Method and apparatus for processing hbm parameters

Abstract

The application provides a processing method and device of HBM parameters, the method is applied to a switching chip, the switching chip comprises a DRAM composed of an HBM, the switching chip is in a single board, and the method comprises the following steps: when the switching chip is restarted, it is judged whether the HBM parameters are retrained this time; if it is determined that the HBM parameters are not retrained this time, it is judged whether an HBM parameter storage path is acquired; if the HBM parameter storage path is acquired, historical HBM parameters are acquired from a storage unit corresponding to the HBM parameter storage path, and the historical HBM parameters are loaded; if the historical HBM parameters are successfully loaded, the DRAM is initialized and configured, and the restarting is continued.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a method and apparatus for processing HBM parameters. Background Technology

[0002] Dynamic Random Access Memory (DRAM) is an external memory module on a chip used to store packets when there is congestion on the board, thereby buffering the packets and reducing packet loss.

[0003] The DRAM consists of two High Bandwidth Memory (HBM) modules. During startup, the chip needs to train some HBM parameters based on the current environment of the board, such as temperature, pressure, and version number. These parameters affect the HBM function, which in turn affects packet storage and forwarding within the chip. Therefore, the chip retrains the HBM parameters each time the board restarts, based on the current environment.

[0004] like Figure 1 As shown, Figure 1 The flowchart illustrates the process of training HBM parameters for existing chips. Training HBM parameters for a single chip takes approximately 3 minutes. If a board contains multiple chips, training HBM parameters for each chip will significantly increase the board's startup time, making it extremely long. Therefore, retraining HBM parameters for each chip on each board every time the network device restarts would waste too much time, severely impacting the network device's restart speed. Summary of the Invention

[0005] In view of this, this application provides a method and apparatus for processing HBM parameters to solve the problem that the training time for HBM parameters of existing chips is long, resulting in excessively long single-board startup time.

[0006] In a first aspect, this application provides a method for processing HBM parameters, the method being applied to a switching chip, the switching chip including DRAM composed of HBM, the switching chip being located within a single board, the method comprising:

[0007] When the switching chip restarts, it is determined whether the HBM parameters need to be retrained during this restart;

[0008] If it is determined that the HBM parameters will not be retrained in this restart, then determine whether the HBM parameter storage path has been obtained;

[0009] If the HBM parameter storage path is obtained, then the historical HBM parameters are retrieved from the storage unit corresponding to the HBM parameter storage path and loaded.

[0010] If the historical HBM parameters are loaded successfully, the DRAM is initialized and configured, and the reboot process continues.

[0011] Secondly, this application provides an apparatus for processing HBM parameters, the apparatus being applied to a switching chip, the switching chip including DRAM composed of HBM, the switching chip being located within a single board, the apparatus comprising:

[0012] The first judgment unit is used to determine whether the HBM parameters need to be retrained during the restart when the switching chip restarts.

[0013] The second judgment unit is used to determine whether the HBM parameter storage path has been obtained if it is determined that the HBM parameters will not be retrained in this restart.

[0014] The first acquisition unit is used to, if the HBM parameter storage path is acquired, retrieve historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and load the historical HBM parameters.

[0015] An initialization unit is used to initialize the DRAM and continue the restart process if the historical HBM parameters are successfully loaded.

[0016] Thirdly, this application provides a network device including a processor and a machine-readable storage medium storing machine-executable instructions that can be executed by the processor, which in turn cause the processor to perform the method provided in the first aspect of this application.

[0017] Therefore, by applying the HBM parameter processing method and apparatus provided in this application, when the switching chip restarts, the switching chip determines whether the HBM parameters need to be retrained during this restart; if it is determined that the HBM parameters will not be retrained during this restart, the switching chip determines whether the HBM parameter storage path has been obtained; if the switching chip obtains the HBM parameter storage path, the switching chip retrieves the historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and loads the historical HBM parameters; if the historical HBM parameters are successfully loaded, the switching chip initializes the DRAM configuration and continues to restart.

[0018] In this way, if the chip determines that it will not retrain the HBM parameters upon restart, it can save chip startup time by acquiring and loading historical HBM parameters, while ensuring the adaptability of the HBM parameters. This solves the problem of excessive startup time for single-board systems with multiple chips due to the long HBM parameter training time of existing chips. Attached Figure Description

[0019] Figure 1 Flowchart for training HBM parameters for existing chips;

[0020] Figure 2 A flowchart illustrating the method for processing HBM parameters provided in this application embodiment;

[0021] Figure 3 A structural diagram of the HBM parameter processing device provided in the embodiments of this application;

[0022] Figure 4 The network device hardware structure provided in the embodiments of this application. Detailed Implementation

[0023] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0024] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the corresponding listed items.

