Heat dissipation regulation and control method and device, computer equipment, storage medium and program product

Abstract

The invention discloses a heat dissipation regulation and control method and device, computer equipment, a storage medium and a program product, and relates to the technical field of computers. According to the heat dissipation regulation and control method, firstly, state changes of all hardware partitions in a server can be detected to obtain state data, and the hardware partitions comprise host systems running independently; then, on the basis of the state data, a control weight of a heat dissipation device can be distributed to the hardware partition, and on the basis of the control weight, the heat dissipation device is controlled to dissipate heat of the hardware partition, so that the state of the hardware partition reaches a target state; the technical problem that a heat dissipation system is poor when global heat dissipation regulation and control are carried out through a management unit in the server is solved, and the technical effects that the control weight of the heat dissipation device is dynamically distributed, so that the heat dissipation power of the global heat dissipation device is dynamically optimized, and local overheating or power consumption waste of the server is avoided are achieved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a heat dissipation control method, device, computer equipment, storage medium and program product. Background Art

[0002] In recent years, with the rapid development of cloud computing, artificial intelligence and big data technologies, multi-channel and multi-core processors have become the core computing units of data centers. In order to further improve the utilization of hardware resources in multi-tenant scenarios in data centers, traditional virtualization technology is used to virtualize multiple virtual machines on a single physical server to achieve resource sharing. However, the additional overhead of the virtual machine hypervisor makes it difficult for real-time services to meet low latency requirements. For this reason, hardware partitioning technology has emerged. Through physical isolation, the computing resources of a single server (such as processors, memory, and input and output devices) are divided into multiple independently running "bare metal" host systems, thereby realizing the partitioning of physical machine performance.

[0003] In related heat dissipation control schemes, the management unit in the server often manages all heat dissipation devices globally and adjusts the rotation speed based on the overall temperature threshold of the server. For example, if the server includes multiple hosts, when the temperature rise in the server is detected, the speed of all fans can be increased through the management unit to dissipate heat for the server. However, since the hardware partitions in the server are often different, this leads to poor global heat dissipation coordination for the server. For example, if the temperature of a partition is high, if the fan speed is increased to reduce the temperature of the partition, it may cause heat accumulation in the adjacent partition with lower temperature. Summary of the invention

[0004] The present application provides a heat dissipation control method, apparatus, computer equipment, storage medium and program product to at least solve the problem of poor heat dissipation system when a server performs global heat dissipation control through a management unit in the related art.

[0005] The present application provides a heat dissipation control method, the method comprising: Detecting status changes of various hardware partitions in the server to obtain status data, wherein the hardware partitions include independently running host systems; Based on the status data, assigning a control weight of a heat sink to the hardware partition; The heat dissipation device is controlled based on the control weight to dissipate heat for the hardware partition, so that the state of the hardware partition reaches a target state.

[0006] The present application also provides a heat dissipation control device, which includes: A detection module, used to detect the state change of each hardware partition in the server and obtain state data, wherein the hardware partition includes an independently running host system; an allocation module, configured to allocate a control weight of a heat dissipation device to the hardware partition based on the status data; The heat dissipation module is used to control the heat dissipation device to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches a target state.

[0007] The present application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of any of the above-mentioned heat dissipation control methods when executing the computer program.

[0008] The present application also provides a computer-readable storage medium, in which a computer program is stored, wherein when the computer program is executed by a processor, the steps of any of the above-mentioned heat dissipation control methods are implemented.

[0009] The present application also provides a computer program product, including a computer program, which implements the steps of any of the above-mentioned heat dissipation control methods when executed by a processor.

[0010] Through the present application, the state changes of each hardware partition in the server can be detected first to obtain state data, wherein the hardware partition contains an independently running host system, and based on the state data, the control weight of the heat sink is allocated to the hardware partition. Then, the heat sink can be controlled to dissipate heat for the hardware partition based on the control weight so that the state of the hardware partition reaches the target state. Therefore, the technical problem of poor heat dissipation system when the server performs global heat dissipation regulation through the management unit can be solved, thereby dynamically allocating the control weight of the heat sink to dynamically optimize the heat dissipation power of the global heat sink to avoid local overheating or power consumption waste of the server. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only 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 heat dissipation control method according to an embodiment of the present invention provided in an embodiment of the present application; Figure 2 It is a schematic diagram of dynamic thermal management program scheduling; Figure 3 is a flow chart of another heat dissipation control method according to an embodiment of the present invention; Figure 4 A schematic diagram of the processor and heat sink arrangement for the hardware partition of a four-way server; Figure 5 is a flow chart of another heat dissipation control method according to an embodiment of the present invention; Figure 6 is a structural block diagram of a heat dissipation control device according to an embodiment of the present invention; Figure 7 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0014] It should be noted that, in the description of this application, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device 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 device. The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence.

[0015] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0016] In conjunction with the specific application environment architecture or the specific hardware architecture on which the execution of the heat dissipation control method depends, the specific application environment architecture or the specific hardware architecture is described herein.

[0017] In recent years, with the rapid development of cloud computing, artificial intelligence and big data technologies, multi-channel and multi-core processors have become the core computing units of data centers. In order to further improve the utilization of hardware resources in multi-tenant scenarios in data centers, traditional virtualization technology is used to virtualize multiple virtual machines on a single physical server to achieve resource sharing. However, the additional overhead of the virtual machine hypervisor makes it difficult for real-time services to meet low latency requirements. For this reason, hardware partitioning technology has emerged. Through physical isolation, the computing resources of a single server (such as processors, memory, and input and output devices) are divided into multiple independently running "bare metal" host systems, thereby realizing the partitioning of physical machine performance.

[0018] In related heat dissipation control solutions, the management unit in the server often manages all heat dissipation devices globally and adjusts the speed based on the overall temperature threshold of the server. For example, if the server includes multiple hosts, when the temperature rise in the server is detected, the speed of all fans can be increased through the management unit to dissipate heat for the server. However, since the hardware partitions in the server are often different, this leads to poor global heat dissipation coordination for the server.

[0019] For example, if a static allocation solution is adopted on the basis of the relevant heat dissipation control solution, and a certain part of the fans are fixedly bound to a specific partition, since the heat dissipation of each hardware partition is jointly determined by multiple fans, this will lead to unclear ownership of control rights, and still cause the problem of poor global heat dissipation coordination for the server. For example, if the temperature of a partition is high, if the fan speed corresponding to the partition is increased to reduce the temperature of the partition, it may interfere with the airflow distribution of the adjacent partition, causing heat accumulation in the adjacent partition with lower temperature.

[0020] The embodiment of the present application provides a heat dissipation control method, and the method is described in detail in combination with the execution process of the heat dissipation control method. The method can first detect the state changes of each hardware partition in the server to obtain state data, wherein the hardware partition contains an independently running host system, and based on the state data, the control weight of the heat dissipation device is allocated to the hardware partition. Then, the heat dissipation device can be controlled to dissipate heat for the hardware partition based on the control weight so that the state of the hardware partition reaches the target state, thereby dynamically allocating the control weight of the heat dissipation device to dynamically optimize the heat dissipation power of the global heat dissipation device to avoid local overheating or power consumption waste of the server.

