Server and power consumption management method thereof, electronic device, medium and program product

Abstract

The application discloses a kind of server and its power consumption management method, electronic equipment, medium and program product, it is related to server technical field.Processing, method includes according to the current power consumption configuration mode of each power supply unit state information and the power supply configuration parameter information of pre-set determined in server running process, if it is first power consumption configuration mode, based on working current safety range and each power supply unit rated power consumption Determine target power consumption ceiling value, and update power consumption limit value strategy according to target power consumption ceiling value;If it is second power consumption configuration mode, according to the rated power consumption of each normal power supply power supply unit and the preset power consumption safety adjustment factor Determine transition power consumption ceiling value, and update power consumption limit value strategy according to transition power consumption ceiling value.The application can solve the problem that related art cannot simultaneously satisfy that power supply does not occur failure and power supply can exert maximum performance, can dynamically adjust server power consumption limit value according to power supply state.

Description

Technical Field

[0001] This invention relates to the field of server technology, and in particular to a server and its power management method, electronic device, media, and software product. Background Technology

[0002] The high availability of servers depends on the normal operation of power supply units such as PSUs (Power Supply Units). In order to avoid power failure, related technologies set fixed power consumption limits based on the PSU status, which cannot simultaneously ensure that the power supply does not fail and can achieve maximum performance. Summary of the Invention

[0003] This invention provides a server power consumption and management method, electronic device, non-volatile storage medium, and computer program product that can dynamically adjust the server power consumption limit according to the power supply status. This not only effectively avoids unexpected server shutdowns caused by power failures, but also maximizes power performance.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0005] This invention provides a server power consumption management method, applied to a management controller, comprising:

[0006] Obtain power configuration parameter information and the safe operating current range of the power supply unit; switch power consumption configuration modes according to the power configuration parameter information and the status information of each power supply unit during operation; in the first power consumption configuration mode, based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range, determine the target power consumption cap value according to the rated power consumption of each power supply unit, and update the power consumption limit strategy according to the target power consumption cap value; in the second power consumption configuration mode, determine the transition power consumption cap value according to the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, and update the power consumption limit strategy according to the transition power consumption cap value, where the preset power consumption safety adjustment factor is less than 1.

[0007] This invention provides a server power management method, applied to logic devices, comprising:

[0008] Upon receiving power configuration parameter information, the system polls each power supply unit to obtain its status information. If the number of power supply units in the "in-place" state is less than or equal to the number of required power supply units, and at least one power supply unit's status information includes an alarm signal, a notification is sent indicating that the power configuration mode corresponding to the power supply status is the second power configuration mode. This determines the transitional power consumption cap based on the rated power consumption of each normally powered power supply unit and a preset power safety adjustment factor, and updates the power limit strategy based on the transitional power consumption cap. The preset power safety adjustment factor is less than 1. The power configuration mode corresponds to the first power configuration mode, which, based on triggering overcurrent warning protection within the operating current safety range without triggering overcurrent protection within the operating current safety range, determines the target power consumption cap based on the rated power consumption of each power supply unit, and updates the power limit strategy based on the target power consumption cap. If the status information of each power supply unit matches the power configuration parameter information, then, based on triggering overcurrent warning protection within the operating current safety range without triggering overcurrent protection within the operating current safety range, the target power consumption cap is determined based on the rated power consumption of each power supply unit, and the power limit strategy is updated based on the target power consumption cap.

[0009] The present invention also provides an electronic device, including a memory and a processor, wherein the processor is used to implement the steps of any of the above-described server power management methods when executing a computer program stored in the memory.

[0010] The present invention also provides a non-volatile storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of any of the above-described server power management methods.

[0011] The present invention also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of any of the above-described server power management methods.