[0025] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0026] The method for processing HBM parameters provided in the embodiments of this application will be described in detail below. See also Figure 2 , Figure 2 This is a flowchart illustrating a method for processing HBM parameters provided in an embodiment of this application. This method is applied to a switching chip. The method for processing HBM parameters provided in an embodiment of this application may include the following steps.

[0027] Step 210: When the switching chip restarts, determine whether the HBM parameters need to be retrained during this restart;

[0028] Specifically, when the switching chip starts to restart, it first initializes the DRAM and calls a judgment function to determine whether the HBM parameters need to be retrained.

[0029] The switching chip first acquires historical and current record values. Historical record values ​​include the historical version number, historical temperature value, and historical magic letter; current record values ​​include the current version number, current temperature value, and current magic letter. Then, the switching chip uses the acquired historical and current record values ​​to determine whether the HBM parameters need to be retrained.

[0030] In this embodiment, the historical record value is the version number, temperature value, and magic word that match the HBM parameter recorded during the last training of the HBM parameter (the historical record value can be stored in the storage unit corresponding to the HBM parameter storage path, or in a separate storage unit for storing historical record values. In this embodiment, the historical record value is stored in the storage unit corresponding to the HBM parameter storage path as an example).

[0031] The current recorded values ​​are the version number, temperature value, and magic text obtained this time. The version number and temperature value can be obtained from the device's hardware resources, while the magic text can be obtained from the device's software resources.

[0032] In this embodiment of the application, the switching chip determines whether the HBM parameters need to be retrained based on the following three conditions: Condition 1: whether the historical version number matches the current version number; Condition 2: whether the historical temperature value matches the current temperature value; Condition 3: whether the historical magic word matches the current magic word.

[0033] If all three conditions are met, the switching chip determines that the HBM parameters will not be retrained during this restart; otherwise, if any one of the three conditions is not met, the switching chip determines that the HBM parameters will be retrained during this restart.

[0034] It should be noted that when the switching chip judges the three judgment conditions respectively, the above-mentioned matching or non-matching refers to whether the two parameters are the same or different, that is, whether the current version number is the same as the historical version number, whether the current temperature value is the same as the historical temperature value, and whether the current magic word is the same as the historical magic word.

[0035] Furthermore, any mismatch among the three judgment conditions mentioned above can trigger the retraining of HBM parameters.

[0036] For example, the version number referred to in condition 1 is the version number of the single board, including upgrade and downgrade version numbers. Different version numbers require different HBM parameters. In this case, the switching chip is restarted and the HBM parameters are retrained.

[0037] The temperature value referred to in condition 2 is monitored by HBM's own temperature monitoring system. Different temperature values ​​require different HBM parameters. In this case, the switching chip is restarted and the HBM parameters are retrained.

[0038] Of course, the temperature monitoring system can also determine whether the current temperature value exceeds or falls below the temperature threshold. If the current temperature value exceeds the first temperature threshold or falls below the second temperature threshold, it indicates that the temperature change of the current environment of the board is too large. At this time, the HBM parameters are no longer suitable for the current environment, which will lead to DRAM failure or packet loss. The switching chip needs to be restarted and the HBM parameters retrained.

[0039] The magic word mentioned in condition 3 is a newly added global variable in the code. When all other conditions are met, the magic word can be modified to force a retraining of the HBM parameters, increasing flexibility and facilitating problem localization.

[0040] For example, the magic word might be a specific value, such as 133. The magic word can be changed based on the current status of the board. For instance, if the magic word value is 133, and other recorded values ​​remain unchanged (version number, temperature value, etc.), and packet loss occurs in the network device, the HBM parameters may no longer be suitable for the network device. The switching chip can then update the magic word value to 134.

[0041] Step 220: If it is determined that the HBM parameters will not be retrained in this restart, then determine whether the HBM parameter storage path has been obtained;

[0042] Specifically, according to the description in step 210, if it is determined that the HBM parameters will not be retrained during this restart, the switching chip determines whether it has obtained the HBM parameter storage path. If the HBM parameter storage path is obtained, then step 230 is executed.

