide backend services to the clients of the target application.

[0060] In some embodiments, the server 200 described above can also be implemented as a node in a blockchain system. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and cryptographic algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0061] Terminal 100 and server 200 can communicate via a network, such as a wired or wireless network.

[0062] In an exemplary embodiment of this application, server 200 follows Figure 4 or Figure 5 The device switching method shown is used to complete the device switching.

[0063] Those skilled in the art should understand that the above-described terminal 100 and server 200 are merely illustrative examples. Other existing or future terminals or servers that are applicable to this application should also be included within the scope of protection of this application, and are hereby incorporated by reference.

[0064] It should be noted that in the device switching method provided in this application embodiment, the executing entity for each step can be a computer device, which refers to an electronic device with data computing, processing, and storage capabilities. While the above-described device switching method is executed by server 200, those skilled in the art should understand that the device switching method can also be executed by terminal 100 alone (e.g., by a client of the target application installed and running on terminal 100), or by terminal 100 and server 200 interacting and cooperating to execute the device switching method. This application does not limit this approach.

[0065] Figure 3 This is a schematic diagram of another computer system architecture provided in an embodiment of this application. This system architecture can be applied to perform training tasks. Figure 3 As shown, the system architecture includes a scheduling system 301, an operation and maintenance system 302, a production equipment pool 303, and a backup equipment pool 304.

[0066] Both production equipment pool 303 and standby equipment pool 304 include servers, spine switches, and leaf switches. The servers and spine switches are connected to each other through leaf switches.

[0067] The scheduling system 301 is used to schedule the servers and switches in the production equipment pool 303. The operation and maintenance system 302 is used to manage the servers and switches in the production equipment pool 303 and the standby equipment pool 304.

[0068] For example, the operation and maintenance system 302 can obtain the status parameters of the devices in the production device pool 303 and the standby device pool 304 respectively, thereby obtaining the operating status of the devices in the production device pool 303 and the standby device pool 304. If it is determined from the status parameters that a production device in the production device pool 303 has a probability of failure, the scheduling system 301 can stop calling the production device, that is, isolate the production device, and add the standby device in the standby device pool 304 to the production device pool 303 for calling. Figure 3 The number of switches and servers shown is for illustrative purposes only and is not intended to limit the number of switches and servers.

[0069] Figure 4 This is a flowchart illustrating a device switching method provided in an exemplary embodiment of this application. The method can be executed by a computer device, which may be a server, such as... Figure 2 The server 200 is illustrated in the diagram. This method is used to execute tasks during the training process on production equipment. The processing procedure is the same for different production equipment; the following embodiment only describes the processing procedure of the computer equipment for one production equipment, and others will not be repeated. Figure 4 As shown, the method includes:

[0070] Step 402: Obtain the status parameters of the production equipment that is performing the training task.

[0071] In this context, production equipment refers to equipment that performs training tasks. Alternatively, production equipment refers to equipment that performs computational tasks within the training task. Specifically, production equipment can be a server.

[0072] The status parameters of production equipment are used to describe the operating status of the production equipment.

[0073] For example, the production equipment described in step 402 can be any production equipment managed by a computer device, and this application embodiment does not specifically limit it.

[0074] For example, the timing of acquiring the status parameters of the production equipment may include one or more of the following timings, but is not limited thereto, and the embodiments of this application do not specifically limit this:

[0075] • During the training process, periodically acquire the status parameters of the production equipment. For example, acquire the status parameters of the production equipment every 30 seconds during the training process.

[0076] • During the training process, the status parameters of each production device are acquired according to the preset acquisition frequency. For example, if the preset acquisition frequency of production device A is once every 30 seconds and the preset acquisition frequency of production device B is once every 15 seconds, then the status parameters of production device A are acquired once every 30 seconds, and the status parameters of production device B are acquired once every 15 seconds.

[0077] • During the training process, state parameters are acquired at different frequencies according to the weight parameters of each production device. For example, if the weight value of production device A is larger, the frequency of acquiring the state parameters of production device A will be correspondingly higher.