[0012] Finally, the present invention also provides a server, including a logic device, a management controller, and a processor; the logic device is connected to the management controller via a first bus and is used to implement the steps of any of the above-described server power management methods when executing a computer program; the logic device includes at least a power supply status monitoring interface and a frequency reduction interaction interface, the power supply status monitoring interface being connected to each power supply unit, the frequency reduction interaction interface being connected to the processor, and the frequency reduction interaction interface being used to send a frequency reduction throttling trigger signal to the processor; the management controller is used to implement the steps of any of the server power management methods when executing a computer program; the management controller includes a power execution manager, the power execution manager being connected to the processor, the power execution manager updating the power limit policy according to a target power cap or a transitional power cap, and sending it to the processor.

[0013] The advantages of the technical solution provided by this invention are that by configuring the number of power supply units, it can not only support dynamic configuration of the power consumption cap, but also ensure the stability of the server's power supply. By comparing the currently available power supply units and their status with the power parameter configuration information, it is determined whether the server is in a normal or abnormal state, and then the power consumption configuration mode to be switched is determined. Combined with the power supply situation under different states, the power consumption limit matching the current power supply state is adjusted, realizing dynamic adjustment of the server's power consumption limit according to the power supply state. This not only effectively avoids unexpected server downtime caused by power failure, ensuring business continuity, but also ensures that hot replacement of power supply units does not affect business continuity, improving the high availability of the server. Furthermore, it maximizes power supply performance, reduces server power consumption, and maximizes resource utilization, which is beneficial for deployment in edge scenarios with limited power supply environments.

[0014] Furthermore, this invention also provides corresponding electronic devices, non-volatile storage media, computer program products, and servers for implementing server power management methods, further making the method more practical. The electronic devices, non-volatile storage media, computer program products, and servers have corresponding advantages. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the present invention or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic diagram of the hardware composition framework applicable to the server power consumption management method provided by the present invention;

[0017] Figure 2 A flowchart illustrating a server power consumption management method provided by the present invention;

[0018] Figure 3 A schematic diagram illustrating the execution flow of the power management process of the management controller provided by the present invention;

[0019] Figure 4 This is a schematic diagram of the power callback process provided by the present invention;

[0020] Figure 5 A flowchart illustrating another server power consumption management method provided by the present invention;

[0021] Figure 6 This is a structural framework diagram of an exemplary embodiment of the server power management device provided by the present invention;

[0022] Figure 7 A structural framework diagram of another exemplary embodiment of the server power management device provided by the present invention;

[0023] Figure 8 A structural diagram of an exemplary embodiment of the electronic device provided by the present invention;

[0024] Figure 9 This is a structural framework diagram of an exemplary embodiment of the server provided by the present invention;

[0025] Figure 10 This is a schematic diagram of the data interaction process of internal server components provided by the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. In this specification and the aforementioned drawings, the terms "first," "second," "third," "fourth," etc., are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. The term "exemplary" means "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as superior to or better than other embodiments.

[0027] With the rapid development of cloud computing, artificial intelligence, big data analytics, and high-performance computing, the scale and computing density of data centers are constantly increasing, and they are being applied in various business scenarios such as real-time financial trading platforms, global e-commerce platforms, various enterprise information systems, life science research, and autonomous driving.

[0028] To meet the needs of various business scenarios, servers used to support these services require continuous availability and long-term stable operation. Continuous availability refers to providing near-uninterrupted service to business applications. Server downtime, even for a brief interruption lasting only a few minutes, can have serious consequences. For example, a financial trading platform might miss an instruction, leading to user losses; a large-scale online service outage could damage reputation and user trust; disruptions to industrial control processes could cause safety accidents or production losses; and a medical system shutdown could endanger patient safety. Long-term stable operation means that servers need to run 24 / 7 without interruption, with a lifespan of several years. This not only requires the server to function normally at startup but also to maintain high stability, performance consistency, and reliability during operation, even with factors such as component aging, environmental temperature fluctuations, and cyclical load changes. Any performance degradation or unplanned downtime of the server will significantly increase maintenance costs and affect service quality agreements.

