Control method of many-core processor, server and storage medium

Abstract

The invention provides a control method of a many-core processor, a server and a storage medium, and belongs to the technical field of processors. The method comprises the following steps: acquiring a current temperature and a preset temperature threshold value of the many-core processor, and determining a power consumption budget value corresponding to each processing unit in the many-core processor according to the current temperature and the preset temperature threshold value; performing power consumption value matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption value matching processing result; and executing a corresponding temperature protection action on each processing unit according to a power consumption value matching processing result. According to the method and the device, the problems that the temperature monitoring granularity is large, and hysteresis exists in execution of the temperature protection action on the processor are solved, and when the many-core processor executes the temperature protection action, the situation that the performance of the many-core processor is seriously reduced is avoided.

Description

Technical Field

[0001] The present invention relates to the field of processor technology, and in particular to a control method, a server and a storage medium for a multi-core processor. Background Art

[0002] During the operation of the multi-core processor, heat will be generated. In order to protect the multi-core processor, a device for monitoring the temperature of the multi-core processor is usually designed outside the multi-core processor to monitor the temperature of the multi-core processor and trigger the corresponding protection action when the temperature of the multi-core processor is higher than the temperature protection threshold. However, this temperature control method uses the entire multi-core processor as the temperature monitoring and control object. The temperature sensing of the temperature monitoring device is slow. When the temperature adjustment is performed, the multi-core processor may have been working in a high temperature state for a period of time, which may easily lead to accelerated aging of the multi-core processor. The triggered protection action is to reduce the frequency of the entire multi-core processor, which may easily lead to a significant decrease in the working performance of the multi-core processor when it is hot. Summary of the invention

[0003] The present application provides a control method, server and storage medium for a many-core processor, which solves the problem of large granularity of temperature monitoring and lag in executing temperature protection actions on the processor, and avoids the situation where the performance of the many-core processor is seriously degraded when the many-core processor executes temperature protection actions.

[0004] In a first aspect, the present application provides a control method for a many-core processor, comprising:

[0005] Obtaining a current temperature and a preset temperature threshold of the many-core processor, and determining a power consumption budget value corresponding to each processing unit in the many-core processor according to the current temperature and the preset temperature threshold;

[0006] Perform power consumption value matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption value matching processing result;

[0007] The corresponding temperature protection action is performed on each processing unit according to the power consumption value matching processing result.

[0008] In a second aspect, the present application also provides a server, comprising a processor, a memory, a computer program stored in the memory and executable by the processor, and a data bus for realizing connection and communication between the processor and the memory, wherein when the computer program is executed by the processor, the steps of any control method of a multi-core processor provided in the specification of the present invention are implemented.

[0009] In a third aspect, the present application also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any control method of a multi-core processor provided in the specification of the present application.

[0010] The present application provides a control method, server and storage medium for a multi-core processor. The present application obtains the current temperature and preset temperature threshold of the multi-core processor, and determines the power consumption budget value corresponding to each processing unit in the multi-core processor according to the current temperature and the preset temperature threshold; performs power consumption matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption matching processing result; and performs corresponding temperature protection actions on each processing unit according to the power consumption matching processing result. By converting the detected temperature into power consumption, the instantaneous nature of power consumption change is utilized to achieve faster temperature control. At the same time, the temperature control method is embedded in the management core of the multi-core processor to run, effectively solving the problems of device aging and shortened life caused by temperature control delay, and by performing power consumption matching processing on each processing unit, each processing unit is controlled to perform corresponding temperature protection actions according to the power consumption matching processing result, solving the problem of large temperature control granularity, so that each processing unit in the entire processor does not need to be frequency-reduced, and the performance of the multi-core processor in the temperature control process is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A flowchart of a control method for a multi-core processor provided in one embodiment of the present application;

[0013] Figure 2 A schematic diagram of a scenario of a power consumption value matching process provided in an embodiment of the present application;

[0014] Figure 3 A schematic diagram of a scenario of a first power consumption set and a second power consumption set provided in an embodiment of the present application;

[0015] Figure 4 A schematic diagram of a calculation process scenario of a target power consumption set provided in an embodiment of the present application;

[0016] Figure 5A flowchart of a control method for a multi-core processor provided in an embodiment of the present application;

[0017] Figure 6 A schematic block diagram of the structure of a server provided in this application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in this application to clearly and completely describe the technical solutions in this application. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0020] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0021] The present application provides a control method, a server and a storage medium for a many-core processor. The control method for the many-core processor can be applied to the many-core processor to achieve temperature control of the many-core processor. Specifically, the control method for the many-core processor can be embedded in a chip in the many-core processor, such as in the system management software of the many-core processor system management core.

[0022] In a specific implementation process, the control method of the multi-core processor can be run on the SCP core (System Control Processor) inside the CPU (Central Processing Unit), and the CPU runs on the server motherboard.

[0023] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0024] Please refer to Figure 1 , Figure 1 A flowchart of a control method for a multi-core processor provided in an embodiment of the present application is provided.

[0025] like Figure 1As shown, the control method of the multi-core processor includes steps S101 to S103.

[0026] Step S101: obtaining the current temperature and the preset temperature threshold of the many-core processor, and determining the power consumption budget value corresponding to each processing unit in the many-core processor according to the current temperature and the preset temperature threshold.

[0027] Exemplarily, by obtaining the current temperature and a preset temperature threshold of the multi-core processor, where the preset temperature threshold is the set maximum temperature, specifically, a temperature margin is calculated based on the current temperature and the preset temperature threshold, and the temperature margin is converted into a power consumption margin, thereby determining the power consumption budget value corresponding to the processing unit based on the power consumption margin.

[0028] In some embodiments, a power consumption budget value corresponding to each processing unit in a many-core processor is determined based on a current temperature and a preset temperature threshold, including: dividing the many-core processor into a plurality of power consumption matching areas according to preset area division information, each power consumption matching area including a plurality of processing units; determining a temperature margin corresponding to each power consumption matching area in the many-core processor according to a preset temperature threshold and the current temperature; determining a power consumption margin corresponding to each temperature margin based on a preset discrete thermal model; and determining a power consumption budget value corresponding to each processing unit at the next sampling moment according to each power consumption margin and the current power consumption value corresponding to each processing unit.

[0029] Exemplarily, the multi-core processor is divided into a plurality of power consumption matching regions according to preset region division information, wherein each power consumption matching region includes a plurality of processing units.