[0078] • During the training task, state parameters are acquired at different frequencies depending on the computational load of each production device. For example, if production device A has a large computational load during the training task, the frequency of acquiring its state parameters will increase accordingly.

[0079] For example, the device's status parameters may include one or more of the following: optical link transmit / receive power, optical link flashover frequency, and memory bit isolation status. Wherein:

[0080] Optical link transmit / receive power is an important indicator for evaluating the signal transmission performance in an optical signal link. For example, the transmit / receive power of an optical link can be used to determine the signal loss during transmission.

[0081] Optical link interruption frequency refers to the frequency at which brief interruptions occur in an optical fiber communication link. For example, if the optical link interruption frequency is greater than the interruption threshold, it indicates that the communication status of the optical link is poor.

[0082] Memory bit isolation, also known as high bandwidth memory (HBM) bit isolation, is an error detection and correction mechanism used to reflect multi-bit errors in memory. For example, memory is divided into multiple pages, each containing multiple bits. When a bit error is detected, the specific page where the error occurred is recorded, marked as unavailable, and isolated from memory. When the number of isolated bits exceeds a threshold, it indicates a memory anomaly.

[0083] Of course, the specific content of the device's status parameters can be configured according to actual needs. Any parameter that can reflect the device's operating status can be used as the status parameter described in this application. The specific content of the status parameters is not limited in the embodiments of this application.

[0084] Step 404: If the status parameters meet the troubleshooting conditions, switch from the production equipment to the first standby equipment to perform the training task.

[0085] Among them, the troubleshooting conditions are used to indicate the probability of equipment failure.

[0086] For example, after obtaining the status parameters of the production equipment, the production equipment is checked based on the status parameters to identify production equipment that meets the check criteria, i.e., production equipment with a probability of failure. After identifying the production equipment with a probability of failure, this production equipment is switched to the first backup equipment to perform the training task. Specifically, after identifying the production equipment with a probability of failure, the production equipment stops performing the training task, and the first backup equipment uses the backup data of the training task to continue performing the training task.

[0087] The first backup device is a backup device in an idle state, meaning it is a usable backup device. The first backup device is the equipment that can continue to perform training tasks in the event of a production equipment failure or risk of failure, and is used to replace the production equipment in performing training tasks.

[0088] Optionally, the backup equipment and the production equipment can be the same model of equipment. Alternatively, the backup equipment and the production equipment can be equipment capable of performing the same function. Alternatively, the backup equipment and the production equipment can be different models of equipment, but capable of performing the same function.

[0089] In one possible implementation, the first backup device can be any device in a backup device pool. The computer device can select any idle backup device from the backup device pool as the first backup device. Alternatively, the computer device can use a polling method to select an idle polling device from the backup device pool. This application embodiment does not limit the process of selecting the first backup device.

[0090] In another possible implementation, the first backup device is a device associated with the production equipment in the backup device pool.

[0091] For example, when configuring the backup device pool, an association relationship is established between each backup device and the production device. A backup device is used to perform the training task in place of its associated production device if the production device may fail.

[0092] For example, obtaining a first backup device from the backup device pool includes one or more of the following methods a to c, but is not limited thereto, and the embodiments of this application do not specifically limit this:

[0093] Method a: Randomly select a backup device from the backup device pool as the first backup device.

[0094] Method b: Sequentially acquire backup devices from the backup device pool as the first backup device.

[0095] Method c: Select the backup device associated with the production equipment from the backup device pool as the first backup device. For example, taking production equipment A as an example, configure a dedicated backup device corresponding to production equipment A in the backup device pool. Then, if the status parameters of production equipment A meet the troubleshooting conditions, the dedicated backup device corresponding to production equipment A will be used as the first backup device.

[0096] In one possible implementation, the screening condition is the operating parameters when the equipment malfunctions; if the status parameter reaches the operating parameter, then the status parameter satisfies the screening condition. For example, taking the optical link transmit / receive power as the status parameter, the operating parameter is set as a power threshold. After obtaining the optical link transmit / receive power of the production equipment, the optical link transmit / receive power is compared with the power threshold. If the optical link transmit / receive power is less than the power threshold, then the screening condition is met, and the production equipment is identified as meeting the screening condition.