[0029] The Power Supply Unit (PSU), as a server power supply component, efficiently, stably, and safely converts AC mains power into DC power required by various computing components within the server, providing precise voltage rails (such as 12V, 3V, etc.). The power electronic components (such as capacitors and magnets), control circuits, and cooling systems within the PSU, operating under abnormal conditions such as high temperature, high load, and power grid fluctuations, are susceptible to failures such as capacitor aging, power device failure, and fan stoppage. PSU failure can impact the server's business continuity and long-term stable operation. Therefore, high server availability depends on a properly functioning power supply. To prevent unexpected server downtime due to power failure, related technologies set fixed power consumption limits (i.e., power caps) based on the PSU's status. These limits prevent excessive power consumption from causing power failure. However, to ensure that power failure is not caused by excessive server power consumption in any scenario, these power consumption limits are usually quite conservative, which means the power supply's maximum performance cannot be fully utilized. In addition, due to power supply aging, instantaneous power consumption spikes, phase imbalance, and other scenarios, the actual power consumption of the server may not reach the specified power consumption cap, which may also lead to server power outages. It can be seen that the relevant technologies cannot simultaneously ensure that the power supply does not fail and can achieve maximum performance.

[0030] Therefore, this invention supports configuring the required number of power supply units and the number of redundant power supply units to maintain normal server operation. In other words, it allows configuration of the total number of power supply units and the minimum number of power supply units required for normal operation. The management controller provides switching between two operating modes: stable and abnormal. Based on the acquired power supply unit status information and the aforementioned power configuration parameters, if the current normal power supply status meets the power supply status provided by the power configuration parameters, it operates in a stable state. In this case, the power consumption limit setting process is as follows: based on triggering the overcurrent warning protection corresponding to the operating current safety range without triggering the overcurrent protection corresponding to the operating current safety range, the target power consumption cap is determined according to the rated power consumption of each power supply unit. If this is not met, the power consumption cap is re-determined based on the rated power consumption of each normally powered power supply unit and a preset power consumption safety adjustment factor. This allows for dynamic adjustment of the server power consumption limit according to the power supply status, effectively preventing unexpected server shutdowns due to power failures and maximizing power supply performance.

[0031] The specific application environment architecture or hardware architecture upon which the server power management method depends is described here. The following section will combine... Figure 1 Examples of possible application scenarios related to the technical solutions of this invention are provided below:

[0032] For servers in cloud data centers, or servers in data centers used to support services with high business continuity requirements (such as financial transactions), or servers in distributed AI training platforms, or servers in edge computing platforms, the hardware architecture used for power management in these servers is as follows:

[0033] The BMC (Baseboard Management Controller) visualization page adds a dynamic power capping configuration option and defines a user-facing management interface. The BMC also features a power execution manager connected to the processor. The power execution manager updates the power limit policy based on the target or transitional power cap and sends the updated policy to the central processing unit. The BMC on the motherboard is connected to the CPLD (Complex Programmable Logic Device) via I2C (Inter-Integrated Circuit). The BMC uses this bus to send power limit policies, power configuration information, and throttling interrupt signals to the CPLD and receives throttling notification signals from the CPLD. The CPLD includes a power status monitoring interface, a frequency reduction interaction interface, and a direct-connect frequency reduction interface. Each PSU must connect at least the signal lines corresponding to PS_PRESENT# (present signal), PSU_ALERT_N (alarm signal), VIN Good (power supply voltage signal), and PSU_PWROK (power supply unit voltage signal) to the power supply status monitoring interface. The CPLD connects to the PRC HOT (Processor Hot) and MEM HOT (Memory Hot) input pins of the central processing unit (CPU) via a frequency reduction interface. Upon receiving a frequency reduction signal from the CPLD, the CPU can be triggered to perform frequency reduction and throttling operations. Furthermore, the CPLD has reserved lines to connect to the B30 of a PCIE (Peripheral Component Interconnect Express) card with power brake functionality and the A70 of an OCP (Open Compute Project) 3.0 network interface card. Upon receiving a frequency reduction signal from the CPLD, the frequency reduction and throttling operation can be triggered.