[0043] Step 230: If the HBM parameter storage path is obtained, retrieve the historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and load the historical HBM parameters.

[0044] Specifically, according to the description of step 220, if the HBM parameter storage path is obtained, the switching chip obtains the historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and loads the historical HBM parameters.

[0045] If the historical HBM parameters are loaded successfully, proceed to step 240.

[0046] Specifically, the aforementioned storage unit can be a storage structure within Flash memory.

[0047] Step 240: If the historical HBM parameters are loaded successfully, initialize the DRAM configuration and continue the restart process.

[0048] Specifically, according to the description of step 230, if the historical HBM parameters are successfully retrieved, the switching chip initializes the DRAM configuration and continues to restart the switching chip.

[0049] It should be noted that, in the embodiments of this application, when the DRAM external to the switching chip includes HBM, the switching chip can execute the various processes described in the embodiments of this application. The embodiments of this application do not limit the specific type or series of switching chips. A single board specifically refers to a card inserted into the switch chassis, and the single board includes a hardware-fixed forwarding chip.

[0050] Therefore, by applying the HBM parameter processing method provided in this application, when the switching chip restarts, the switching chip determines whether the HBM parameters should be retrained during this restart; if it is determined that the HBM parameters will not be retrained during this restart, the switching chip determines whether the HBM parameter storage path has been obtained; if the switching chip obtains the HBM parameter storage path, the switching chip retrieves the historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and loads the historical HBM parameters; if the historical HBM parameters are successfully loaded, the switching chip initializes the DRAM configuration and continues to restart.

[0051] In this way, if the chip determines that it will not retrain the HBM parameters upon restart, it can save chip startup time by acquiring and loading historical HBM parameters, while ensuring the adaptability of the HBM parameters. This solves the problem of excessive startup time for single-board systems with multiple chips due to the long HBM parameter training time of existing chips.

[0052] Optionally, before step 210 in the embodiments of this application, the process of training HBM parameters, creating HBM parameter storage paths, and storing the trained HBM parameters in the corresponding storage unit is further included when the switching chip is first started.

[0053] Specifically, when the switching chip is started for the first time, it obtains the original recorded values ​​of the board at the time of the first startup. These original recorded values ​​include the original version number, original temperature value, and original magic letter of the board.

[0054] The switching chip trains the HBM parameters based on the original version number, original temperature value, and original magic word. The switching chip creates an HBM parameter storage path and stores the trained HBM parameters, along with the original version number, original temperature value, and original magic word, in the corresponding storage.

[0055] It is understandable that the HBM parameters stored above are the same as the historical HBM parameters mentioned in the preceding steps.

[0056] Optionally, in this embodiment of the application, the process of the switching chip determining that the HBM parameters will be retrained during this restart is also included.

[0057] Specifically, according to the description of step 210, if it is determined that the HBM parameters will be retrained during this restart, the switching chip will train the HBM parameters based on the current version number, the current temperature value, and the current magic word.

[0058] It is understandable that the process of training HBM parameters by the switching chip is the same as the existing training process, and will not be repeated here.

[0059] After the switching chip completes the training of the HBM parameters, it replaces the historical HBM parameters with the trained HBM parameters, and replaces the historical version number, historical temperature value, and historical magic word with the current version number, current temperature value, and current magic word.

[0060] In other words, the switching chip deletes the historical HBM parameters and historical values ​​already stored in the storage unit, and stores the trained HBM parameters, current version number, current temperature value, and current magic word in the storage unit. The trained HBM parameters, current version number, current temperature value, and current magic word can be used as historical HBM parameters and historical values ​​to be obtained during the next restart.

[0061] Optionally, this application embodiment also includes a process where the switching chip fails to obtain the HBM parameter storage path, or the switching chip fails to successfully load the historical HBM parameters.

[0062] Specifically, if the switching chip fails to obtain the HBM parameter storage path, or if the switching chip fails to load the historical HBM parameters successfully, the switching chip will train the HBM parameters based on the current version number, the current temperature value, and the current magic word.

[0063] It is understandable that the process of training HBM parameters by the switching chip is the same as the existing training process, and will not be repeated here.

[0064] After the switching chip completes the training of the HBM parameters, it replaces the historical HBM parameters with the trained HBM parameters, and replaces the historical version number, historical temperature value, and historical magic word with the current version number, current temperature value, and current magic word.