[0097] For example, suppose the investigation conditions are that the optical link transmit / receive power is less than or equal to 20dBm and the optical link interruption frequency is greater than or equal to 50Hz. Suppose that the status parameters of production equipment A obtained at a certain moment are: optical link transmit / receive power 10dBm and optical link interruption frequency 100Hz. Then, it is determined that production equipment A meets the investigation conditions.

[0098] In another possible implementation, the screening condition includes a status score greater than or equal to a first threshold. Based on the status parameters, the production equipment is scored to obtain a status score; if the status score of the production equipment is greater than or equal to the first threshold, then the status parameters meet the screening condition.

[0099] In some embodiments, after obtaining the status parameters of multiple production devices, the multiple production devices are scored according to the status parameters to obtain a status score for each production device. The status scores of the multiple production devices are sorted from largest to smallest, and the production devices with the top n status scores are identified as production devices that meet the investigation criteria, that is, production devices that may fail. Here, n is a positive integer.

[0100] Optionally, the size of n can be determined in one or more of the following methods 1 to 3, but is not limited thereto, and the embodiments of this application do not specifically limit this:

[0101] Method 1: The size of n can be dynamically configured based on the number of backup devices in the backup device pool. For example, if there are only 5 backup devices in the backup device pool, the value of n can be configured to be less than 5 to ensure that there are enough backup devices for replacement.

[0102] Method 2: The size of n is a user-preset value or a default value.

[0103] Method 3: The value of n can be dynamically configured based on the number of production devices acquired. For example, if the status parameters of 12 production devices are acquired, and one-third of these devices are identified as meeting the screening criteria based on the status parameters, then n is 4.

[0104] Furthermore, in step 404, if the status parameters meet the screening conditions, the training task is switched from the production equipment to the first backup equipment. Specifically, if the status parameters meet the screening conditions, the production equipment stops executing the training task, and the first backup equipment uses the latest backup data of the training task to continue executing the training task.

[0105] Another possible implementation, in step 404, is as follows: if the status parameters meet the screening conditions, the training task is switched from the production equipment to the first backup equipment. Specifically, if the status parameters meet the screening conditions, after completing the checkpoint backup operation of the training task, the training task is switched from the production equipment to the first backup equipment. That is, after determining that the status parameters of the production equipment meet the screening conditions, the checkpoint backup operation of the training task is performed again, and then the backup data from this backup operation is used to switch to the first backup equipment to perform the training task.

[0106] For example, the step of switching the production equipment to the first standby equipment can be specifically implemented as follows: disconnecting the production equipment from other equipment and linking the first standby equipment to other equipment. Loading the equipment data of the production equipment into the first standby equipment. The first standby equipment uses the backup data to continue performing the training task.

[0107] Other equipment refers to equipment that communicates and / or exchanges data with production equipment during the execution of training tasks.

[0108] After identifying production equipment meeting the screening criteria during the training task, the first link node related to this training task is calculated using a pipeline data tensor (PDT) segmentation strategy. The first link node between the production equipment meeting the screening criteria and other equipment is disconnected, while the second link node between the production equipment meeting the screening criteria and other equipment is retained. After disconnecting the first link node, the first backup device is linked to other equipment through the first link node. That is, in this embodiment, when replacing the production equipment meeting the screening criteria with the first backup device, it is not necessary to disconnect all link nodes of the production equipment meeting the screening criteria; only the first link node related to this training task needs to be disconnected. Similarly, the first backup device is only linked to the first link node. This partial linking method allows for a faster switching speed and higher equipment switching efficiency between production equipment meeting the screening criteria and the first backup device.

[0109] In this training task, the first link node refers to the node that enables communication and / or data exchange between production equipment meeting the investigation criteria and other equipment. Alternatively, the first link node refers to the link node between production equipment meeting the investigation criteria and other equipment that is relevant to this training task. The second link node refers to the link node between production equipment meeting the investigation criteria and other equipment that is unrelated to this training task.