[0021] According to an embodiment of the present invention, an embodiment of a heat dissipation control method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0022] In this embodiment, a heat dissipation control method is provided, which can be used for a server, such as a server in a data center. Figure 1 is a flow chart of a heat dissipation control method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps: Step S101, detecting the status change of each hardware partition in the server to obtain status data, wherein the hardware partition includes an independently running host system.

[0023] In the embodiments of the present disclosure, hardware partitioning technology can be used in the server, that is, dividing the server into multiple hardware partitions, where the hardware partitions can be multiple independently running "bare metal" host systems obtained by dividing the computing resources (such as CPU, memory, input and output devices, etc.) of a single server through physical isolation. Each hardware partition has exclusive hardware resources and can directly run the native operating system, such as Linux system, Windows system, etc.

[0024] During the operation of the server, it is necessary to monitor the status data of each hardware partition in real time. The status data can be used to indicate the load and temperature of the processor, memory, etc. in the hardware partition, so that the cooling device corresponding to the hardware partition can be adjusted in real time according to the status data to cool the hardware partition.

[0025] Step S102: allocating control weights of heat dissipation devices to hardware partitions based on the status data.

[0026] In the disclosed embodiment, the management unit (baseboard management controller, hereinafter referred to as BMC) in the server can be partitioned in advance, and each hardware partition is assigned a corresponding BMC partition, so that the BMC partition controls the heat dissipation device based on the control weight of the heat dissipation device. In addition, the heat dissipation device can be a fan.

[0027] It should be understood that for the same heat dissipation device, multiple hardware partitions may be corresponding, and the sum of the control weights corresponding to all hardware partitions is 1. For example, if heat dissipation device 1 corresponds to hardware partition 1 and hardware partition 2, the control weight corresponding to hardware partition 1 may be 0.6, and the control weight corresponding to hardware partition 2 may be 0.4.

[0028] Here, the control weight of the heat dissipation device corresponding to the hardware partition can be adjusted dynamically. For example, the higher the real-time temperature corresponding to the hardware partition determined according to the status data, the greater the heat dissipation demand of the hardware partition, and the higher the control weight allocated to the heat dissipation partition.

[0029] Step S103: controlling the heat dissipation device to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches the target state.

[0030] In an embodiment of the present disclosure, when the heat dissipation device includes a fan, the control weight can be used to control the rotation speed of the heat dissipation device so that the state of the hardware partition reaches a target state, wherein the target state can be used to indicate that the hardware temperature in the hardware partition has returned to normal temperature, or the hardware load has returned to normal load.

[0031] Specifically, when controlling the heat sink based on the control weight, the PWM (Pulse Width Modulation) duty cycle of the heat sink can be calculated based on the control weight, so as to adjust the fan frequency based on the PWM duty cycle, thereby controlling the speed. Here, if a heat sink serves two enabled hardware partitions at the same time, its PWM duty cycle control right will be split according to this ratio, and multiple logical channels will be established through the remapping of the BMC physical PWM pins, and each will be independently adjusted by the corresponding partitions.

[0032] For example, if the control weight of hardware partition 1 corresponding to the fan is 0.6, and the control weight of hardware partition 2 corresponding to the fan is 0.4, the PWM duty cycle output by the BMC partition of hardware partition 1 is PWM1, and the PWM duty cycle output by the BMC partition of hardware partition 2 is PWM2. Then, the total PWM duty cycle corresponding to the fan can be expressed as 0.6PWM1+0.4PWM2.

[0033] It can be seen from the above description that in the embodiment of the present disclosure, the state change of each hardware partition in the server can be detected first to obtain state data, wherein the hardware partition includes an independently running host system, and based on the state data, the control weight of the heat sink is allocated to the hardware partition. Then, the heat sink can be controlled to dissipate heat for the hardware partition based on the control weight so that the state of the hardware partition reaches the target state, thereby dynamically allocating the control weight of the heat sink to dynamically optimize the heat dissipation power of the global heat sink to avoid local overheating or power consumption waste of the server.

[0034] In some optional implementations, before allocating control weights of heat sinks to hardware partitions, the above Figure 1 The corresponding embodiments also include: Step S11, partitioning the management unit in the server based on hardware partitioning to obtain sub-management units, wherein the management unit is used to perform global heat dissipation management based on the running status of the server.

[0035] Step S12, assigning a corresponding sub-management unit to the hardware partition, wherein the sub-management unit is used to control a heat dissipation device corresponding to the hardware partition.

[0036] In the disclosed embodiment, the management unit is the above-mentioned BMC. Here, the software and hardware collaborative design of BMC and CPLD (complex programmable logic device, a programmable logic device) can be used to realize an intelligent heat dissipation management system that is collaboratively composed of a dynamic topology-aware fan allocation module and a cross-partition thermal collaborative closed-loop control module. Among them, the operating status of each hardware partition can be carried out through the BMC, and the CPLD can aggregate the temperature, power consumption and historical data of each hardware partition, run the heat conduction prediction model and emergency priority evaluation, generate a globally optimized fan speed regulation strategy, and distribute the control instructions to the corresponding heat dissipation device by controlling the PWM signal to control the fan.

[0037] Specifically, the BMC hardware in the server can be partitioned first to obtain multiple sub-management units (i.e., the above-mentioned BMC units), where each sub-management unit can be used to manage one or more corresponding hardware partitions and control the cooling device corresponding to the hardware partition. Here, the sub-management unit can assign the control weight of the cooling device to the hardware partition by running a dynamic cooling management program (the dynamic cooling management program is run through a dynamic topology-aware fan allocation module) to control the cooling device.

[0038] Next, in view of the load imbalance characteristics of the sub-management unit in the hardware partitioning scenario, an elastically scalable heat dissipation control module scheduling mechanism can be used to run the above dynamic heat dissipation management program, which specifically includes the following process: Step S21: determining a target sub-management unit whose load is lower than a preset threshold among the sub-management units.

[0039] Step S22: running a dynamic heat dissipation management program based on the target sub-management unit, so as to execute the step of allocating control weights of heat dissipation devices to hardware partitions based on the state data based on the dynamic heat dissipation management program.

[0040] In the disclosed embodiment, the target sub-management unit with a load lower than a preset threshold may be a sub-management unit with the lowest load. For example, a sub-management unit when the hardware partition is in a shutdown state is in a low-load state because no real-time monitoring task is running.

[0041] Here, the load level of each sub-management unit can be directly associated with the operating status of the corresponding hardware partition. For example, when a hardware partition is in a shutdown state, its associated sub-management unit is in a low-load state due to no real-time monitoring tasks. At this time, the system automatically dispatches the dynamic heat dissipation management program to the low-load sub-management unit to run, so as to fully utilize idle computing resources to generate a global control strategy.