[0034] During server operation, users manually enable the dynamic power capping configuration function through the power capping enable option on the visual page, and input power configuration parameter information through the user management interface. This power configuration parameter information includes the required number of power supply units and the number of redundant power supply units to maintain normal server operation. The Power Management Control (BMC) obtains the power configuration parameter information and the operating current safety range of the power supply units. The BMC periodically polls the presence status and status word of each PSU. If the status information of each power supply unit during operation matches the power configuration parameter information, then based on triggering the overcurrent warning protection corresponding to the operating current safety range without triggering the overcurrent protection corresponding to the operating current safety range, the target power capping value is determined according to the rated power consumption of each power supply unit, and the power limit strategy is updated according to the target power capping value. If the status information of each power supply unit during operation does not match the power configuration parameter information, then based on the rated power consumption of each normally powered power supply unit and the preset power safety adjustment factor, a transitional power capping value is determined, and the power limit strategy is updated according to the transitional power capping value. The preset power safety adjustment factor is less than 1. The CPLD receives power configuration parameter information from the BMC and polls each power supply unit in real time based on this information to obtain its status. If the number of power supply units in the active state is less than or equal to the required number, and at least one power supply unit's status information includes an alarm signal, a frequency reduction and throttling trigger signal and a throttling notification signal are generated. If the operating frequency to be reduced is that of the CPU, the frequency reduction and throttling trigger signal is sent directly to the CPU's PRC HOT and MEM HOT input pins via the frequency reduction interface. If it is a peripheral device, the signal is sent to the corresponding peripheral device via the direct-connect frequency reduction interface, and a throttling notification signal is sent to the BMC so that the BMC can determine that the status information of each power supply unit does not match the power configuration parameter information. The BMC polls the motherboard CPLD via I2C to record dynamic throttling operations triggered after the dynamic capping configuration function is enabled.

[0035] Through the aforementioned three-level collaborative architecture (Hardware-level Real-time Monitoring (CPLD) → Intelligent Decision-Making (BMC) → Policy Execution Power Consumption Manager), the response time of this invention is shortened to 100ns, supporting real-time dynamic power cap adjustment and full hot maintenance. In contrast, independent BMC management requires a response time >800ms, offers limited dynamic adjustment, and does not support hot maintenance. For the hardware protection circuit implementation, the response time is 10μs, and it does not support dynamic adjustment of the power cap or hot maintenance. For the software-defined power supply implementation, the response time is greater than 1s, supporting dynamic adjustment of the power cap and partial hot maintenance. Therefore, this invention can meet the 99.999% availability and millisecond-level fault response requirements of cloud data centers, achieving zero downtime maintenance and ensuring business continuity. It can also meet the burst power management needs of high-performance devices such as image processors and accelerator cards in AI distributed training systems, achieving overcurrent protection, collaborative power control of heterogeneous computing units, and meeting the power constraints of edge computing platforms, enabling dynamic derating.

[0036] It should be noted that the above application scenarios are only shown to facilitate understanding of the ideas and principles of the present invention, and the embodiments of the present invention are not limited in any way. On the contrary, the embodiments of the present invention can be applied to any applicable scenario. After introducing the technical solution of the present invention, various non-limiting embodiments of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] Please see first. Figure 2 , Figure 2 This is a flowchart illustrating a server power management method provided in this embodiment. This embodiment applies to the server's management controller, which may be, for example, a BMC and may include the following:

[0038] S201: Obtain power configuration parameter information and the safe operating current range of the power supply unit, and switch the power consumption configuration mode according to the power configuration parameter information and the status information of each power supply unit during operation.