[0110] Optionally, the equipment data of the production equipment may include one or more of weight parameters, activation functions, and binary operators, but is not limited thereto, and the embodiments of this application do not specifically limit this.

[0111] In summary, the solution provided in this embodiment proposes a device switching method. By acquiring the status parameters of the production equipment during the execution of the training task, and determining that a production equipment failure is possible, a backup device is used to replace the production equipment to perform the training task, without interrupting the training task. This avoids the long switching time and low execution efficiency of the training task caused by switching to the backup device only after the production equipment failure and the training task interruption. It shortens the switching time of the equipment performing the training task and improves the execution efficiency of the training task.

[0112] Figure 5 This is a flowchart illustrating a device switching method provided in an exemplary embodiment of this application. The method can be executed by a computer device, which may be a server, such as... Figure 2 The server 200 is illustrated in the diagram. This method is used to execute tasks during the training process on production equipment. The processing procedure is the same for different production equipment; the following embodiment only describes the processing procedure of the computer equipment for one production equipment, and others will not be repeated. Figure 5 As shown, the method includes steps 502-508. Steps 502 and 504 are performed in parallel, that is, while the computer device is executing step 502, the computer device is also executing step 504.

[0113] In some embodiments, steps 502 and 504 may be performed alternately. For example, the computer device may perform step 504 after or during step 502. Alternatively, the computer device may perform step 502 after or during step 504. This application does not specifically limit the scope of these embodiments.

[0114] Step 502: Obtain the status parameters of the production equipment that is performing the training task.

[0115] For a description of this step, please refer to step 402; it will not be elaborated upon here.

[0116] Step 504: Obtain the status parameters of the standby devices in the standby device pool.

[0117] The backup equipment in the backup pool can also be called hot backup equipment. Hot backup equipment refers to backup equipment in the backup pool that is always in an active state and can be called up at any time during the production equipment's training tasks.

[0118] The backup equipment pool includes one or more backup devices. Optionally, the number of backup devices in the backup equipment pool can be configured based on the number of production devices performing the training task. For example, if five production devices are performing the training task, five backup devices can be configured in the backup equipment pool, thus enabling the backup equipment pool to cope with the extreme situation where all five production devices fail.

[0119] For example, the timing of acquiring the status parameters of the standby device includes one or more of the following timings, but is not limited thereto, and the embodiments of this application do not specifically limit this:

[0120] • During the training task, periodically acquire the status parameters of the backup device. For example, acquire the status parameters of the backup device every 30 seconds during the training task.

[0121] • During the training task, status parameters are acquired according to the preset acquisition frequency of each backup device. For example, if the preset acquisition frequency of backup device A is once every 30 seconds and the preset acquisition frequency of backup device B is once every 15 seconds, then the status parameters of backup device A are acquired once every 20 seconds, and the status parameters of backup device B are acquired once every 15 seconds.

[0122] • During the training process, the status parameters of the standby equipment are acquired according to the acquisition frequency of the production equipment's status parameters. That is, the status parameters of both the production equipment and the standby equipment are acquired simultaneously.

[0123] Step 506: Maintain the standby equipment in the standby equipment pool according to the status parameters of the standby equipment.

[0124] For example, after obtaining the status parameters of the backup devices, abnormal backup devices in the backup device pool are identified based on these parameters. The abnormal backup devices are then maintained according to their corresponding status parameters, ensuring that all backup devices in the backup device pool are always in normal operating condition and readily available for use.

[0125] Optionally, maintenance of the backup equipment includes: resetting and repairing the backup equipment according to its status parameters; and / or issuing alarms and warnings for the backup equipment according to its status parameters; and / or conducting performance evaluation tests on the backup equipment according to its status parameters.

[0126] For example, if the backup equipment has an optical link anomaly, the optical link of the backup equipment will be reset and repaired. If the optical link anomaly is not resolved after the reset and repair, an alarm will be issued to prompt the replacement of the components on the optical link as soon as possible.