[0065] In other words, the switching chip deletes the historical HBM parameters and historical values ​​already stored in the storage unit, and stores the trained HBM parameters, current version number, current temperature value, and current magic word in the storage unit. The trained HBM parameters, current version number, current temperature value, and current magic word can be used as historical HBM parameters and historical values ​​to be obtained during the next restart.

[0066] Optionally, in this embodiment of the application, a process is also included to force the switching chip to restart if the temperature of the environment where the board is located changes significantly after the switching chip restarts.

[0067] Specifically, after step 230, once the switching chip has restarted, HBM's own temperature monitoring system periodically detects the current temperature of the environment in which the board is located.

[0068] After the temperature monitoring system obtains the current temperature value of the environment in which the board is located, it determines whether the current temperature value exceeds the first temperature threshold (for example, the first temperature threshold may be 60 degrees), or whether the current temperature value is lower than the second temperature threshold (for example, the second temperature threshold may be 0 degrees).

[0069] If the current temperature exceeds the first temperature threshold, or if the current temperature is lower than the second temperature threshold, the temperature monitoring system determines that the temperature change of the current environment of the single board is too large. At this time, the HBM parameters existing in the switching chip are no longer applicable to the current environment, which will lead to DRAM failure or packet loss.

[0070] The temperature monitoring system generates an alarm message and sends it to the switching chip. This alarm message includes HBM parameters that are above or below a threshold. For example, the temperature value that is above or below the threshold.

[0071] After receiving the alarm information, the switching chip forces itself to restart based on the HBM parameters included in the alarm information.

[0072] Understandably, the restart operation of the switching chip can refer to the process described in steps 210-240 above, and will not be repeated here. At this time, if the switching chip determines that this restart will retrain the HBM parameters, replace the historical HBM parameters with the trained HBM parameters, and replace the historical version number, historical temperature value, and historical magic word with the current version number, current temperature value, and current magic word, it will ensure that the DRAM can be used normally.

[0073] Based on the same inventive concept, embodiments of this application also provide an HBM parameter processing apparatus corresponding to the HBM parameter processing method. See [link to related documentation]. Figure 3 , Figure 3The HBM parameter processing apparatus provided in this application embodiment is applied to a switching chip, the switching chip including DRAM composed of HBM, the switching chip being located within a single board, the apparatus comprising:

[0074] The first judgment unit 310 is used to determine whether the HBM parameters need to be retrained during the restart when the switching chip restarts.

[0075] The second judgment unit 320 is used to determine whether the HBM parameter storage path has been obtained if it is determined that the HBM parameters will not be retrained in this restart.

[0076] The first acquisition unit 330 is used to, if the HBM parameter storage path is acquired, retrieve historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and load the historical HBM parameters.

[0077] The initialization unit 340 is used to initialize the DRAM and continue the restart process if the historical HBM parameters are successfully loaded.

[0078] Optionally, the first determination unit 310 is specifically used to obtain historical record values ​​and current record values. The historical record values ​​include historical version numbers, historical temperature values, and historical magic words. The current record values ​​include current version numbers, current temperature values, and current magic words.

[0079] Determine whether the current version number, the current temperature value, and the current magic word match the corresponding historical version number, the historical temperature value, and the historical magic word, respectively;

[0080] If the current version number matches the historical version number, the current temperature value matches the historical temperature value, and the current magic word matches the historical magic word, then it is determined that the HBM parameters will not be retrained during this restart.

[0081] If any one of the current version number, the current temperature value, and the current magic word does not match the corresponding historical version number, the historical temperature value, and the historical magic word, then it is determined that the HBM parameters will be retrained in this restart.

[0082] Optionally, the device further includes: a training unit (not shown in the figure), configured to train the HBM parameters based on the current version number, the current temperature value, and the current magic word if it is determined that the HBM parameters will be retrained during this restart;

[0083] An update unit (not shown in the figure) is used to replace the historical HBM parameters with the trained HBM parameters, and to replace the historical version number, the historical temperature value, and the historical magic word with the current version number, the current temperature value, and the current magic word.

[0084] Optionally, the device further includes: a second acquisition unit (not shown in the figure), used to acquire the original recorded values ​​of the board at the time of the first startup when the switching chip is first started, the original recorded values ​​including the original version number, original temperature value and original magic letter of the board;

[0085] The training unit (not shown in the figure) is also used to train the HBM parameters based on the original version number, the original temperature value, and the original magic word.