[0039] The power configuration parameters include the number of power supply units. Users can configure this parameter through the power supply unit number configuration function. For example, the power configuration parameters may include the required number of power supply units to maintain normal server operation and the number of redundant power supply units. The management controller can provide a management interface, such as Redfish (interface name), through which users can configure the server's power parameters. The required number of power supply units refers to the number of power supply units needed for normal server operation. The number of redundant power supply units is equivalent to backup power supply units. A power supply unit is a device that provides power to the server, such as a PSU. Accordingly, the power configuration parameters include the number of PSUs N required for normal server operation and the number of redundant PSUs M. The operating current safe range refers to the safe range of the power supply unit's operating input current and the current values ​​and response times corresponding to various protection mechanisms, such as OPP (Over Power Protection), OCP (Over Current Protection), and OCW (Over-Current Warning). The operating current safe range is obtained through testing in scenarios where the power supply unit has no redundant power supply configuration. The response time includes alarm delay time and fault delay under 12V latching. An example operating safe range is shown in Table 1.

[0040] Table 1. Safe Operating Current Range of Power Supply Unit

[0041]

[0042] This embodiment refers to the function of supporting dynamic power consumption cap configuration. The management controller polls the presence and status of each power supply unit at a fixed frequency. Of course, it can also be triggered by an instruction to perform the operation of obtaining the status information of each power supply unit. The status information of each power supply unit includes at least the presence status of the power supply unit and whether there is an alarm status. The status information of each power supply unit can be obtained by obtaining the status word of each power supply unit. The status word can contain alarm information such as over-temperature, over-current, and over-voltage. The status information of each power supply unit reflects the current power supply status of the server. The power configuration parameter information is used to reflect the server's configured power supply status or the supply status that can ensure the stable and continuous operation of the server. Each time the management controller obtains the status information of each power supply unit, it compares the current status information of each power supply unit with the power configuration parameter information. If the current status information of each power supply unit meets the power configuration parameter information, that is, the two match, it indicates that the current state is normal. The power consumption configuration mode corresponding to the current power supply status is the first power consumption configuration mode. According to S202, the power consumption cap value matching the current power supply status is determined. For ease of description, it can be defined as the target power consumption cap value. If the current status information of each power supply unit does not meet the power configuration parameter information, i.e., the two do not match, the server's power supply may fail, i.e., it is in an abnormal state. In this case, the power consumption configuration mode corresponding to the current power supply state is the second power consumption configuration mode. S203 can be used to determine the power consumption capping value that matches the current power supply state, which can be defined as the transitional power consumption capping value. That is, by comparing the power configuration parameter information and the status information of each power supply unit during operation, it can be determined whether the current power consumption configuration mode is the first power consumption configuration mode or the second power consumption configuration mode. If the current power consumption configuration mode is the first power consumption configuration mode, and the power consumption configuration mode corresponding to the current power supply state is the second power consumption configuration mode, then switch to the second power consumption configuration mode, and use S203 to determine the power consumption capping value that matches the current power supply state to configure the power consumption limit policy. If the current power consumption configuration mode is the first power consumption configuration mode, and the power consumption configuration mode corresponding to the current power supply state is the first power consumption configuration mode, then there is no need to switch, and therefore no need to reconfigure the power consumption limit policy.

[0043] S202: In the first power consumption configuration mode, based on the overcurrent warning protection corresponding to the safe operating current range and without triggering the overcurrent protection corresponding to the safe operating current range, the target power consumption cap value is determined according to the rated power consumption of each power supply unit, and the power consumption limit strategy is updated according to the target power consumption cap value.

[0044] S203: In the second power consumption configuration mode, the transition power consumption cap value is determined based on the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, and the power consumption limit strategy is updated based on the transition power consumption cap value.