[0030] It should be noted that the present application does not limit the size of the power consumption matching area and the number of processing units in each power consumption matching area.

[0031] Exemplarily, the current temperature of the power consumption matching area is obtained to determine the temperature margin of the power consumption matching area according to the current temperature and a preset temperature threshold. The calculation process is as follows:

[0032] ΔT=T set -T now

[0033] Among them, ΔT is used to indicate the temperature margin, T set Used to indicate the preset temperature threshold, T now Used to indicate the current temperature.

[0034] After the temperature margin is calculated, based on a preset discrete thermal model and according to the temperature margin of the power consumption matching area, the power consumption margin of the power consumption matching area is determined to complete the process of converting the temperature margin into the power consumption margin.

[0035] Specifically, the discrete heat model is

[0036] in,

[0037] Y p =[Y max1 T , Y max2 T , Y max3 T , ....Y maxp T ] T , R=rI q×p , r is the weight adjustment coefficient. is the state matrix of the discrete thermal model.

[0038] In the calculation of ΔP k After that, the power consumption budget value corresponding to the next sampling moment is calculated by the following formula.

[0039]

[0040] in, is the power consumption budget value corresponding to the next sampling moment, P is the current power consumption value, ΔP k is the power consumption margin.

[0041] The power consumption budget value corresponding to each processing unit at the next sampling moment is calculated through the above calculation process, so that power consumption value matching processing can be performed according to the current power consumption value of each processing unit and the power consumption budget value of each processing unit to obtain the power consumption value matching processing result.

[0042] Step S102: Perform power consumption value matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption value matching processing result.

[0043] Exemplarily, power consumption matching processing is performed on the current power consumption value of each processing unit and the power consumption budget value corresponding to each processing unit, that is, the current power consumption value and the power consumption budget value are matched to obtain a power consumption matching processing result, so that the corresponding processing unit can be controlled according to the power consumption value matching processing result.

[0044] In some embodiments, power consumption value matching processing is performed on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption value matching result, including: power consumption value matching processing is performed on the current power consumption value and the power consumption budget value corresponding to the processing units in each power consumption matching area to obtain a power consumption value matching processing result.

[0045] In the specific implementation process, the processing units in each power consumption matching area are determined, and the current power consumption value and the power consumption budget value corresponding to the processing units in each power consumption matching area are obtained, so as to use the current power consumption value and the power consumption budget value corresponding to the processing units in the same power consumption matching area for matching processing, thereby achieving matching of the current power consumption value and the power consumption budget value in each power consumption matching area.

[0046] In some embodiments, power consumption matching processing is performed on current power consumption values ​​and power consumption budget values ​​corresponding to processing units in each power consumption matching area to obtain power consumption value matching processing results, including: matching each current power consumption value with the power consumption budget value based on a preset Hungarian algorithm; associating the processing unit corresponding to the successfully matched current power consumption value with the processing unit corresponding to the power consumption budget value; and performing power consumption re-matching processing on unmatched current power consumption values ​​and unmatched power consumption budget values ​​to obtain power consumption value matching processing results.

[0047] Exemplarily, by matching the current power consumption value with the power consumption budget value, when a current power consumption value is matched successfully, the processing unit corresponding to the current power consumption value is associated with the processing unit of the power consumption budget value that matches the current power consumption value, and stored in the power consumption value matching processing result; and, the unmatched current power consumption value is subjected to power consumption re-matching processing with the unmatched power consumption budget value to obtain the power consumption value matching processing result.

[0048] For example, P1 is used to indicate the current power consumption value of processing unit a. is the power consumption budget value of processing unit a, P2 is the current power consumption value of processing unit b, is the power consumption budget of processing unit b, if P1 and If the match is successful, the processing unit a is associated with the processing unit b and stored in the power consumption value matching result, and the unmatched Rematch the power consumption with the unmatched P2.

[0049] In some embodiments, each current power consumption value is matched with each power consumption budget value based on a preset Hungarian algorithm, including: establishing a weight matrix, the number of rows in the weight matrix corresponds to the number of current power consumption values, the number of columns in the weight matrix corresponds to the number of power consumption budget values, and the elements in the weight matrix are used to indicate the absolute value of the difference between the power consumption budget value corresponding to the row where the element is located and the power consumption budget value corresponding to the column where the element is located; the absolute value of the difference greater than a first preset threshold is changed to a target parameter, the target parameter is used to indicate an infinite value, and the absolute value of the difference less than or equal to the first preset threshold is retained to obtain an updated weight matrix; in the updated weight matrix, each element in each row is subtracted from the minimum value of the row where the element is located to obtain a target weight matrix; the minimum value in each row of the target weight matrix is ​​determined, and when the minimum value is less than the second preset threshold, it is determined that the power consumption budget value corresponding to the column where the minimum value is located matches the current power consumption value corresponding to the row where the minimum value is located successfully, and the first preset threshold is greater than the second preset threshold.

[0050] Exemplarily, taking a certain power consumption matching area including processing unit a, processing unit b, processing unit c and processing unit d as an example, the power consumption value matching process is described.

[0051] See also Figure 2 , Figure 2 A schematic diagram of a scenario of a power consumption value matching process provided in an embodiment of the present application.

[0052] like Figure 2 As shown, in the power consumption matching area, the current power consumption value of processing unit a is P1, and the power consumption budget value is The current power consumption value of processing unit b is P2, and the power consumption budget value is The current power consumption value of processing unit c is P3, and the power consumption budget value is The current power consumption value of processing unit d is P4, and the power consumption budget value is A weight matrix is ​​established through the current power consumption value and the power consumption budget value of each processing unit, wherein the number of rows of the weight matrix is ​​the number of current power consumption values ​​in the power consumption matching area, and the number of columns of the weight matrix is ​​the number of power consumption budget values ​​in the power consumption matching area, that is, the weight matrix corresponding to the power consumption matching expectation is a 4X4 matrix, and the element is the absolute value of the difference between the current power consumption value corresponding to the row where the element is located and the power consumption budget value corresponding to the column where the element is located, and in the calculation process, P1 to P4 correspond to the first to fourth rows, to Corresponding to the first to fourth columns. P1 = 19, P2 = 11, P3 = 14, P4 = 17, For example, the weight matrix established is as follows:

[0053]

[0054] The established weight matrix is ​​updated to obtain a target weight matrix. Specifically, elements whose element values ​​are greater than a first preset threshold are updated to target parameters, and the value of the target parameter is infinite; elements whose element values ​​are less than or equal to the first preset threshold are retained to obtain an updated weight matrix.