[0127] If the backup device exhibits an increased memory bit isolation anomaly, HBM diagnostics are performed on the backup device. If the anomaly persists after diagnostics, a stress test is conducted, subjecting the backup device's memory to extreme operating conditions. The memory's performance under these extreme conditions is used to identify potential hardware problems or defects. If the stress test fails, an alarm is triggered to prompt prompt replacement of the faulty memory.

[0128] Step 508: If the status parameters meet the troubleshooting conditions, after completing the checkpoint backup operation of the training task, switch from the production equipment to the first standby equipment to execute the training task.

[0129] Among them, checkpoint backup operation refers to the operation of backing up the process and equipment of the training task.

[0130] For example, the operation time of a checkpoint backup operation includes one or more of the following, but is not limited thereto, and the embodiments of this application do not specifically limit it:

[0131] • Perform checkpoint backups periodically during training tasks. For example, perform a checkpoint backup every minute during training tasks.

[0132] • During the training mission, after identifying production equipment that meets the inspection criteria, a checkpoint backup operation is performed.

[0133] For example, during the execution of a training task, if the status parameters of the production equipment meet the checkpoint conditions, the production equipment is stopped from being called after the checkpoint backup operation of the training task is completed, and the first standby equipment is called. The first standby equipment continues to execute the training task based on the backup data backed up by the checkpoint backup operation. After the checkpoint backup operation of the training task is completed, the backup data at this point is up-to-date. Switching the equipment after this node can avoid the problem of repeated training caused by restarting the training task from the most recent checkpoint.

[0134] In some embodiments, after the checkpoint backup operation of the training task is completed, backup data is obtained. Production equipment that meets the screening conditions is isolated; that is, production equipment that meets the screening conditions stops executing the training task. Then, by disconnecting the production equipment from other equipment, a first standby equipment is linked to other equipment. The equipment data of the production equipment is loaded into the first standby equipment. The first standby equipment uses the backup data to continue executing the training task. The steps for switching the production equipment to the first standby equipment are described in step 404 and will not be elaborated further here.

[0135] In some embodiments, when the first backup device performs a training task in place of a production device that meets the screening conditions, the production device that meets the screening conditions is reset and repaired to obtain a repaired production device. A repaired production device refers to a production device that has been successfully reset and repaired. After the checkpoint backup operation of the training task is completed, the link between the first backup device and other devices is disconnected, and the link between the repaired production device and other devices is restored. The device data of the first backup device is loaded into the repaired production device. The repaired production device uses the backup data to continue performing the training task. When the first backup device performs a training task in place of a production device that meets the screening conditions, by repairing the production device that meets the screening conditions, and using the checkpoint backup operation as a time node, after the checkpoint backup operation is completed, the repaired production device is reused to continue performing the training task, thereby releasing the first backup device. This keeps the devices in the backup device pool in a dynamic balance, so that other production devices can be called upon when they meet the screening conditions.

[0136] For example, Figure 6 This is a schematic diagram illustrating a device switching method provided in an exemplary embodiment of this application. The solution provided in this application can be executed by a scheduling system and an operation and maintenance system. The scheduling system and the operation and maintenance system are located on a server, and this embodiment of the application does not limit their use. The scheduling system is used to schedule the use of production equipment and standby equipment. The operation and maintenance system is used to manage and maintain the production equipment and standby equipment.