[0045] As shown in Table 1, the threshold value for triggering overcurrent warning protection within the safe operating current range but not triggering overcurrent protection within the safe operating current range is 120%, which means that the overcurrent warning protection will be triggered when the input current is 1.2 times the rated current. This also triggers the minimum value for overcurrent protection. To ensure that overcurrent protection is not triggered, a smaller safety value can be set, and the sum of the two values ​​can be adjusted to obtain the current maximum power consumption limit. The preset power consumption safety adjustment factor is a number less than 1, such as 0.9, 0.85, 0.88, or 0.8. By adjusting the sum of the rated power consumption of each normally powered unit using the preset power consumption safety adjustment factor as the maximum power consumption value, safe and reliable operation can be ensured in the event of a potential fault. The power consumption limit strategy is a strategy for setting power consumption limits. Whenever a new power consumption cap is generated, the corresponding power consumption cap value of the power consumption limit strategy will be updated. Taking Table 1 as an example, if the PSU has a rated power consumption of 1000W, and the current does not exceed 140%, the power consumption will be multiplied by 1.4, allowing it to last for approximately 11 seconds. If the current exceeds 150%, it may only last for 5ms before power failure, posing a risk of failure. If the PSU has redundancy, using a maximum power limit of 1.2 * 0.97 will not trigger OCW and will maximize performance. If the PSU fails and loses redundancy, a power limit of 0.8 * 0.97 will be used for safety priority control. The total system power consumption must not exceed 1.4 times the PSU power consumption to ensure reliable control.

[0046] In the technical solution provided in this embodiment, configuring the number of power supply units not only supports dynamic configuration of the power consumption cap but also ensures the stability of the server's power supply. By comparing the currently available power supply units and their status with the power parameter configuration information, it is determined whether the server is in a normal or abnormal state, and then the power consumption configuration mode to be switched is determined. The power consumption limit matching the current power supply state is adjusted based on the power supply status, achieving dynamic adjustment of the server's power consumption limit according to the power supply state. This not only effectively avoids unexpected server downtime due to power failures, ensuring business continuity, but also ensures that hot replacement of power supply units does not affect business continuity, improving server high availability. Furthermore, it maximizes power supply performance, reduces server power consumption, and maximizes resource utilization, which is beneficial for deployment in edge scenarios with limited power supply environments.

[0047] Before performing the dynamic power capping adjustment process described in the above embodiments, the dynamic capping configuration function can be configured. In this embodiment, an enable and disable option for the dynamic capping configuration function (powercapping) can be added to the management controller's visualization page. Users can manually enable the dynamic capping configuration function, or set a default option, which is disabled. When the dynamic capping configuration function is disabled, the logic device can determine whether frequency reduction and throttling are needed based on the minimum redundancy. Figure 3As shown, when the dynamic capping value configuration function is enabled, a power consumption limiting strategy is set. First, it is necessary to determine whether the dynamic capping value configuration function is supported. The determination process is as follows:

[0048] The system obtains the number of power supply units in place and their rated power consumption. If the server's maximum system power consumption is less than or equal to the power demand under overload protection corresponding to the safe operating current range, the power configuration parameter information is sent to the logic device, and the dynamic power cap configuration function is activated. If the number of power supply units in place does not meet the power configuration parameter information, a redundancy loss alarm is triggered. If the server's maximum system power consumption is greater than the power demand, a prompt message indicating that dynamic power cap configuration is not supported is generated.

[0049] The maximum system power consumption of the server is the estimated maximum power consumption required by the server, which can be estimated using server model information and asset information sent by the BIOS. For example, the power consumption hardware and its corresponding configuration information can be determined based on the server asset information. Power consumption hardware includes CPU / GPU / memory / hard drive and expansion cards, and configuration information includes their respective operating parameters such as rated power consumption. The rated power consumption of each power consumption hardware is determined from its configuration information. Based on the server model information (i.e., form factor, such as 1U or 4U) and thermal layout parameters (number of fans), the server's base power consumption is determined, which includes the losses of the motherboard, system fans, various cables, and PCB. Based on each power consumption hardware and its corresponding rated power consumption, the server's base power consumption, and a margin factor, the maximum system power consumption is determined. For example, the maximum system power consumption = (power consumption hardware 1 × rated power consumption 1 + ... power consumption hardware n × rated power consumption n + server base power consumption) × margin factor, where the margin factor is used to cover the peak value within the peak value, and can be taken as 1.1~1.2.