[0055] Taking the first preset threshold value 2 as an example, the updated weight matrix obtained by updating the weight matrix established above is as follows:

[0056]

[0057] Among them, INF is used to indicate the target parameter.

[0058] In the updated weight matrix, the elements in each row are subtracted from the minimum element value in the row where the element is located, and the target weight matrix obtained is as follows:

[0059]

[0060] Determine the minimum value of each row in the target weight matrix, and when the minimum value is less than a second preset threshold, determine that the power consumption budget value indicated by the column where the element corresponding to the minimum value is located successfully matches the current power consumption value indicated by the row where the element corresponding to the minimum value is located.

[0061] Specifically, the second preset threshold is a value that has only one value in each row and column, or a preset value. In this embodiment, the second preset threshold is 1, there is no minimum value in the first row, and it is determined that P1 is not matched; the element value corresponding to the element in the first column of the second row is the minimum value in the second row and is less than 1, so it is determined that P2 is matched. The match is successful; the element value corresponding to the element in the third row and second column is the minimum value in the third row and is less than 1, confirming that P3 matches The match is successful; the element value corresponding to the element in the fourth row and third column is the minimum value in the fourth row and is less than 1, confirming that P4 matches The match is successful. The connected blocks in the figure represent the current power consumption value and the power consumption budget value that have been successfully matched, while the unconnected blocks indicate the current power consumption value or the power consumption budget value that have not been matched.

[0062] After the above processing, the power consumption matching pair is determined: and The unmatched P1 and It is understandable that in the subsequent steps, the corresponding temperature protection action can be performed on the processing unit corresponding to the current power consumption value based on the power consumption matching.

[0063] It should be noted that the same power consumption matching area may include more processing units, and the present application does not limit the number of processing units in the same power consumption matching area.

[0064] In some embodiments, power consumption rematching processing is performed on unmatched current power consumption values ​​and unmatched power consumption budget values ​​to obtain power consumption value matching processing results, including: based on a preset minimum iterative cut algorithm, group classification processing is performed on the unmatched current power consumption values ​​and unmatched power consumption budget values ​​obtained from each power consumption matching area to obtain a group classification processing result; based on the group classification processing result, the unmatched current power consumption value and the unmatched power consumption budget value are rematched to obtain a power consumption value matching processing result.

[0065] Exemplarily, unmatched current power consumption values ​​and unmatched power consumption budget values ​​are obtained from each power consumption matching area to perform power consumption rematching processing on all unmatched current power consumption values ​​and unmatched power consumption budget values ​​in the multi-core processor. Specifically, the unmatched current power consumption values ​​and unmatched power consumption budget values ​​are respectively divided into a first power consumption set and a second power consumption set, and the first power consumption set and the second power consumption set are processed based on the minimum iterative cut algorithm to obtain a final set classification processing result, wherein the set classification processing result is represented by a first target power consumption set and a second target power consumption set, so as to match the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​based on the first target power consumption set and the second target power consumption set to obtain a power consumption value matching processing result.

[0066] In some embodiments, based on a preset minimum iterative cut algorithm, the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​obtained in each power consumption matching area are subjected to set classification processing to obtain a set classification processing result, including: determining a first power consumption set and a second power consumption set, the first power consumption set including at least one unmatched current power consumption value and at least one unmatched power consumption budget value, and the second power consumption set including unmatched current power consumption values ​​and unmatched power consumption budget values ​​other than the first power consumption set; performing power consumption value movement processing on the first power consumption set and the second power consumption set; based on a preset set cut calculation rule, determining a cut value change value according to the first power consumption set and the second power consumption set after performing the power consumption value movement processing; determining the maximum cut value change value among the cut value change values ​​corresponding to each power consumption value movement processing, and determining the first target power consumption set and the second target power consumption set according to the comparison result of the maximum cut value change value and the third preset threshold.

[0067] In a specific implementation process, the case where the current power consumption values ​​and the power consumption budget values ​​corresponding to the processing units e, f, g, h, and i are matched is taken as an example for explanation.

[0068] See also Figure 3 , Figure 3 A schematic diagram of a scenario of a first power consumption set and a second power consumption set provided in an embodiment of the present application.

[0069] For example, the current power consumption value Pe and the power consumption budget value of the processing unit e are Current power consumption value Pf and power consumption budget value of processing unit f Current power consumption value Pg and power consumption budget value of processing unit g into the first power consumption set; the current power consumption value Ph of the processing unit h and the power consumption budget value Current power consumption value Pi and power consumption budget value of processing unit i to the second power consumption set.

[0070] After completing the allocation of the current power consumption value and the power consumption budget value, power consumption value moving processing is performed on the first power consumption set and the second power consumption set. Specifically, a current power consumption value in the first power consumption set, such as Pe, is moved to the second power consumption set. It should be understood that in the process of executing the power consumption value moving processing, other current power consumption values ​​such as Pg can also be moved, or a power consumption budget value can be moved. This application is not limited to this. In order to explain the calculation process in detail, the following will be explained by taking the movement of Pe as an example.

[0071] See also Figure 3 After moving Pe, the first power consumption set after the execution power consumption value movement processing includes Pf, Pg and The second power consumption set after the power consumption value movement process is performed includes Pe, Ph, Pi and And calculate the cut-off value change value corresponding to the power consumption value movement process, wherein the calculation process of the cut-off value change value is shown in the following formula:

[0072]

[0073] It should be understood that when the current power consumption value and / or the power consumption budget value in the first power consumption set are moved to the second power consumption set, the power consumption value movement processing of the first power consumption set and the second power consumption set is terminated, and the maximum cut value change value is determined among the cut value change values ​​corresponding to each power consumption movement processing, and the maximum cut value change value is compared with the third preset threshold to determine the first target power consumption set and the second target power consumption set.

[0074] The minimum iterative cut algorithm can reduce the algorithm complexity and calculation amount while meeting the matching accuracy requirements, ensure the response speed and real-time performance of the algorithm, thereby ensuring a rapid response to temperature changes in the processing unit and reducing device aging and shortened device life caused by temperature control delays.