[0137] Step 601: Inspect Production Equipment. The operation and maintenance system obtains the status parameters of the production equipment currently executing training tasks and inspects the equipment based on these parameters. If the status parameters meet the inspection criteria, the production equipment is identified as meeting these criteria, meaning it is identified as having a risk of failure. For example, if the status parameters of production equipment 3 in the production equipment pool meet the inspection criteria, then production equipment 3 is identified as having a risk of failure. Step 602: Inspect Backup Equipment. While the operation and maintenance system obtains the status parameters of the production equipment currently executing training tasks, it also obtains the status parameters of the backup equipment in the backup equipment pool. Based on these parameters, the system maintains the backup equipment, ensuring it is always in normal operating condition and can be readily used to replace production equipment. Step 603: Prepare for Equipment Switchover. After identifying the production equipment with a risk of failure, the system sends information to the first backup equipment in the backup equipment pool, initiating equipment switchover preparation for the first backup equipment. Step 604: Send Information about the Production Equipment Meeting the Inspection Criteria and the First Backup Equipment. The operations and maintenance system sends information about the production equipment meeting the investigation criteria and the first standby equipment to the scheduling system, enabling the scheduling system to know which production equipment in the production equipment pool needs to be replaced and which standby equipment in the standby equipment pool will be used for replacement. Step 605: Perform checkpoint backup operation. The scheduling system performs checkpoint backup operation periodically. After the operations and maintenance system receives information about the production equipment meeting the investigation criteria and the first standby equipment, it suspends the training task after completing the checkpoint backup operation. Step 606: Partially link the first standby equipment and synchronize data. After completing the checkpoint backup operation for the training task, the link between the production equipment meeting the investigation criteria and other equipment is disconnected, and the first standby equipment is partially linked to other equipment. Simultaneously, the equipment data of the production equipment meeting the investigation criteria is loaded into the first standby equipment to achieve data synchronization. Step 607: The first standby equipment replaces the production equipment meeting the investigation criteria to execute the training task. After data synchronization, the first standby equipment uses the backup data to continue executing the training task in place of the production equipment meeting the investigation criteria.

[0138] In summary, the solution provided in this embodiment proposes a device switching method that identifies potentially faulty production equipment during the execution of a training task. After identifying the potentially faulty production equipment, a checkpoint backup operation is used as the time node. After the checkpoint backup operation is completed, a first standby device replaces the production equipment at risk of failure to perform the training task. Since the identification of potentially faulty production equipment and the identification of the first standby device are both implemented during the execution of the training task, this avoids the long switching time and low execution efficiency of the training task caused by the interruption of the training task due to a production equipment failure. This shortens the switching time of the equipment performing the training task and improves the execution efficiency of the training task.

[0139] For example, Figure 7 This is a flowchart of a device switching method provided in an exemplary embodiment of this application. The solution provided in this application can be executed by a scheduling system and an operation and maintenance system. The scheduling system and the operation and maintenance system can be located in a server, and this embodiment of the application does not limit this. The scheduling system is used to schedule the use of production equipment and standby equipment. The operation and maintenance system is used to manage and maintain the production equipment and standby equipment.

[0140] Step 701: Conduct regular inspections of production equipment.

[0141] For example, the operation and maintenance system obtains the status parameters of the production equipment that is performing training tasks, and checks the production equipment based on the status parameters of the production equipment. The purpose is to identify the production equipment in the production equipment pool that has the risk of failure, that is, to identify the production equipment in the production equipment pool that has the probability of failure.

[0142] Step 702: Regularly inspect standby equipment.

[0143] For example, while the operation and maintenance system obtains the status parameters of the production equipment that is performing training tasks, the operation and maintenance system also obtains the status parameters of the backup equipment in the backup equipment pool. Based on the status parameters of the backup equipment, the backup equipment is investigated to identify faulty or abnormal backup equipment in the backup equipment pool.

[0144] Step 703: Perform maintenance on the backup equipment.

[0145] For example, after the operation and maintenance system identifies abnormal backup devices in the backup device pool, it maintains the abnormal backup devices according to their corresponding status parameters. This includes resetting and repairing the abnormal backup devices, issuing alarms and warnings, and conducting performance evaluation tests. This ensures that the backup devices in the backup device pool are always in normal operating condition, making them readily available to replace the generated devices.

[0146] Step 704: Evaluate the production equipment to identify those that meet the screening criteria.

[0147] For example, after obtaining the status parameters of multiple production devices, the multiple production devices are scored according to the status parameters to obtain a status score for each production device. The status scores of the multiple production devices are sorted from largest to smallest, and the production devices with the top n status scores are identified as production devices that meet the investigation conditions, that is, production devices with a risk of failure.

[0148] Step 705: Perform checkpoint backups periodically.