[0042] For example, if all hardware partitions are in the power-on state, the system monitors the processor utilization of each hardware partition in real time through BMC inter-core communication, migrates the dynamic cooling management program to the target sub-management unit with the lowest current core load, and uses state snapshot and incremental synchronization technology to achieve switching, ensuring that the program always occupies the minimum resources.

[0043] Specifically, the above step S21, determining the target sub-management unit whose load is lower than the preset threshold in the sub-management unit, includes: When the running time of the dynamic heat dissipation management program meets the load migration condition, the target sub-management unit is determined based on the sub-management unit with the current lowest load.

[0044] In the embodiments of the present disclosure, it is considered that the load change rate of the server is closely related to the usage scenario and the operating time period. For example, for some usage scenarios, certain specific time periods every day are usually peak times for user access, such as shopping time in the evening, working time during the day, etc. Taking the social media platform as an example, seven or eight o'clock in the evening is a period of high user activity, and a large number of users will log in, post content, and interact at the same time. Therefore, a load migration condition can be established based on the usage scenario and the operating time period, and the load migration operation of the dynamic thermal management program can be performed when the load migration condition is met, so as to avoid too frequent migration of the dynamic thermal management program.

[0045] Specifically, a migration detection program may be pre-set, and when the running time of the dynamic thermal management program meets the load migration condition, the migration detection program is run to detect the target sub-management unit and migrate the dynamic thermal management program.

[0046] After the target sub-management unit is determined, the above step S22, running the dynamic heat dissipation management program based on the target sub-management unit, includes: Step a1, when there are multiple target sub-management units, obtain the inter-core communication efficiency of the target sub-management units, and determine whether the inter-core communication efficiency meets the parallel processing conditions, if so, execute step a2, if not, execute step a3.

[0047] Step a2: jointly running a dynamic heat dissipation management program based on the target sub-management unit.

[0048] Step a3: running a dynamic heat dissipation management program based on any target sub-management unit.

[0049] In the disclosed embodiment, the BMC may be multi-core, wherein the sub-management units transmit data through inter-core communication. Therefore, when the inter-core communication efficiency is high, it means that the communication bus is less occupied, and the dynamic thermal management program can be run in parallel through multiple target sub-management units. Conversely, the dynamic thermal management program can be run separately through any target sub-management unit.

[0050] Specifically, Figure 2 The figure shows a schematic diagram of the dynamic thermal management program scheduling, where all hardware partitions in (a) are turned off, and the thermal control module does not need to be scheduled at this time. In (b), partitions 1 and 2 are in the power-on state, so the dynamic thermal management program is scheduled on the target sub-management units of partitions 3 and 4. Specifically, the dynamic thermal management program can be run through the target sub-management units of partitions 3 and 4 at the same time, or through the target sub-management unit of partition 3 or partition 4. In (c), hardware partitions 1-3 are in the startup state, and partition 4 is turned off. At this time, the dynamic thermal management program runs on the target sub-management unit of partition 4. In (d), all hardware partitions are in the power-on state, and the dynamic thermal management program will migrate between the four sub-management units to ensure that it always runs in the target sub-management unit with the lowest load. At the same time, the four sub-management units use inter-core communication to synchronize the load status.

[0051] In the disclosed embodiment, a corresponding sub-management unit can be assigned to each hardware partition in the server, thereby providing a basis for dynamic control of the partitions of the cooling device, and the dynamic cooling management program can be automatically mobilized to run in a low-load target sub-management unit, thereby making full use of idle computing resources to generate a global control strategy.

[0052] In this embodiment, another heat dissipation control method is provided, which can be used for a server, such as a server in a data center. Figure 3 is a flow chart of a heat dissipation control method according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps: Step S301, detecting the status change of each hardware partition in the server, and obtaining status data, wherein the hardware partition includes an independently running host system. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0053] Step S302: allocating control weights of heat dissipation devices to hardware partitions based on the status data.

[0054] Specifically, the above step S302 includes: Step S3021, obtaining a preset weight matrix, wherein the preset weight matrix is ​​used to indicate the original weights of controlling the corresponding heat dissipation devices pre-allocated to each hardware partition.

[0055] Step S3022: adjusting the original weight corresponding to the hardware partition in the preset weight matrix based on the state data to obtain the control weight.

[0056] In the dynamic topology-aware fan allocation mechanism proposed in the embodiment of the present disclosure, the associated weights of the cooling device and the hardware partition can be constructed based on the real-time perception of the server's physical cooling topology and the hardware partition status to obtain a preset weight matrix, thereby achieving dynamic and accurate allocation of control rights based on the preset weight matrix.

[0057] Specifically, the following process may be included when determining the preset weight matrix: Step S31, obtaining position information corresponding to the heat dissipation device.

[0058] Step S32: determining the distance data between the heat dissipation device and the hardware partition based on the location information.

[0059] Step S33 , performing simulation based on the distance data to obtain the original weights of the heat sinks corresponding to the hardware partition controls, and determining a preset weight matrix according to the original weights corresponding to the hardware partitions.

[0060] In the disclosed embodiment, the heat dissipation contribution of the fan to each partition can be quantified by a thermodynamic correlation model. Specifically, first, the distance data is determined according to the physical distance between the heat sink and the hardware partition, so as to define the spatial distance attenuation factor according to the distance data, where the weight value is inversely proportional to the square of the distance. Here, simulation can be performed based on the distance data to obtain the original weight of the hardware partition control corresponding heat sink to produce the best heat dissipation effect. It should be understood that the original weight of a single heat sink to all hardware partitions can be ensured to be 1 by normalization processing.

[0061] For example, if Figure 4 The figure shows a schematic diagram of the arrangement of processors and heat sinks in the hardware partitions of a four-way server, wherein processors 1 to 4 correspond to hardware partitions 1 to 4, and F1 to F6 correspond to heat sinks 1 to 6.

[0062] For example Figure 4For the server arrangement shown, the simulation result corresponding to the heat sink F1 can be: the original weight for processor 1 is 0.6, the original weight for processor 2 is 0, the original weight for processor 3 is 0, and the original weight for processor 4 is 0.4. The simulation result corresponding to the heat sink F3 can be: the original weight for processor 1 is 0.2, the original weight for processor 2 is 0.3, the original weight for processor 3 is 0.3, and the original weight for processor 4 is 0.2. The simulation result corresponding to the heat sink F5 can be: the original weight for processor 1 is 0, the original weight for processor 2 is 0.7, the original weight for processor 3 is 0.3, and the original weight for processor 4 is 0.

[0063] Then, the preset weight matrix determined according to the original weights of the hardware partitions corresponding to the heat sink 1, the heat sink 3 and the heat sink 4 can be expressed as: .