[0075] In one embodiment, a maximum cut value change value is determined among the cut value change values ​​corresponding to each power consumption value movement process, and a first target power consumption set and a second target power consumption set are determined based on a comparison result between the maximum cut value change value and a third preset threshold value, including: if the maximum cut value change value is less than or equal to the third preset threshold value, then the first power consumption set corresponding to the maximum cut value change value is determined as the first target power consumption set, and the second power consumption set corresponding to the maximum cut value change value is determined as the second target power consumption set.

[0076] Exemplarily, the third preset threshold is 0. When the maximum cutoff value change value is less than or equal to 0, the first power consumption set after the power consumption value corresponding to the maximum cutoff value change value is moved is determined to be the first target power consumption set and the second power consumption set after the power consumption value corresponding to the maximum cutoff value change value is moved is determined to be the second target power consumption set, and the power consumption value and the power consumption budget value are matched in the first target power consumption set, and the power consumption value and the power consumption budget value are matched in the second target power consumption set.

[0077] In other embodiments, if the maximum cut value change value is greater than the third preset threshold, the power consumption value movement processing is performed on the first power consumption set and the second power consumption set again until the maximum cut value change value among the cut value change values ​​corresponding to the power consumption value movement processing is less than or equal to the third preset threshold, thereby ending the power consumption value movement processing for the first power consumption set and the second power consumption set.

[0078] It can be understood that when the maximum cut value change value is greater than 0, the power consumption value moving process is performed on the first power consumption set and the second power consumption set again. For example, if the corresponding maximum cut value change value after moving Pe is greater than 0, Pf in the first power consumption set is moved to the second power consumption set, and the corresponding cut value change value is calculated until the maximum cut value change value is found to be less than or equal to 0, the corresponding first target power consumption set and the second target power consumption set are determined, and the power consumption value moving process is ended.

[0079] In one embodiment, based on the set classification processing result, the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​are re-matched to obtain the power consumption value matching processing result, including: based on the preset Hungarian algorithm, matching the power consumption budget values ​​in the first target power consumption set with the current power consumption values ​​in the first target power consumption set, and matching the power consumption budget values ​​in the second target power consumption set with the current power consumption values ​​in the second target power consumption set; associating the processing unit corresponding to the successfully matched current power consumption value with the processing unit corresponding to the power consumption budget value, and storing the processing unit corresponding to the failed current power consumption value and the failed power consumption budget value into the power consumption value matching processing result.

[0080] Exemplarily, after determining the first target power consumption set and the second target power consumption set, the current power consumption value in the first target set is matched with the power consumption budget value. Specifically, the matching process is the same as the matching process of matching the current power consumption value corresponding to the processing unit in the power consumption matching area with the power consumption budget value, and will not be repeated here.

[0081] After the Hungarian algorithm matches the current power consumption value in the first target power consumption set with the power consumption budget value, and matches the current power consumption value in the second target power consumption set with the power consumption budget value, the processing unit of the successfully matched current power consumption value and the processing unit corresponding to the power consumption budget value that matches the current power consumption value are associated, and the processing unit corresponding to the current power consumption value that fails to match and the power consumption budget value that fails to match are stored in the power consumption value processing result, so as to perform corresponding temperature protection actions on each processing unit according to the power consumption value processing result.

[0082] Step S103: executing corresponding temperature protection actions on each processing unit according to the power consumption value matching processing result.

[0083] Exemplarily, the power consumption value matching processing result determined by the above steps performs corresponding temperature protection actions on each processing unit. Specifically, the power consumption value matching processing result includes the processing unit corresponding to the current power consumption value, the processing unit associated with the processing unit corresponding to the current power consumption value, and / or the current power consumption value that failed to match and its corresponding processing unit, as well as the power consumption budget value that failed to match and its corresponding processing unit.

[0084] Exemplarily, the temperature protection action to be performed on the processing unit is determined according to the matching result of the current power consumption value of the processing unit.

[0085] In some embodiments, a corresponding temperature protection action is performed on each processing unit according to the power consumption value matching processing result, including: migrating the currently executed task of the processing unit corresponding to the current power consumption value to a processing unit associated with the processing unit corresponding to the current power consumption value.

[0086] Exemplarily, after completing power consumption matching value processing for processing units in a power consumption matching area, and in the power consumption value matching processing result, if a processing unit has an associated processing unit, the currently executing task of the processing unit is migrated to the processing unit associated with the processing unit to reduce the power consumption of the processing unit, thereby achieving the purpose of cooling.

[0087] For example, the current power consumption value P1 of the processing unit a in a certain power consumption matching area is equal to the power consumption budget value P2 of the processing unit b in the same power consumption matching area. If the temperature protection action is executed, the current execution task of the processing unit a is migrated to the processing unit b to reduce the tasks executed on the processing unit a, thereby reducing the power consumption of the processing unit a and achieving the purpose of reducing the temperature of the processing unit a.

[0088] It can be understood that if the current power consumption value of the processing unit in the power consumption matching area does not match the power consumption budget value corresponding to the processing unit in the same power consumption matching area, the power consumption rematching processing step is entered. For the specific power consumption rematching processing steps, please refer to step S102, which will not be repeated here.

[0089] The power consumption value matching processing result obtained after the power consumption rematching processing is completed includes the successfully matched current power consumption value and its corresponding processing unit, the processing unit associated with the processing unit of the successfully matched current power consumption value, and / or the failed matching current power consumption value and its corresponding processing unit, and the unsuccessfully matched power consumption budget value. Based on the power consumption value matching processing result after the power consumption rematching processing, the corresponding temperature protection action is performed on each processing unit in the multi-core processor.

[0090] Exemplarily, after the power consumption rematching process, the processing unit corresponding to the successfully matched current power consumption value and the processing unit corresponding to the successfully matched power consumption budget value are associated, and the currently executed task of the processing unit corresponding to the successfully matched current power consumption value is migrated to the associated processing unit.

[0091] In other embodiments, corresponding temperature protection actions are performed on each processing unit according to the power consumption value matching processing results, including: based on the size relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match, controlling the processing unit corresponding to the power consumption budget value of the failed match to perform the corresponding temperature protection action.

[0092] See also Figure 4 , Figure 4 A schematic diagram of a calculation process scenario of a target power consumption set provided in an embodiment of the present application.