[0149] For example, during the execution of a training task, checkpoint backups are performed periodically. By performing checkpoint backups periodically, backup data can be obtained periodically.

[0150] Step 706: Suspension training task.

[0151] For example, after the checkpoint backup operation is completed, the training task is suspended based on the checkpoint backup operation as the time node.

[0152] That is, the training task is temporarily paused, and at this time, the data from each device is up-to-date.

[0153] Step 707: Isolate production equipment that meets the screening criteria.

[0154] For example, after suspending the training task, the link between the production equipment that meets the investigation conditions and other equipment is disconnected, and the first backup equipment is linked to other equipment.

[0155] When disconnecting the production equipment that meets the investigation conditions from other equipment, only the link nodes related to this training task between the production equipment that meets the investigation conditions and other equipment should be disconnected; it is not necessary to disconnect all nodes of the production equipment that meets the investigation conditions.

[0156] Step 708: Complete the partial connection of the first backup device.

[0157] For example, after disconnecting the link nodes related to this training task between the production equipment that meets the investigation conditions and other equipment, the first backup equipment is linked to the link nodes related to this training task, thereby realizing the partial linking of the first backup equipment.

[0158] Step 709: Synchronize the equipment data of production equipment that meets the investigation conditions.

[0159] For example, after partially linking the first backup device with the production device that meets the screening conditions, the equipment data of the production device that meets the screening conditions is loaded into the first backup device to achieve data synchronization.

[0160] Step 710: Unsustain the training task and continue executing the training task.

[0161] For example, after the first backup device is partially linked to the production device that meets the screening conditions and the data is synchronized, the first backup device uses the backup data to continue performing the training task in place of the production device that meets the screening conditions.

[0162] The foregoing mainly describes the solution provided in this application. Accordingly, this application also provides a device switching apparatus for implementing the above-described method embodiments.

[0163] In some embodiments, the device switching apparatus includes hardware structures and / or software modules corresponding to the execution of each function in order to achieve the above-described functions. Those skilled in the art will readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0164] This application embodiment can divide the device switching device into functional modules according to the above method embodiment. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0165] In some embodiments, this application provides a device switching apparatus, which is used to implement the functions of the device switching apparatus in the above method embodiments. For example... Figure 8 The diagram shows the structure of the equipment switching device. The equipment switching device may include an acquisition module 801, a switching module 802, and a maintenance module 803.

[0166] Module 801 is used for execution Figure 4 The illustrated method includes step 402, and the execution of... Figure 5 The illustrated method includes steps 502 and 504. Switching module 802 is used to execute... Figure 4The illustrated method includes step 404, and the execution of... Figure 5 The illustrated method includes step 508. Maintenance module 803 is used to execute... Figure 5 The illustrated method includes step 506.

[0167] like Figure 9 As shown, the computer device provided in this application embodiment may include a processor 901, a bus 902, a communication interface 903, and a memory 904. The processor 901, memory 904, and communication interface 903 communicate with each other via the bus 902. It should be understood that this application does not limit the number of processors and memories in the computer device.

[0168] The 902 bus can be a PCI bus, an Extended Industry Standard Architecture (EISA) bus, or a UB bus, etc. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 8 The bus 902 is represented by only one line, but this does not mean that there is only one bus or one type of bus. The bus 902 may include a path for transmitting information between various components of a computer device (e.g., memory 904, processor 901, communication interface 903).

[0169] Processor 901 may include any one or more processors such as CPU, graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0170] The memory 904 may include volatile memory, such as random access memory (RAM). The processor 901 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0171] The communication interface 903 uses transceiver modules, such as, but not limited to, network interface cards and transceivers, to enable communication between computer devices and other devices or communication networks.

[0172] The memory 904 stores executable program code, which the processor 901 executes to implement the functions of the device switching device or the CPU core in the aforementioned method embodiments. That is, the memory 904 stores instructions for executing the aforementioned device switching method.

[0173] In another aspect, embodiments of this application provide a computer device, the computer device including: a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the device switching method provided in the above method embodiments.