[0064] Step S303: Based on the control weight, the heat dissipation device is controlled to dissipate heat for the hardware partition, so that the state of the hardware partition reaches the target state. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0065] In the embodiments of the present disclosure, the distance data between the heat sink and the hardware partition can be determined based on the position information corresponding to the heat sink, and simulation can be performed based on the distance data to obtain the original weight of the heat sink corresponding to the hardware partition control, and a preset weight matrix can be determined according to the original weights corresponding to each hardware partition. Thus, by combining the hardware position, air duct direction and radiator distribution, a preset weight matrix for the heat sink control that is strongly associated with the physical structure is generated, and higher weights are given to the heat sinks in adjacent hardware partitions to directionally enhance the heat dissipation efficiency, laying the foundation for the optimal allocation of heat dissipation resources.

[0066] In some optional implementations, the above step S3022, adjusting the original weight corresponding to the hardware partition in the preset weight matrix based on the state data to obtain the control weight, includes: Step S41: Based on the status data, determine in the hardware partitions a first hardware partition where the processor is powered on and a second hardware partition where the processor is dormant or powered off.

[0067] Step S42, increasing the original weight corresponding to the first hardware partition in the preset weight matrix, and decreasing the original weight corresponding to the second hardware partition in the preset weight matrix, to obtain a control weight.

[0068] In the disclosed embodiment, the state data may be used to indicate the power-on state, power-off state, and sleep state of the processor in the hardware partition. Here, when the processor is in the power-off state or sleep state, the original weight corresponding to the processor hardware partition may be adjusted to zero, and the extra weight may be allocated to other hardware partitions corresponding to the heat sink, so as to keep the sum of the control weights corresponding to the heat sink unchanged.

[0069] Specifically, the hardware partition where the processor is powered on can be determined as the first hardware partition, and the hardware partition where the processor is dormant or powered off can be determined as the second hardware partition. Then, the above step S42 can be performed to increase the original weight corresponding to the first hardware partition in the preset weight matrix, and reduce the original weight corresponding to the second hardware partition in the preset weight matrix to obtain the control weight, which specifically includes the following process: Step S421: determining, in the heat dissipation device, a target heat dissipation device corresponding to the first hardware partition and the second hardware partition.

[0070] Step S422, reducing the original weight corresponding to the second hardware partition to a preset weight, and increasing the original weight of the first hardware partition based on the proportion of the original weight of the first hardware partition in the target cooling device to obtain a control weight, wherein the preset weight matches the state of the second hardware partition.

[0071] In the embodiment of the present disclosure, if the target heat dissipation device corresponding to the first hardware partition and the second hardware partition is the same, the sum of the control weights corresponding to the first hardware partition and the second hardware partition is 1. On this basis, the above step S422 can be performed to increase the original weight of the first hardware partition based on the proportion of the original weight of the first hardware partition in the target heat dissipation device to obtain the control weight, which specifically includes: Step b1, obtaining the original weight sum corresponding to the first hardware partition.

[0072] Step b2: based on the proportion of the original weight of each first hardware partition in the sum of weights, increase the original weight of the corresponding first hardware partition to obtain the control weight.

[0073] In an embodiment of the present disclosure, the total weight sum of the original weights corresponding to the above-mentioned target heat dissipation devices can be 1, and the original weight sum corresponding to the first hardware partition can be the total weight sum minus the original weight of the second hardware partition. For example, if the weight of the second hardware partition is 0.2, the original weight sum corresponding to the first hardware partition can be 1-0.2=0.8.

[0074] Then, the proportion of the original weight of each first hardware partition in the original weight corresponding to the first hardware partition can be calculated respectively to obtain the control weight corresponding to the first hardware partition in the target cooling device. Specifically, if the original weight corresponding to the first hardware partition is , the original weight corresponding to the second hardware partition is , then the control weight corresponding to the first hardware partition can be expressed as: , ..., .

[0075] For example, Figure 4 Taking the corresponding server arrangement as an example, the rotation speed of the target heat sink F4 is jointly determined by processors 1-4. If the system state changes at this time, processor 3 is in sleep or shut down, a weight redistribution will be triggered, and the control weight of processor 3 will be adjusted to 0, and the control weights of the hardware partitions corresponding to processors 1, 2 and 4 will be redistributed. Here, it can be seen from the above that the simulation results corresponding to the heat sink F3 can be: the original weight for processor 1 is 0.2, the original weight for processor 2 is 0.3, the original weight for processor 3 is 0.3, and the original weight for processor 4 is 0.2. Therefore, the control weight corresponding to hardware partition 1 can be expressed as , the control weight corresponding to hardware partition 2 can be expressed as , the control weight corresponding to hardware partition 4 can be expressed as That is, the rotation speed of the target heat sink F3 is collaboratively controlled by processor 1 (29%), processor 2 (43%), and processor 4 (29%).

[0076] For another example, the rotation speed of the target heat sink F5 is jointly determined by processor 1 and processor 3. After processor 3 goes into sleep or shuts down, the control weight of processor 3 is adjusted to 0, and the control weight of processor 2 is adjusted to 1. That is, the target heat sink F5 will be completely controlled by the local duty cycle of CPU2. At this time, F5 still dissipates heat for processor 2, but is no longer affected by the needs of processor 3.

[0077] It should be understood that the operating state, in addition to indicating the above-mentioned machine state, shutdown state and sleep state, may also include the load state of the hardware partition, such as processor load, memory load, etc. Therefore, the original weight and control weight of the hardware partition may also be adjusted based on the load state of the hardware partition.

[0078] For example, if the load of processor 2 suddenly increases, the system detects that the temperature rise rate of processor 2 exceeds the threshold, and temporarily increases the control weight of the heat sink associated with it. As can be seen from the above, the heat sink associated with processor 2 may be heat sink F3 and heat sink F5. Then, the control weight of the heat sink associated with processor 2 may be adjusted.

[0079] Specifically, the control weight of processor 2 corresponding to heat sink F3 can be increased from 0.43 to 0.6, and the control weights of processor 1 and processor 2 can be reduced to 0.2. In addition, considering that the control weight corresponding to heat sink F5 is already 100%, the duty cycle of heat sink F5 can be allowed to exceed the safety upper limit, for example, from 70% to 90%. It should be understood that the duty cycle of the heat sink can be close to 100%, which is subject to the heat dissipation requirements during actual use.

[0080] In the disclosed embodiment, based on the status data, a first hardware partition where the processor is turned on and a second hardware partition where the processor is dormant or shut down are determined in the hardware partition, so as to increase the original weight corresponding to the first hardware partition in the preset weight matrix, and reduce the original weight corresponding to the second hardware partition in the preset weight matrix, thereby obtaining the control weight, realizing the innovative design of dynamic weights, ensuring the reasonable migration of fan control rights when any processor is dormant or shut down, avoiding systemic heat dissipation risks caused by local failures, and adjusting the server heat dissipation efficiency by further optimizing the initial preset weight matrix, thereby achieving the best heat dissipation device control efficiency on the multi-hardware partition platform.