[0093] For example, the first power consumption set is processed multiple times to obtain a first target power consumption set, and the second power consumption set is processed multiple times to obtain a second target power consumption set, wherein each power consumption value movement process only moves one current power consumption value or one power consumption budget value in the first power consumption set, such as Figure 4 As shown, the cut value change value corresponding to each power consumption value movement process is calculated to determine the maximum cut value change value among multiple cut value change values, and when the maximum cut value change value is less than or equal to 0, the corresponding first target power consumption set and the second target power consumption set are determined, for example, the first target power consumption set obtained includes Pe, Pf, and Pi; the second target power consumption set includes Ph, Pg, and

[0094] Power consumption matching processing is performed in the first target power consumption set, and power consumption matching processing is performed in the second target power consumption set, and the power consumption value matching processing result is obtained as the current power consumption value Pe and the power consumption budget value Matching is successful, Pf and Matching is successful, Pg and Matching success; Pi in the first target power consumption set and Ph in the second target power consumption set, All the matches failed (the connected blocks in the figure indicate the current power consumption values ​​and the power consumption budget values ​​that have successfully matched, and the unconnected blocks indicate the current power consumption values ​​and the power consumption budget values ​​that have failed to match).

[0095] Migrate the currently executed task of the processing unit e corresponding to the successfully matched current power consumption value Pe to the power consumption budget value The corresponding processing unit g should be understood to execute the same processing as the processing unit e on the processing unit corresponding to the successfully matched current power consumption value, and complete the current execution task migration processing of the processing unit corresponding to the successfully matched current power consumption value.

[0096] The current power consumption value Pi that failed to match is compared with the power consumption budget value and the size relationship between the current power consumption value Ph that fails to match and the power consumption budget value to determine the temperature protection action to be performed by the processing unit i corresponding to the power consumption budget value Pi and the processing unit h corresponding to the current power consumption value Ph according to the comparison result.

[0097] In some embodiments, based on the size relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match, the processing unit corresponding to the current power consumption value of the failed match is controlled to perform corresponding temperature protection actions, including: when the power consumption budget value of the failed match is less than the current power consumption value of the failed match, adjusting the current operating frequency and current operating voltage of the processing unit corresponding to the power consumption budget value of the failed match, and migrating the current execution task of the processing unit corresponding to the current power consumption value of the failed match to the processing unit corresponding to the corresponding power consumption budget value of the failed match; when the power consumption budget value of the failed match is greater than or equal to the current power consumption value of the failed match, maintaining the current working state of the processing unit corresponding to the power consumption budget value of the failed match, and migrating the current execution task of the processing unit corresponding to the current power consumption value of the failed match to the processing unit corresponding to the corresponding power consumption budget value of the failed match.

[0098] For example, there is a corresponding relationship between the current power consumption value and the power consumption budget value for comparison, such as Pi and Compare the size, then the processing unit corresponding to the current power consumption value Pi and the corresponding processing unit corresponding to the power consumption budget value is the power consumption budget value The corresponding processing unit e.

[0099] In the specific implementation process, Pi = 19, Then reduce The current operating frequency and current operating voltage of the corresponding processing unit e are used to implement frequency and voltage modulation processing on the processing unit, and the current execution task on the processing unit i corresponding to Pi is migrated to The corresponding processing unit e should be understood that after the frequency and voltage modulation processing is performed on the processing unit, the processing unit will not be damaged when receiving the tasks migrated by other processing units, and the temperature of the processing unit will also decrease due to the decrease in operating load, thereby achieving the purpose of temperature control of the processing unit; in another implementation process, Ph=5, Maintain Corresponding to the current working status of the processing unit h, since the current power consumption value and the power consumption budget value correspond to the same processing unit, there is no need to perform task migration action; if the current power consumption value and the power consumption budget value correspond to different processing units, it is still necessary to migrate the currently executed task of the processing unit corresponding to the current power consumption value to the processing unit corresponding to the power consumption budget value.

[0100] like Figure 4 As shown, the current power consumption value Pg and the power consumption budget value If the match is successful, the current execution task of the processing unit g corresponding to Pg is migrated to The corresponding processing unit i, and the current power consumption value Pi fails to match, and the current power consumption value Pi that fails to match the power consumption budget value that fails to match Compare the size; determine Afterwards, the frequency and voltage modulation processing is performed on the processing unit e, and the current execution task of the processing unit i is controlled to migrate to the processing unit e; that is, the processing unit i migrates the originally executed task a to the processing unit e after the frequency and voltage modulation processing. At the same time, the processing unit i receives the current execution task b of the processing unit g, so that the processing unit i will not be idle, thereby realizing the task migration of the global processing unit.

[0101] It should be noted that the specific control process of the remaining processing units is the same as the steps mentioned in the above embodiment, which will not be described here. After the processing unit corresponding to the power consumption budget value that failed to match performs frequency and voltage regulation processing or maintains the current working state processing, it receives the current execution task of the processing unit corresponding to the current power consumption value that failed to match, realizes the task migration of the global processing unit, and avoids the problem of too many processing tasks or idle state of the processing unit.

[0102] See also Figure 5 , Figure 5 This is a flow chart of a control method for a multi-core processor provided in one embodiment of the present application.

[0103] like Figure 5 As shown, the control method of the multi-core processor provided by the present application divides the multi-core processor into multiple power consumption matching areas, determines the temperature margin of each power consumption matching area according to the current temperature of the multi-core processor, the preset temperature threshold and the area division information; based on the preset discrete thermal model, determines the power consumption margin corresponding to each power consumption matching area according to the temperature margin of each power consumption matching area, converts the temperature margin into the power consumption margin, and thus determines the power consumption budget value corresponding to each processing unit in each power consumption matching area based on the power consumption margin. The temperature margin is converted into the power consumption margin using the discrete thermal model, and the instantaneous nature of the power consumption change is used to improve the timeliness of the temperature control, thereby reducing the problems of device aging and shortened service life caused by the delay of the temperature control. At the same time, the multi-core processor is divided into multiple power consumption matching areas, which can achieve the local optimal matching of the power consumption margin in the area, and can solve the problem that the large granularity of temperature control leads to a serious decline in the performance of the multi-core processor.

[0104] After determining the power consumption budget value of each processing unit, the current power consumption value and the power consumption budget value of the processing units in the same power consumption matching area are matched based on the Hungarian algorithm to determine the power consumption budget value that matches the current power consumption value. The successfully matched current power consumption value is associated with the processing unit corresponding to the power consumption budget value and stored in the power consumption value matching processing result, while the unmatched current power consumption value and the unmatched power consumption budget value enter the power consumption rematching processing step.