[0086] A creation unit (not shown in the figure) is used to create the HBM parameter storage path and store the trained HBM parameters, the original version number, the original temperature value, and the original magic word in the storage unit.

[0087] Optionally, the training unit (not shown in the figure) is further configured to train the HBM parameters based on the current version number, the current temperature value, and the current magic word if the HBM parameter storage path is not obtained, or if the historical HBM parameters are not successfully loaded.

[0088] The update unit (not shown in the figure) is also used to replace the historical HBM parameters with the trained HBM parameters, and to replace the historical version number, the historical temperature value, and the historical magic word with the current version number, the current temperature value, and the current magic word.

[0089] Optionally, the device further includes a restart unit (not shown in the figure), used to restart the switching chip again according to the alarm information if an alarm information is received, the alarm information including a current recorded value that exceeds or falls below a threshold.

[0090] Therefore, by applying the HBM parameter processing method and apparatus provided in this application, when the switching chip restarts, the switching chip determines whether the HBM parameters need to be retrained during this restart; if it is determined that the HBM parameters will not be retrained during this restart, the switching chip determines whether the HBM parameter storage path has been obtained; if the switching chip obtains the HBM parameter storage path, the switching chip retrieves the historical HBM parameters from the storage unit corresponding to the HBM parameter storage path and loads the historical HBM parameters; if the historical HBM parameters are successfully loaded, the switching chip initializes the DRAM configuration and continues to restart.

[0091] In this way, if the chip determines that it will not retrain the HBM parameters upon restart, it can save chip startup time by acquiring and loading historical HBM parameters, while ensuring the adaptability of the HBM parameters. This solves the problem of excessive startup time for single-board systems with multiple chips due to the long HBM parameter training time of existing chips.

[0092] Based on the same inventive concept, embodiments of this application also provide a network device, such as... Figure 4 As shown, the system includes a processor 410, a transceiver 420, and a machine-readable storage medium 430. The machine-readable storage medium 430 stores machine-executable instructions that can be executed by the processor 410. The processor 410 is prompted by the machine-executable instructions to execute the HBM parameter processing method provided in the embodiments of this application. (The foregoing...) Figure 3 The HBM parameter processing device shown can be as follows: Figure 4 The hardware structure of the network device shown is implemented.

[0093] The aforementioned computer-readable storage medium 430 may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the computer-readable storage medium 430 may also be at least one storage device located remotely from the aforementioned processor 410.

[0094] The processor 410 mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0095] In this embodiment of the application, the processor 410 reads the machine-executable instructions stored in the machine-readable storage medium 430, and is prompted by the machine-executable instructions to enable the processor 410 itself and the transceiver 420 to execute the HBM parameter processing method described in the aforementioned embodiment of the application.

[0096] In addition, this application provides a machine-readable storage medium 430 that stores machine-executable instructions. When called and executed by the processor 410, the machine-executable instructions cause the processor 410 itself and the transceiver 420 to execute the HBM parameter processing method described in the aforementioned application.

[0097] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0098] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0099] For the embodiments of the HBM parameter processing apparatus and the machine-readable storage medium, since the methods involved are basically similar to those in the aforementioned method embodiments, the description is relatively simple, and relevant details can be found in the descriptions of the method embodiments.

[0100] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method of processing HBM parameters, the method comprising: The method is applied to a switch chip, the switch chip comprises a DRAM composed of an HBM, the switch chip is in a single board, and the method comprises: When the switch chip is restarted, it is judged whether the current restart is for retraining of HBM parameters; If it is determined that the current restart is not for retraining of HBM parameters, it is judged whether an HBM parameter storage path is obtained; If the HBM parameter storage path is obtained, historical HBM parameters are obtained from a storage unit corresponding to the HBM parameter storage path, and the historical HBM parameters are loaded; If the historical HBM parameters are successfully loaded, the DRAM is initialized and configured, and the restart is continued; The method further comprises: If it is determined that the current restart is for retraining of the HBM parameters, the HBM parameters are trained according to the current version number, the current temperature value and the current magic word; The trained HBM parameters replace the historical HBM parameters, and the current version number, the current temperature value and the current magic word replace the historical version number, the historical temperature value and the historical magic word. Before the step of, when the switch chip is restarted, it is judged whether the current restart is for retraining of HBM parameters, the method further comprises: When the switch chip is started for the first time, original record values of the single board at the first start are obtained, the original record values comprising original version numbers, original temperature values and original magic words of the single board; 2. The method of claim 1, wherein, The original record values are stored in a storage unit corresponding to the HBM parameter storage path. ​ ​ 3. The method of claim 2, wherein, ​ ​ ​ 4. The method of claim 1, wherein, ​ ​ According to the original version number, the original temperature value, and the original magic word, the HBM parameter is trained. The HBM parameter storage path is created, and the trained HBM parameter and the original version number, the original temperature value, and the original magic word are stored in the storage unit.