[0081] In this embodiment, another heat dissipation control method is provided, which can be used for a server, for example, a server in a data center, etc. Figure 5 is a flow chart of a heat dissipation control method according to an embodiment of the present invention. Figure 5 As shown, the process includes the following steps: Step S501, detecting the status change of each hardware partition in the server and obtaining status data, wherein the hardware partition includes an independently running host system. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0082] Step S502: Based on the status data, control weights of the cooling devices are allocated to the hardware partitions. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0083] Step S503: controlling the heat dissipation device to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches the target state.

[0084] Specifically, the above step S503, controlling the heat dissipation device to dissipate heat for the hardware partition based on the control weight, includes: Step S5031, obtaining pulse width modulation parameters output by the hardware partition corresponding to the heat dissipation device, wherein the pulse width modulation parameters are used to control the rotation speed of the fan in the heat dissipation device.

[0085] Step S5032: determining control parameters corresponding to the heat dissipation device based on the control weights corresponding to the pulse width modulation parameters, and controlling the heat dissipation device to dissipate heat for the hardware partition based on the control parameters.

[0086] In the disclosed embodiment, the pulse width modulation parameter may be the above-mentioned PWM duty cycle, and the pulse width modulation parameter may be determined in real time by the sub-management unit corresponding to the hardware partition and output to the heat dissipation device to control the fan speed.

[0087] Specifically, when determining the pulse width modulation parameters corresponding to the hardware partition, the sub-management unit can monitor the temperature of the hardware partition. For example, the sub-management unit will monitor the temperature of the key components in the hardware partition in real time, such as the processor, memory, hard disk, etc. When the temperature of these components rises, the sub-management unit will increase the pulse width modulation parameters according to the preset strategy to increase the fan speed, thereby enhancing the heat dissipation effect and ensuring that the key components operate within a safe temperature range. For another example, the sub-management unit can also consider the overall temperature distribution of the hardware partition. Through the data collected by multiple temperature sensors, the sub-management unit can determine the temperature conditions of different areas in the hardware partition and adjust the pulse width modulation parameters as needed to achieve overall heat dissipation balance. For example, if the temperature at the back of the hardware partition is higher and the temperature at the front is lower, the sub-management unit may adjust the pulse width modulation parameters to allow cold air to enter the server better and hot air to be discharged faster.

[0088] Alternatively, when determining the pulse width modulation parameters corresponding to the hardware partition, the sub-management unit can also monitor the power consumption of the hardware partition, for example, separately monitor the device power consumption and power supply power consumption (if any), and combine other temperature data to comprehensively judge the heat dissipation requirements of the hardware partition, and then adjust the pulse width modulation parameters corresponding to the heat dissipation device to control the fan speed.

[0089] Alternatively, when determining the pulse width modulation parameters corresponding to the hardware partition, the sub-management unit may also monitor the system load of the hardware partition, for example, monitoring the utilization rate of the processor and content. When the hardware partition runs more tasks and the utilization rate of the processor and memory is high, more heat will be generated. At this time, the sub-management unit will increase the pulse width modulation parameters and increase the fan speed to ensure the stable operation of the system. For example, in a data center, when a large number of users access a hardware partition at the same time, the utilization rate of the processor and memory will increase significantly. The sub-management unit will automatically adjust the fan speed to ensure that the hardware partition will not fail due to overheating.

[0090] After the sub-management unit of the hardware partition determines the control weight, the above step S5032 may be performed to determine the control parameter corresponding to the heat dissipation device based on the control weight corresponding to the pulse width modulation parameter, including: Step c1, performing weighted summation on the pulse width modulation parameters based on the control weights to obtain a summation result.

[0091] Step c2, determining the control parameters of the heat dissipation device according to the summation result.

[0092] In the embodiment of the present disclosure, the control parameter may include a PWM duty cycle of a fan in the heat dissipation device, and the control parameter of each heat dissipation device may be expressed as a weighted sum of pulse width modulation parameters output by its corresponding hardware partition.

[0093] For example, it can be seen from the above that the control weight of hardware partition 1 corresponding to the heat sink F1 is 0.6, and the control weight of hardware partition 4 is 0.4. If the pulse width modulation parameter output by hardware partition 1 is , the pulse width modulation parameters of the hardware partition 4 output are Then, the control parameters of the heat sink F1 are It can be expressed as: ( ×0.6)+( ×0.4).

[0094] For another example, it can be seen from the above that the control weight of hardware partition 1 corresponding to heat sink F3 is 0.2, the control weight of hardware partition 2 is 0.3, the control weight of hardware partition 3 is 0.3, and the control weight of hardware partition 4 is 0.2. If the pulse width modulation parameter output by hardware partition 1 is , if the pulse width modulation parameters output by hardware partition 2 are , if the pulse width modulation parameters output by hardware partition 3 are , the pulse width modulation parameters of the hardware partition 4 output are Then, the control parameters of the heat sink F3 are It can be expressed as: ( ×0.2)+( ×0.3)+( ×0.3)+( ×0.2).

[0095] In the embodiments of the present disclosure, the control weights corresponding to the pulse width modulation parameters output by the hardware partition corresponding to the heat sink can be used to determine the control parameters corresponding to the heat sink, and the heat sink can be controlled to cool the hardware partition based on the control parameters, so as to achieve local heat dissipation optimization according to the physical layout of the server through a dynamic topology-aware heat dissipation distribution method. While reducing the average speed of the fans in the heat sink and extending its service life, it can also reduce hardware protective downtime triggered by over-temperature, thereby significantly improving server availability.

[0096] In some optional implementations, the above step S503 further includes: Step S51 : based on the load change of the hardware partitions, predict a first target hardware partition whose change rate exceeds a load change threshold.

[0097] Step S52: Obtain a load prediction result for the first target hardware partition.

[0098] Step S53: adjusting the pulse width modulation parameters output by the hardware partition based on the load prediction result.

[0099] In the disclosed embodiment, considering that the hardware partition with a faster load change rate means that the temperature will rise or fall faster in the future, the first target hardware partition whose change rate exceeds the load change threshold can be predicted based on the load change of the hardware partition. Here, if the load prediction result is a load oyster, it means that the temperature of the hardware partition will rise, so the pulse width modulation parameter can be increased to increase the speed of the fan in the heat sink, otherwise, the pulse width modulation parameter can be reduced to reduce the speed of the fan in the heat sink.

[0100] Here, the predicted change rate of the hardware partition can be predicted by a machine learning model or the like to obtain a load prediction result. The machine learning model can be a neural network, a decision tree, a random forest, or other model that has strong learning ability and adaptability and can handle complex nonlinear relationships.

[0101] Specifically, first, hardware partition data can be collected, such as performance indicator data, application data, system event data, etc. Among them, performance indicator data can be obtained through server monitoring tools or operating system indicators, such as processor utilization, memory utilization, network traffic and other key performance indicators. Historical data, these data can intuitively reflect the load situation of the hardware partition and are the basis for prediction.

[0102] Then, the hardware partition data can be preprocessed, for example, outliers in the data can be identified and removed. These outliers may be caused by server failures, network attacks or other abnormal situations, which will interfere with the prediction results. For missing values ​​in the data, methods such as average filling and interpolation filling can be used to ensure the integrity of the data. In order to reduce data fluctuations and noise, the data can be smoothed, such as the moving average method.