[0105] During the power consumption rematching process, the unmatched current power consumption values ​​and unmatched power consumption budget values ​​obtained from each power consumption matching area are first divided into a first power consumption set and a second power consumption set, and the first target power consumption set and the second target power consumption set are determined based on the minimum iterative cut algorithm; based on the Hungarian algorithm, the current power consumption values ​​and the power consumption budget values ​​in the first target power consumption set are matched, and based on the Hungarian algorithm, the current power consumption values ​​and the power consumption budget values ​​in the second target power consumption set are matched to obtain the final power consumption value matching processing result. It should be understood that the power consumption value matching processing result includes the current power consumption value that has been successfully matched and its corresponding processing unit, the power consumption budget value that matches the current power consumption value and its corresponding processing unit, and the above two processing units also have an associated relationship, the current power consumption value that has failed to match and its corresponding processing unit, and the power consumption budget value that has failed to match and its corresponding processing unit.

[0106] The current task migration is performed on the processing unit corresponding to the current power consumption value that successfully matches, and the current working frequency and current working voltage of the processing unit corresponding to the power consumption budget value that failed to match are adjusted or the current working state of the processing unit corresponding to the power consumption budget value that failed to match is maintained according to the size relationship between the current power consumption value that failed to match and the power consumption budget value that failed to match, and after frequency and voltage modulation processing or maintaining working state processing, the current execution task of the processing unit corresponding to the current power consumption value that failed to match is migrated to the processing unit corresponding to the corresponding power consumption budget value that failed to match. The processing unit corresponding to the current power consumption value that successfully matches adopts task scheduling processing between processing units to ensure the performance of the multi-core processor, while the processing unit corresponding to the current power consumption value that failed to match maintains the current working state of the processing unit or performs frequency and voltage modulation processing based on the size relationship between the current power consumption value that failed to match and the power consumption budget value that failed to match. In the end, only a few processing units are subjected to frequency and voltage modulation processing, thereby avoiding the problem of serious performance degradation of the multi-core processor during temperature control.

[0107] It should be noted that the implementation process of the above steps can refer to the specific implementation methods provided in the previous embodiments, and will not be repeated here.

[0108] See also Figure 6 , Figure 6 A schematic block diagram of the structure of a server provided in this application.

[0109] like Figure 6 As shown, the server 300 includes a processor 301 and a memory 302 , and the processor 301 and the memory 302 are connected via a bus 303 , such as an I2C (Inter-integrated Circuit) bus.

[0110] Specifically, the processor 301 is used to provide computing and control capabilities to support the operation of the entire server. The processor 301 can be a central processing unit (CPU), and the processor 301 can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0111] Specifically, the memory 302 may be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a mobile hard disk.

[0112] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the server to which the solution of the present application is applied. The specific server may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0113] The processor is used to run a computer program stored in the memory, and implement any one of the control methods of the multi-core processor provided in the present application when executing the computer program.

[0114] In one embodiment, the processor is used to run a computer program stored in the memory, and implements the following steps when executing the computer program:

[0115] Obtaining a current temperature and a preset temperature threshold of the many-core processor, and determining a power consumption budget value corresponding to each processing unit in the many-core processor according to the current temperature and the preset temperature threshold;

[0116] Perform power consumption value matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain a power consumption value matching processing result;

[0117] The corresponding temperature protection action is performed on each processing unit according to the power consumption value matching processing result.

[0118] In one embodiment, when the processor determines the power consumption budget value corresponding to each processing unit in the multi-core processor according to the current temperature and the preset temperature threshold, it is used to implement:

[0119] Dividing the multi-core processor into a plurality of power consumption matching regions according to preset region division information, each power consumption matching region including a plurality of processing units;

[0120] Determine the temperature margin corresponding to each power consumption matching area in the many-core processor according to a preset temperature threshold and the current temperature;

[0121] Based on a preset discrete thermal model, determine the power consumption margin corresponding to each temperature margin;

[0122] According to each power consumption margin and the current power consumption value corresponding to each processing unit, a power consumption budget value corresponding to each processing unit at the next sampling moment is determined.

[0123] In one embodiment, when the processor performs power consumption value matching processing on the current power consumption value corresponding to each processing unit and the power consumption budget value corresponding to each processing unit to obtain the power consumption value matching processing result, it is used to implement:

[0124] Power consumption value matching processing is performed on the current power consumption value and the power consumption budget value corresponding to the processing unit in each power consumption matching area to obtain a power consumption value matching processing result.

[0125] In one embodiment, when the processor performs power consumption value matching processing on the current power consumption value and the power consumption budget value corresponding to the processing unit in each power consumption matching area and obtains the power consumption value matching processing result, it is used to implement:

[0126] Match each current power consumption value with the power consumption budget value based on the preset Hungarian algorithm;

[0127] Associating the processing unit corresponding to the successfully matched current power consumption value with the processing unit corresponding to the power consumption budget value;

[0128] The unmatched current power consumption value and the unmatched power consumption budget value are subjected to power consumption rematching processing to obtain a power consumption value matching processing result.

[0129] In one embodiment, when the processor matches each current power consumption value with the power consumption budget value based on a preset Hungarian algorithm, it is configured to implement:

[0130] Establish a weight matrix, the number of rows in the weight matrix corresponds to the number of current power consumption values, the number of columns in the weight matrix corresponds to the number of power consumption budget values, and the elements in the weight matrix are used to indicate the absolute value of the difference between the current power consumption value corresponding to the row where the element is located and the power consumption budget value corresponding to the column where the element is located;

[0131] The absolute value of the difference greater than the first preset threshold is changed to a target parameter, where the target parameter is used to indicate an infinite value, and the absolute value of the difference less than or equal to the first preset threshold is retained to obtain an updated weight matrix;

[0132] In the updated weight matrix, each element in each row is subtracted from the minimum value of the row where the element is located to obtain the target weight matrix;

[0133] Determine the minimum value in each row of the target weight matrix, and when the minimum value is less than the second preset threshold, determine that the power consumption budget value corresponding to the column where the minimum value is located successfully matches the current power consumption value corresponding to the row where the minimum value is located, and the first preset threshold is greater than the second preset threshold.

[0134] In one embodiment, when the processor performs a corresponding temperature protection action on each processing unit according to the power consumption value matching processing result, it is used to implement:

[0135] The currently executed task of the processing unit corresponding to the current power consumption value is migrated to the processing unit associated with the processing unit corresponding to the current power consumption value.