5. The method of claim 2, wherein, The method further comprises: If the HBM parameter storage path is not acquired, or if the historical HBM parameter is not loaded successfully, according to the current version number, the current temperature value, and the current magic word, the HBM parameter is trained. The trained HBM parameter replaces the historical HBM parameter, and the current version number, the current temperature value, and the current magic word replace the historical version number, the historical temperature value, and the historical magic word.

6. The method of claim 1, wherein, The method further comprises: If the alarm information is received, the switching chip is restarted again according to the alarm information, and the alarm information includes a current record value that exceeds or is lower than a threshold value.

7. A processing device of HBM parameters, characterized in that, The device is applied to a switching chip, the switching chip includes a DRAM composed of an HBM, and the switching chip is in a single board. A first judgment unit is configured to determine whether the HBM parameter is retrained when the switching chip is restarted. A second judgment unit is configured to determine whether the HBM parameter storage path is acquired if it is determined that the HBM parameter is not retrained this time. A first acquisition unit is configured to acquire the historical HBM parameter from a storage unit corresponding to the HBM parameter storage path and load the historical HBM parameter if the HBM parameter storage path is acquired. An initialization unit is configured to initialize and configure the DRAM and continue to restart if the historical HBM parameter is loaded successfully. The first judgment unit is specifically configured to determine whether the HBM parameter is retrained when the switching chip is restarted, and specifically includes: Acquiring a historical record value and a current record value, the historical record value including a historical version number, a historical temperature value, and a historical magic word, and the current record value including a current version number, a current temperature value, and a current magic word. Determining whether the current version number, the current temperature value, and the current magic word match the corresponding historical version number, historical temperature value, and historical magic word, respectively. If the current version number matches the historical version number, the current temperature value matches the historical temperature value, and the current magic word does not match the historical magic word, it is determined that the HBM parameter is retrained this time. The magic word is a global variable newly added in the code, and in the case that the current temperature value and the historical temperature value match, the HBM parameter is forcibly retrained by modifying the magic word.

8. The apparatus of claim 7, wherein, The first judging unit is specifically configured to determine that the HBM parameter is not retrained this time if the current version number matches the historical version number, the current temperature value matches the historical temperature value, and the current magic word matches the historical magic word. The first judging unit is specifically configured to determine that the HBM parameter is retrained this time if any one of the current version number, the current temperature value, and the current magic word does not match the corresponding historical version number, historical temperature value, and historical magic word.

9. The apparatus of claim 8, wherein, The device further comprises: A training unit configured to train the HBM parameter according to the current version number, the current temperature value, and the current magic word if it is determined that the HBM parameter is retrained this time. An updating unit configured to replace the historical HBM parameter with the trained HBM parameter, and replace the historical version number, the historical temperature value, and the historical magic word with the current version number, the current temperature value, and the current magic word.

10. The apparatus of claim 9, wherein, The device further comprises: A second obtaining unit configured to obtain original record values of the single board when the switching chip is started for the first time when the switching chip is started for the first time, the original record values including original version numbers, original temperature values, and original magic words of the single board. The training unit is further configured to train the HBM parameter according to the original version number, the original temperature value, and the original magic word. A creating unit configured to create the HBM parameter storage path, and store the trained HBM parameter and the original version number, the original temperature value, and the original magic word in the storage unit.

11. The apparatus of claim 9, wherein, The training unit is further configured to train the HBM parameter according to the current version number, the current temperature value, and the current magic word if the HBM parameter storage path is not obtained, or if the historical HBM parameter is not loaded successfully. The updating unit is further configured to replace the historical HBM parameter with the trained HBM parameter, and replace the historical version number, the historical temperature value, and the historical magic word with the current version number, the current temperature value, and the current magic word.

12. The apparatus of claim 7, wherein, The device further comprises: A restarting unit configured to restart the switching chip again according to the alarm information if the alarm information is received, the alarm information including current record values that exceed or are lower than threshold values.

Images