[0103] Next, data feature extraction can be performed, such as time feature extraction, historical load feature extraction, etc. Here, time is an important feature, including date, hour, minute, etc. The server load usually shows a certain time pattern. For example, the morning and afternoon of each day may be the peak working hours, while the evening is relatively low; the load is higher on weekdays and lower on weekends. By analyzing the time characteristics, these periodic changes can be captured. Past load data is an important basis for predicting future loads. Statistics such as average load, maximum load, minimum load, and load change trends and fluctuations over a period of time can be calculated. Historical load characteristics can help understand the load characteristics and change patterns of hardware partitions.

[0104] After selecting the corresponding machine learning model, the machine learning model can be monitored and updated in real time. Specifically, during the actual operation, the server load and other relevant indicators can be monitored in real time, and the real-time data can be input into the prediction model to obtain real-time prediction results. In this way, the changing trend of the load can be discovered in time, and corresponding measures can be taken to make adjustments. At the same time, as time goes by and the business changes, the load characteristics of the server may also change. Therefore, it is necessary to regularly update and optimize the prediction model and retrain the model to adapt to new data and situations. The incremental learning method can be used to learn and update only the newly added data without retraining the entire model, thereby improving the update efficiency of the model.

[0105] In the disclosed embodiment, considering that a hardware partition with a faster load change rate means that the temperature will rise or fall faster in the future, the first target hardware partition whose change rate exceeds the load change threshold can be predicted based on the load change of the hardware partition, so that the pulse width modulation parameters output by the hardware partition can be adjusted based on the load prediction result for the first target hardware partition, thereby dynamically adjusting the pulse width modulation parameters of the hardware partition, so as to adjust the fan speed before the temperature of the hardware partition rises or falls, thereby compensating for the lag in controlling the heat dissipation device according to the temperature.

[0106] In some optional embodiments, the above Figure 1 The corresponding implementation method also includes: Step S61, obtaining a preset heat dissipation model corresponding to the server.

[0107] Step S62: Based on the preset heat dissipation model, predict a second target hardware partition in the hardware partition whose temperature exceeds the temperature threshold.

[0108] Step S63: Obtain a temperature prediction result for the second target hardware partition.

[0109] Step S64: adjusting the pulse width modulation parameters output by the second target hardware partition based on the temperature prediction result.

[0110] In the disclosed embodiment, although the control weight allocation of dynamic topology perception can achieve local heat dissipation optimization according to the physical layout of the server, it is still open-loop control in essence, and it is difficult to cope with complex scenarios such as sudden thermal load surge, multi-partition thermal coupling effect and global power consumption constraints. Specifically, first of all, the problem of thermal inertia and hysteresis, dynamic topology perception only relies on the current temperature state to allocate fan resources, and the delay of heat conduction may cause local intervention to lag behind the actual thermal risk. Secondly, there is competition and conflict among multiple heat sources. When multiple partitions need to speed up heat dissipation at the same time, dynamic topology perception may cause fan resource competition due to the lack of a global coordination mechanism, causing the power consumption of the whole machine to exceed the limit or the heat dissipation efficiency to decrease. At the same time, due to the strong coupling of the temperature fields of each partition in the server, the static weight allocation of dynamic topology perception cannot correct the cross-zone thermal interference in real time. Therefore, based on closed-loop control based on real-time temperature feedback and PID (Proportional-Integral-DerivativeControl Algorithm, i.e., proportional-integral-differential control) algorithm, the control weight of the hardware partition and the output pulse width modulation parameters can be dynamically adjusted to suppress the temperature oscillation caused by thermal coupling and ensure that the system converges quickly to a steady state.

[0111] Specifically, the heat dissipation of the server can be modeled in advance to obtain a preset heat dissipation model to predict the second target hardware partition whose temperature exceeds the temperature threshold in the hardware partition. Here, the optimal scheduling of heat dissipation resources for multiple heat source partitions in the server can be achieved through dynamic weight allocation and global power consumption constraints. For example, a preset heat dissipation model is established based on the partial differential equation of the temperature field distribution inside the server based on Fourier's law of heat conduction. The preset heat dissipation model can be expressed as: , where T is the temperature field distribution function, α is the thermal diffusion coefficient, Q is the total heat generation power of the processor, and ρ is the air density. is the specific heat capacity of air.

[0112] Then, the above step S62 may be performed to predict, based on the preset heat dissipation model, a second target hardware partition whose temperature exceeds the temperature threshold in the hardware partition, specifically including: Step d1, discretize the internal space of the server into a grid to obtain a target grid corresponding to each hardware partition.

[0113] Step d2, predicting the temperature increase of the target grid within a preset period of time based on a preset heat dissipation model, and obtaining a temperature prediction result corresponding to the hardware partition.

[0114] Step d3, obtaining a temperature threshold, and determining in the hardware partition a second target hardware partition whose temperature prediction result exceeds the temperature threshold.

[0115] In the disclosed embodiment, the internal space of the server can be grid discretized to divide the internal space of the server into non-uniform grid units, where the grid densities of different areas in the server are different. For example, the target grid corresponding to the hardware partition can be denser to improve the accuracy of temperature prediction, and the idle area and the corresponding grid can be sparser to save computing power consumption.

[0116] Specifically, the internal space of the server can be discretized by the finite difference method to solve the future The temperature change trend over time is used to predict the second target hardware partition with potential thermal risk. Here, the temperature prediction result of the target grid can be expressed as ,in, Indicates grid cells at time step The temperature value, is the spatial step length, is the time step. Through iterative solution, the preset period is predicted Then, the pulse width modulation parameters of the second target hardware partition output are generated by the proportional-integral-derivative (PID) algorithm.

[0117] Specifically, the pulse width modulation parameters output by the second target hardware partition can be expressed as . Where u(t) is the PWM duty cycle output, is the temperature deviation, Set the temperature threshold.

[0118] Then, based on the comparison result between the temperature prediction result and the temperature threshold, the pulse width modulation parameters output by the second target hardware partition can be adjusted. Specifically, the above CPLD can receive the temperature data of the hardware partition sent by each sub-management unit in real time, and compare it with the predicted value. When the temperature change rate of a hardware partition exceeds the temperature threshold (such as ), the fan speed adjustment is triggered to increase the control weight of the heat dissipation device associated with the thermal emergency zone. When the temperature of the thermal emergency zone falls back to the safe range ( ), the CPLD restores the weight distribution in an exponential decay manner.

[0119] In the disclosed embodiment, the pulse width modulation parameters of the second target hardware partition output can be adjusted based on the temperature prediction result of the temperature prediction model, thereby realizing a cross-partition thermal collaborative closed-loop control method. Through the collaborative design of thermal field prediction modeling, dynamic speed regulation and closed-loop feedback correction, multi-partition collaborative decision-making is achieved with the help of CPLD, and ambient temperature fluctuations are suppressed through PID algorithm control.