[0136] In one embodiment, when the processor performs power consumption rematching processing on the unmatched current power consumption value and the unmatched power consumption budget value to obtain the power consumption value matching processing result, it is used to implement:

[0137] Based on a preset minimum iterative cut algorithm, the unmatched current power consumption values ​​and unmatched power consumption budget values ​​obtained from each power consumption matching area are subjected to collective classification processing to obtain a collective classification processing result;

[0138] Based on the set classification processing result, the unmatched current power consumption value and the unmatched power consumption budget value are re-matched to obtain a power consumption value matching processing result.

[0139] In one embodiment, when the processor implements a preset minimum iterative cut algorithm, performs a collection classification process on the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​obtained in each power consumption matching area, and obtains a collection classification process result, it is used to implement:

[0140] Determine a first power consumption set and a second power consumption set, the first power consumption set including at least one unmatched current power consumption value and at least one unmatched power consumption budget value, and the second power consumption set including unmatched current power consumption values ​​and unmatched power consumption budget values ​​other than the first power consumption set;

[0141] Performing power consumption value moving processing on the first power consumption set and the second power consumption set;

[0142] Based on a preset set cut value calculation rule, determining a cut value change value according to the first power consumption set and the second power consumption set after the power consumption value movement process is performed;

[0143] The maximum cut value change value is determined among the cut value change values ​​corresponding to each power consumption value movement process, and the first target power consumption set and the second target power consumption set are determined according to the comparison result between the maximum cut value change value and the third preset threshold.

[0144] In one embodiment, when the processor determines the maximum cut value change value among the cut value change values ​​corresponding to each power consumption value movement process, and determines the first target power consumption set and the second target power consumption set according to the comparison result between the maximum cut value change value and the third preset threshold value, it is used to implement:

[0145] If the maximum cut value change value is less than or equal to the third preset threshold, determining the first power consumption set corresponding to the maximum cut value change value as the first target power consumption set, and the second power consumption set corresponding to the maximum cut value change value as the second target power consumption set;

[0146] If the maximum cut-off value change value is greater than the third preset threshold, the power consumption value movement processing is performed on the first power consumption set and the second power consumption set again until the maximum cut-off value change value among the cut-off value change values ​​corresponding to the power consumption value movement processing is less than or equal to the third preset threshold, and the power consumption value movement processing of the first power consumption set and the second power consumption set is ended.

[0147] In one embodiment, when the processor performs rematching processing on the unmatched current power consumption value and the unmatched power consumption budget value based on the set classification processing result to obtain the power consumption value matching processing result, it is used to implement:

[0148] Based on a preset Hungarian algorithm, matching the power consumption budget value in the first target power consumption set with the current power consumption value in the first target power consumption set, and matching the power consumption budget value in the second target power consumption set with the current power consumption value in the second target power consumption set;

[0149] The processing unit corresponding to the successfully matched current power consumption value is associated with the processing unit corresponding to the power consumption budget value, and the processing unit corresponding to the failed matching current power consumption value and the failed matching power consumption budget value are stored in the power consumption value matching processing result.

[0150] In one embodiment, when the processor performs a corresponding temperature protection action on each processing unit according to the power consumption value matching processing result, it is used to implement:

[0151] Migrating the currently executed task of the processing unit corresponding to the current power consumption value to the processing unit associated with the processing unit of the current power consumption value;

[0152] According to the magnitude relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match, the processing unit corresponding to the power consumption budget value of the failed match is controlled to perform a corresponding temperature protection action.

[0153] In one embodiment, when the processor controls the processing unit corresponding to the power consumption budget value of the failed match to perform the corresponding temperature protection action according to the size relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match, it is used to implement:

[0154] When the power consumption budget value of the failed match is less than the current power consumption value of the failed match, adjust the current operating frequency and current operating voltage of the processing unit corresponding to the power consumption budget value of the failed match, and migrate the current execution task of the processing unit corresponding to the current power consumption value of the failed match to the processing unit corresponding to the corresponding power consumption budget value of the failed match;

[0155] When the power consumption budget value of the failed match is greater than or equal to the current power consumption value of the failed match, the current working state of the processing unit corresponding to the power consumption budget value of the failed match is maintained, and the current execution task of the processing unit corresponding to the current power consumption value of the failed match is migrated to the processing unit corresponding to the corresponding power consumption budget value of the failed match.

[0156] It should be noted that technicians in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working process of the server described above can refer to the corresponding process in the aforementioned multi-core processor control method embodiment, and will not be repeated here.

[0157] The present application also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any control method of a multi-core processor provided in the specification of the present application.

[0158] The storage medium may be an internal storage unit of the server described in the foregoing embodiment, such as a hard disk or memory of the server. The storage medium may also be an external storage device of the server, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the server.

[0159] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0160] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0161] The above application serial numbers are only for description and do not represent the advantages and disadvantages of the embodiments. The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A control method for a multi-core processor, It is characterized in that include: Acquire a current temperature and a preset temperature threshold of the many-core processor, and determine a power consumption budget value corresponding to each processing unit in the many-core processor according to the current temperature and the preset temperature threshold; Perform power consumption value matching processing on the current power consumption value corresponding to each of the processing units and the power consumption budget value corresponding to each of the processing units to obtain a power consumption value matching processing result; A corresponding temperature protection action is performed on each of the processing units according to the power consumption value matching processing result.

2. The control method of the multi-core processor according to claim 1, It is characterized in that The determining, according to the current temperature and the preset temperature threshold, a power consumption budget value corresponding to each processing unit in the many-core processor includes: Dividing the multi-core processor into a plurality of power consumption matching regions according to preset region division information, each of the power consumption matching regions including a plurality of processing units; Determine a temperature margin corresponding to each of the power consumption matching regions in the many-core processor according to the preset temperature threshold and the current temperature; Based on a preset discrete thermal model, determining a power consumption margin corresponding to each of the temperature margins; The power consumption budget value corresponding to each processing unit at the next sampling moment is determined according to each power consumption margin and the current power consumption value corresponding to each processing unit.

3. The control method of the multi-core processor according to claim 1 or 2, It is characterized in that The performing power consumption value matching processing on the current power consumption value corresponding to each of the processing units and the power consumption budget value corresponding to each of the processing units to obtain the power consumption value matching processing result includes: Power consumption value matching processing is performed on the current power consumption value and the power consumption budget value corresponding to the processing unit in each power consumption matching area to obtain a power consumption value matching processing result.