[0174] In another aspect, a computer-readable storage medium is provided, wherein at least one computer program is stored in the computer-readable storage medium, and the at least one computer program is loaded and executed by a processor to implement the device switching method as provided in the above-described method embodiments.

[0175] In another aspect, a computer program product is provided, comprising a computer program stored in a computer-readable storage medium; the computer program is read from and executed by a processor of a computer device from the computer-readable storage medium, causing the computer device to perform the device switching method provided in the above-described method embodiments.

[0176] The method steps in this embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Alternatively, the ASIC can reside in a computer device. Of course, the processor and storage medium can also exist as discrete components in the computer device.

[0177] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video disc (DVD); or it can be a semiconductor medium, such as a solid-state drive (SSD). The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A device switching method, characterized in that, The method includes: Obtain the status parameters of the production equipment that is performing a training task, the status parameters being used to describe the operating status of the production equipment; If the status parameters meet the screening conditions, the training task is performed by switching from the production equipment to the first backup equipment. The troubleshooting conditions are used to indicate the probability of a device malfunctioning, and the first backup device is a backup device in an idle state.

2. The method according to claim 1, characterized in that, If the status parameters meet the screening conditions, switching from the production equipment to the first backup equipment to execute the training task includes: If the status parameters meet the troubleshooting conditions, after the checkpoint backup operation of the training task is completed, the training task is switched from the production equipment to the first backup equipment.

3. The method according to claim 2, characterized in that, After completing the checkpoint backup operation of the training task, switching from the production equipment to the first standby equipment to execute the training task includes: After the checkpoint backup operation of the training task is completed, backup data is obtained; The production equipment stops executing the training task, and the first backup equipment uses the backup data to continue executing the training task.

4. The method according to claim 3, characterized in that, The production equipment stops executing the training task, and the first backup equipment uses the backup data to continue executing the training task, including: Disconnect the production equipment from other equipment and connect the first backup equipment to the other equipment, wherein the other equipment refers to the equipment that communicates and / or exchanges data with the production equipment during the execution of the training task; Load the equipment data of the production equipment into the first backup equipment; The first backup device uses the backup data to continue performing the training task.

5. The method according to any one of claims 1 to 4, characterized in that, The troubleshooting conditions are the operating parameters when the equipment malfunctions; the status parameters that satisfy the troubleshooting conditions include: The status parameters reach the operating parameters.

6. The method according to any one of claims 1 to 4, characterized in that, The screening criteria include a status score greater than or equal to a first threshold; The status parameters satisfy the investigation conditions, including: The production equipment is scored based on the state parameters to obtain a state score for the production equipment. The status score of the production equipment is greater than or equal to the first threshold.

7. The method according to any one of claims 1 to 6, characterized in that, The first backup device belongs to the backup device pool; the method further includes: Obtain the status parameters of the standby devices in the standby device pool; Maintain the standby devices in the standby device pool based on the status parameters of the standby devices.

8. The method according to claim 7, characterized in that, Maintaining the standby devices in the standby device pool based on the status parameters of the standby devices includes one or more of the following operations: The backup equipment is reset and repaired based on its status parameters. Based on the status parameters of the backup equipment, an alarm and early warning will be issued for the backup equipment; Based on the status parameters of the backup equipment, a performance evaluation test is performed on the backup equipment.

9. A device switching apparatus, characterized in that, The device includes: The acquisition module is used to acquire the status parameters of the production equipment that is executing the training task, and the status parameters are used to describe the operating status of the production equipment. A switching module is used to switch from the production equipment to the first backup equipment to perform the training task if the status parameters meet the investigation conditions. The troubleshooting conditions are used to indicate the probability of a device malfunctioning, and the first backup device is a backup device in an idle state.

10. A computer device, characterized in that, The computer device includes a processor and a memory, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the device switching method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor to implement the device switching method as described in any one of claims 1 to 8.

12. A computer program product, characterized in that, The computer program product includes a computer program stored in a computer-readable storage medium; the computer program is read from and executed by a processor of a computer device from the computer-readable storage medium, causing the computer device to perform the device switching method as described in any one of claims 1 to 8.

Images