[0120] In summary, in the disclosed embodiment, the state change of each hardware partition in the server can be detected first to obtain state data, wherein the hardware partition includes an independently running host system, and based on the state data, the control weight of the heat sink is allocated to the hardware partition. Then, the heat sink can be controlled to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches the target state, thereby dynamically allocating the control weight of the heat sink to dynamically optimize the heat dissipation power of the global heat sink to avoid local overheating or power consumption waste of the server.

[0121] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0122] The embodiments of the present application also provide a heat dissipation control device, which is used to implement the above embodiments and preferred implementation modes, and will not be repeated hereafter. As used below, the term "module" may implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0123] This embodiment provides a heat dissipation control device, such as Figure 6 As shown, including: The detection module 601 is used to detect the status change of each hardware partition in the server and obtain status data, wherein the hardware partition includes an independently running host system; An allocation module 602, for allocating control weights of heat sinks to hardware partitions based on the status data; The heat dissipation module 603 is used to control the heat dissipation device to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches the target state.

[0124] In some optional implementations, the allocation module 602 is further configured to: Obtaining a preset weight matrix, wherein the preset weight matrix is ​​used to indicate the original weights of controlling the corresponding heat sinks that are pre-assigned to each hardware partition; The original weight corresponding to the hardware partition in the preset weight matrix is ​​adjusted based on the state data to obtain the control weight.

[0125] In some optional implementations, the allocation module 602 is further configured to: Obtaining position information corresponding to the heat dissipation device; Based on the location information, determine distance data between the heat sink and the hardware partition; Based on the distance data, simulation is performed to obtain the original weights of the heat sinks corresponding to the hardware partition control, and a preset weight matrix is ​​determined according to the original weights corresponding to the various hardware partitions.

[0126] In some optional implementations, the allocation module 602 is further configured to: Based on the state data, determining in the hardware partitions a first hardware partition in which the processor is powered on and a second hardware partition in which the processor is dormant or powered off; The original weight corresponding to the first hardware partition in the preset weight matrix is ​​increased, and the original weight corresponding to the second hardware partition in the preset weight matrix is ​​decreased to obtain the control weight.

[0127] In some optional implementations, the allocation module 602 is further configured to: Determine, in the heat dissipation device, a target heat dissipation device corresponding to the first hardware partition and the second hardware partition; The original weight corresponding to the second hardware partition is reduced to a preset weight, and based on the proportion of the original weight of the first hardware partition in the target heat dissipation device, the original weight of the first hardware partition is increased to obtain a control weight, wherein the preset weight matches the state of the second hardware partition.

[0128] In some optional implementations, the allocation module 602 is further configured to: Obtaining the original weight sum corresponding to the first hardware partition; Based on the proportion of the original weight of each first hardware partition in the sum of weights, the original weight of the corresponding first hardware partition is increased to obtain the control weight.

[0129] In some optional embodiments, the device is also used for: Before allocating control weights of heat dissipation devices to hardware partitions, partitioning management units in the server based on the hardware partitions to obtain sub-management units, wherein the management units are used to perform global heat dissipation management based on the operating status of the server; A corresponding sub-management unit is allocated to the hardware partition, wherein the sub-management unit is used to control a heat dissipation device corresponding to the hardware partition.

[0130] In some optional embodiments, the device is also used for: Determine, among the sub-management units, a target sub-management unit whose load is lower than a preset threshold; The dynamic thermal management program is run based on the target sub-management unit to perform the step of allocating control weights of the thermal devices to the hardware partitions based on the state data based on the dynamic thermal management program.

[0131] In some optional embodiments, the device is also used for: When the running time of the dynamic heat dissipation management program meets the load migration condition, the target sub-management unit is determined based on the sub-management unit with the current lowest load.

[0132] In some optional embodiments, the device is also used for: When there are multiple target sub-management units, obtaining the inter-core communication efficiency of the target sub-management units, and determining whether the inter-core communication efficiency meets the parallel processing condition; If so, jointly running a dynamic thermal management program based on the target sub-management unit; If not, a dynamic thermal management program is run based on any target sub-management unit.

[0133] In some optional implementations, the heat dissipation module 603 is further used for: Obtaining pulse width modulation parameters output by the hardware partition corresponding to the heat dissipation device, wherein the pulse width modulation parameters are used to control the speed of a fan in the heat dissipation device; Based on the control weight corresponding to the pulse width modulation parameter, the control parameter corresponding to the heat dissipation device is determined, and based on the control parameter, the heat dissipation device is controlled to dissipate heat for the hardware partition.

[0134] In some optional implementations, the heat dissipation module 603 is further used to: perform weighted summation on the pulse width modulation parameters based on the control weight to obtain a summation result; The control parameters of the heat dissipation device are determined according to the summation result.

[0135] In some optional implementations, the heat dissipation module 603 is further used to: predict, based on the load change of the hardware partition, a first target hardware partition whose change rate exceeds a load change threshold; Obtaining a load prediction result for the first target hardware partition; Based on the load prediction results, the pulse width modulation parameters of the hardware partition output are adjusted.

[0136] In some optional implementations, the heat dissipation module 603 is further used to: obtain a preset heat dissipation model corresponding to the server; Based on the preset heat dissipation model, predicting in the hardware partition a second target hardware partition whose temperature exceeds a temperature threshold; Obtaining a temperature prediction result for a second target hardware partition; Based on the temperature prediction result, the pulse width modulation parameters output by the second target hardware partition are adjusted.

[0137] In some optional implementations, the heat dissipation module 603 is further used to: discretize the internal space of the server into a grid to obtain a target grid corresponding to each hardware partition; Based on the preset heat dissipation model, the temperature rise of the target grid within the preset period is predicted to obtain the temperature prediction result corresponding to the hardware partition; A temperature threshold is obtained, and a second target hardware partition whose temperature prediction result exceeds the temperature threshold is determined in the hardware partition.

[0138] In some optional implementations, the heat dissipation module 603 is further configured to: adjust the pulse width modulation parameters output by the second target hardware partition based on a comparison result between the temperature prediction result and the temperature threshold.

[0139] For the description of the features in the embodiment corresponding to the heat dissipation control device, reference can be made to the relevant description of the embodiment corresponding to the heat dissipation control method, which will not be repeated here.

[0140] The embodiment of the present application also provides an electronic device, such as Figure 7 As shown, it includes a memory 10 and a processor 20, the memory 10 stores a computer program, and the processor 20 is configured to run the computer program to execute the steps in any of the above-mentioned heat dissipation control method embodiments.

[0141] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above-mentioned heat dissipation control method embodiments when running.

[0142] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0143] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any one of the above-mentioned heat dissipation control method embodiments are implemented.

[0144] An embodiment of the present application also provides another computer program product, including a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in any of the above-mentioned heat dissipation control method embodiments are implemented.

[0145] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0146] The above is a detailed introduction to a heat dissipation control method, device, computer equipment, storage medium and program product provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection required by the present application.