4. The control method of the many-core processor according to claim 3, It is characterized in that The performing power consumption value matching processing on the current power consumption value and the power consumption budget value corresponding to the processing unit in each power consumption matching area to obtain the power consumption value matching processing result includes: Matching each of the current power consumption values ​​with the power consumption budget value based on a preset Hungarian algorithm; Associating the processing unit corresponding to the successfully matched current power consumption value with the processing unit corresponding to the power consumption budget value; The unmatched current power consumption value and the unmatched power consumption budget value are subjected to power consumption rematching processing to obtain a power consumption value matching processing result.

5. The control method of the many-core processor according to claim 4, It is characterized in that The matching of each of the current power consumption values ​​with the power consumption budget value based on a preset Hungarian algorithm includes: Establish a weight matrix, wherein the number of rows in the weight matrix corresponds to the number of the current power consumption values, the number of columns in the weight matrix corresponds to the number of the power consumption budget values, and the elements in the weight matrix are used to indicate the absolute value of the difference between the current power consumption value corresponding to the row where the element is located and the power consumption budget value corresponding to the column where the element is located; The absolute value of the difference greater than the first preset threshold is changed to a target parameter, where the target parameter is used to indicate an infinite value, and the absolute value of the difference less than or equal to the first preset threshold is retained to obtain an updated weight matrix; In the updated weight matrix, each element in each row is subtracted from the minimum value of the row where the element is located to obtain a target weight matrix; Determine the minimum value in each row of the target weight matrix, and when the minimum value is less than a second preset threshold, determine that the power consumption budget value corresponding to the column where the minimum value is located successfully matches the current power consumption value corresponding to the row where the minimum value is located, and the first preset threshold is greater than the second preset threshold.

6. The control method of the many-core processor according to claim 4, It is characterized in that The performing power consumption rematching processing on the unmatched current power consumption value and the unmatched power consumption budget value to obtain the power consumption value matching processing result includes: Based on a preset minimum iterative cut algorithm, the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​obtained from each of the power consumption matching areas are subjected to collective classification processing to obtain a collective classification processing result; Based on the set classification processing result, the unmatched current power consumption value and the unmatched power consumption budget value are re-matched to obtain a power consumption value matching processing result.

7. The control method of the many-core processor according to claim 6, It is characterized in that Based on the preset minimum iterative cut algorithm, the unmatched current power consumption values ​​and the unmatched power consumption budget values ​​obtained in each power consumption matching area are subjected to collective classification processing to obtain a collective classification processing result, including: Determine a first power consumption set and a second power consumption set, wherein the first power consumption set includes at least one unmatched current power consumption value and at least one unmatched power consumption budget value, and the second power consumption set includes unmatched current power consumption values ​​and unmatched power consumption budget values ​​other than the first power consumption set; Performing power consumption value moving processing on the first power consumption set and the second power consumption set; Based on a preset set cut value calculation rule, determining a cut value change value according to the first power consumption set and the second power consumption set after the power consumption value movement process is performed; A maximum cut value change value is determined among the cut value change values ​​corresponding to each power consumption value movement process, and a first target power consumption set and a second target power consumption set are determined according to a comparison result between the maximum cut value change value and a third preset threshold.

8. The control method of the many-core processor according to claim 7, It is characterized in that The method of determining a maximum cut value change value among the cut value change values ​​corresponding to each power consumption value movement process, and determining a first target power consumption set and a second target power consumption set according to a comparison result between the maximum cut value change value and a third preset threshold, includes: If the maximum cut value change value is less than or equal to the third preset threshold, determining that the first power consumption set corresponding to the maximum cut value change value is the first target power consumption set, and the second power consumption set corresponding to the maximum cut value change value is the second target power consumption set; If the maximum cut value change value is greater than the third preset threshold, the power consumption value movement processing is performed on the first power consumption set and the second power consumption set again until the maximum cut value change value among the cut value change values ​​corresponding to the power consumption value movement processing is less than or equal to the third preset threshold, and the power consumption value movement processing of the first power consumption set and the second power consumption set is terminated.

9. The control method of the many-core processor according to claim 6, It is characterized in that The rematching process is performed on the unmatched current power consumption value and the unmatched power consumption budget value based on the set classification process result to obtain the power consumption value matching process result, including: Based on the preset Hungarian algorithm, matching the power consumption budget value in the first target power consumption set with the current power consumption value in the first target power consumption set, and matching the power consumption budget value in the second target power consumption set with the current power consumption value in the second target power consumption set; The processing unit corresponding to the successfully matched current power consumption value is associated with the processing unit corresponding to the power consumption budget value, and the processing unit corresponding to the failed matching current power consumption value and the failed matching power consumption budget value are stored in the power consumption value matching processing result.

10. The control method of the many-core processor according to claim 1, It is characterized in that The performing corresponding temperature protection actions on each of the processing units according to the power consumption value matching processing result includes: Migrating a currently executed task of a processing unit corresponding to the current power consumption value to a processing unit associated with the processing unit of the current power consumption value; According to the magnitude relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match, the processing unit corresponding to the power consumption budget value of the failed match is controlled to perform a corresponding temperature protection action.

11. The control method of the many-core processor according to claim 10, It is characterized in that The controlling the processing unit corresponding to the power consumption budget value of the failed match to perform a corresponding temperature protection action according to the magnitude relationship between the current power consumption value of the failed match and the power consumption budget value of the failed match includes: When the power consumption budget value of the failed match is less than the current power consumption value of the failed match, adjust the current operating frequency and current operating voltage of the processing unit corresponding to the power consumption budget value of the failed match, and migrate the current execution task of the processing unit corresponding to the current power consumption value of the failed match to the processing unit corresponding to the power consumption budget value of the failed match; When the power consumption budget value of the failed match is greater than or equal to the current power consumption value of the failed match, the current working state of the processing unit corresponding to the power consumption budget value of the failed match is maintained, and the current execution task of the processing unit corresponding to the current power consumption value of the failed match is migrated to the processing unit corresponding to the power consumption budget value of the failed match.

12. A server, It is characterized in that The server includes a central processing unit, a memory, a computer program stored in the memory and executable by the central processing unit, and a data bus for realizing connection and communication between the central processing unit and the memory, wherein when the computer program is executed by the central processing unit, the steps of the control method of the multi-core processor as described in any one of claims 1 to 11 are implemented.

13. A storage medium, It is characterized in that The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the control method of the many-core processor according to any one of claims 1 to 11.

Images