Claims

1. A heat dissipation control method, characterized in that: The method comprises: Detecting status changes of various hardware partitions in the server to obtain status data, wherein the hardware partitions include independently running host systems; Based on the status data, assigning a control weight of a heat sink to the hardware partition; The heat dissipation device is controlled based on the control weight to dissipate heat for the hardware partition, so that the state of the hardware partition reaches a target state.

2. The method according to claim 1, characterized in that The allocating a control weight of a heat dissipation device to the hardware partition based on the status data includes: Obtaining a preset weight matrix, wherein the preset weight matrix is ​​used to indicate the original weights of controlling the corresponding heat sinks that are pre-assigned to each hardware partition; The original weight corresponding to the hardware partition in the preset weight matrix is ​​adjusted based on the state data to obtain the control weight.

3. The method according to claim 2, characterized in that The method further comprises: Obtaining position information corresponding to the heat dissipation device; Based on the location information, determining distance data between the heat dissipation device and the hardware partition; A simulation is performed based on the distance data to obtain the original weights of the heat sinks corresponding to the hardware partition controls, and the preset weight matrix is ​​determined according to the original weights corresponding to the various hardware partitions.

4. The method according to claim 2, characterized in that: The adjusting the original weight corresponding to the hardware partition in the preset weight matrix based on the state data to obtain the control weight includes: Based on the state data, determining in the hardware partitions a first hardware partition in which a processor is turned on and a second hardware partition in which a processor is dormant or turned off; The original weight corresponding to the first hardware partition in the preset weight matrix is ​​increased, and the original weight corresponding to the second hardware partition in the preset weight matrix is ​​decreased to obtain the control weight.

5. The method according to claim 4, characterized in that The increasing the original weight corresponding to the first hardware partition in the preset weight matrix and reducing the original weight corresponding to the second hardware partition in the preset weight matrix to obtain the control weight includes: Determining, in the heat dissipation device, a target heat dissipation device corresponding to the first hardware partition and the second hardware partition; The original weight corresponding to the second hardware partition is reduced to a preset weight, and based on the proportion of the original weight of the first hardware partition in the target cooling device, the original weight of the first hardware partition is increased to obtain the control weight, wherein the preset weight matches the operating state of the second hardware partition.

6. The method according to claim 5, characterized in that The step of increasing the original weight of the first hardware partition based on the proportion of the original weight of the first hardware partition in the target heat dissipation device to obtain the control weight includes: Obtaining the original weight sum corresponding to the first hardware partition; Based on the proportion of the original weight of each of the first hardware partitions in the sum of the weights, the original weight of the corresponding first hardware partition is added to obtain the control weight.

7. The method according to claim 1, characterized in that The method further comprises: Before allocating the control weight of the heat dissipation device to the hardware partition, partitioning the management unit in the server based on the hardware partition to obtain sub-management units, wherein the management unit is used to perform global heat dissipation management based on the operating status of the server; A corresponding sub-management unit is allocated to the hardware partition, wherein the sub-management unit is used to control a heat dissipation device corresponding to the hardware partition.

8. The method according to claim 7, characterized in that The method further comprises: Determining, among the sub-management units, a target sub-management unit whose load is lower than a preset threshold; A dynamic thermal management program is run based on the target sub-management unit to perform the step of allocating control weights of the thermal devices to the hardware partitions based on the status data based on the dynamic thermal management program.

9. The method according to claim 8, characterized in that The determining, in the sub-management unit, of a target sub-management unit whose load is lower than a preset threshold comprises: When the running time of the dynamic heat dissipation management program meets the load migration condition, the target sub-management unit is determined based on the sub-management unit with the current lowest load.

10. The method according to claim 8, characterized in that The running of the dynamic heat dissipation management program based on the target sub-management unit includes: When the number of the target sub-management units is multiple, obtaining the inter-core communication efficiency of the target sub-management units, and determining whether the inter-core communication efficiency meets the parallel processing condition; If so, jointly running the dynamic heat dissipation management program based on the target sub-management unit; If not, the dynamic thermal management program is run based on any target sub-management unit.

11. The method according to claim 1, characterized in that: The controlling the heat dissipation device to dissipate heat for the hardware partition based on the control weight includes: Acquire a pulse width modulation parameter output by a hardware partition corresponding to the heat dissipation device, wherein the pulse width modulation parameter is used to control a rotation speed of a fan in the heat dissipation device; Based on the control weight corresponding to the pulse width modulation parameter, the control parameter corresponding to the heat dissipation device is determined, and based on the control parameter, the heat dissipation device is controlled to dissipate heat for the hardware partition.

12. The method according to claim 11, characterized in that The determining the control parameter corresponding to the heat dissipation device based on the control weight corresponding to the pulse width modulation parameter includes: Performing weighted summation on the pulse width modulation parameters based on the control weight to obtain a summation result; The control parameters of the heat dissipation device are determined according to the summation result.

13. The method according to claim 11, characterized in that The method further comprises: Based on the load change of the hardware partition, predict a first target hardware partition whose change rate exceeds a load change threshold; Obtaining a load prediction result for the first target hardware partition; Based on the load prediction result, the pulse width modulation parameters output by the hardware partition are adjusted.

14. The method according to claim 11, characterized in that The method further comprises: Obtaining a preset heat dissipation model corresponding to the server; Based on the preset heat dissipation model, predicting a second target hardware partition in the hardware partition whose temperature exceeds a temperature threshold; Obtaining a temperature prediction result for the second target hardware partition; Based on the temperature prediction result, a pulse width modulation parameter output by the second target hardware partition is adjusted.

15. The method according to claim 14, characterized in that The step of predicting, based on the preset heat dissipation model, in the hardware partition, a second target hardware partition whose temperature exceeds a temperature threshold, comprises: Discretizing the internal space of the server into a grid to obtain a target grid corresponding to each of the hardware partitions; Predicting the temperature increase of the target grid within a preset period of time based on the preset heat dissipation model to obtain the temperature prediction result corresponding to the hardware partition; A temperature threshold is obtained, and a second target hardware partition whose temperature prediction result exceeds the temperature threshold is determined in the hardware partition.

16. The method according to claim 15, characterized in that The method further comprises: Based on a comparison result between the temperature prediction result and the temperature threshold, an adjustment operation is performed on a pulse width modulation parameter output by the second target hardware partition.

17. A heat dissipation control device, characterized in that: The device comprises: A detection module, used to detect the status change of each hardware partition in the server and obtain status data, wherein the hardware partition includes an independently running host system; an allocation module, configured to allocate a control weight of a heat dissipation device to the hardware partition based on the status data; The heat dissipation module is used to control the heat dissipation device to dissipate heat for the hardware partition based on the control weight, so that the state of the hardware partition reaches a target state.

18. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the heat dissipation control method according to any one of claims 1 to 16 by executing the computer instructions.

19. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the heat dissipation control method according to any one of claims 1 to 16.

20. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions are used to enable a computer to execute the heat dissipation control method according to any one of claims 1 to 16.

Images