[0050] The process for determining the power consumption demand can be as follows: Based on the number of power supply units required and their rated power consumption, determine the total power consumption of the power supply units; if the fault delay time of the overload protection corresponding to the safe operating current range is less than a pre-set first preset delay threshold, and the maximum fault delay time of the overcurrent protection is greater than a pre-set second preset delay threshold, and the second preset delay threshold is significantly greater than the first preset delay threshold, then determine the maximum current triggered by the overcurrent protection corresponding to the maximum fault delay time, and determine the power consumption critical adjustment factor based on the maximum current and the rated current; determine the power consumption demand based on the total power consumption of the power supply units and the power consumption critical adjustment factor. As shown in Table 1, the overload protection response time is 5ms, which is too short and difficult to maintain. The response time of the first-level overcurrent protection is 11.1s. If this happens to be the maximum power consumption state, the remaining PSUs may immediately trigger overcurrent protection due to excessive current, causing the logic devices to fail to throttle. This alarm can inform users of the risk in advance, allowing them to take preventative measures. Therefore, the maximum current triggered by overcurrent protection corresponding to the maximum fault delay time is 140%. Correspondingly, if the maximum system power consumption is greater than N * 1.4 times the rated power consumption of a single power supply unit, a log alarm will be generated. For example, the visualization page may display "Dynamic power capping cannot be supported (The current system configuration cannot be supported due to exceeding the system powercap)," pointing to 4b0000 to disable dynamic power capping. If the current maximum system power consumption is less than or equal to N * 1.4 times the rated power consumption of a single power supply unit, it proves that the dynamic power capping configuration function is supported. The power configuration parameter information will be sent to the logic device. The logic device will obtain the status of each power supply unit based on the received power configuration information. If the number of power supply units in place is less than the required number of power supply units, and there are power supply unit alarms, it indicates that there is a possibility that the actual power consumption exceeds the power capping value. The logic device will generate an alarm and trigger an operation to reduce the operating frequency of the relevant power consumption devices. The management controller will record this frequency reduction and throttling operation in the log for subsequent power management. After the logic device performs frequency reduction and throttling, the power consumption will decrease. At this point, the power consumption has been reduced and the alarm can be cleared. If the alarm pin of the power supply unit can be pulled high, the logic device will continue to perform frequency reduction and throttling until it receives a throttling interrupt signal from the management controller, thereby avoiding excessive loss of server performance.

[0051] The above embodiments do not limit how the target power consumption value is determined. Based on the above embodiments, the present invention also provides an exemplary implementation method, which may include the following:

[0052] The power consumption information is determined based on the status information of each power supply unit; the maximum warning current for triggering overcurrent warning is determined based on the safe range of the operating current, and the maximum power consumption adjustment factor is determined based on the maximum warning current and the rated current; the target power consumption cap is determined based on the power consumption information, the maximum power consumption adjustment factor and the first preset safety margin value.

[0053] The first preset safety margin value is used to establish a safe, stable, and reliable buffer between hardware limits and software control targets to cope with control delays, measurement errors, and instantaneous peaks, ensuring smooth and effective capping operations without triggering hardware protection mechanisms that could cause system crashes. This value can be, for example, 0.97, 0.95, or 0.98, and those skilled in the art can choose according to the actual situation. The power consumption provision value refers to the total rated power consumption that the power supply units can provide. For example, based on the status information of each power supply unit, the target power supply unit in normal working condition among the power supply units in the present state can be determined; the power consumption provision value is determined based on the rated power consumption value of each target power supply unit and the total number of target power supply units. Accordingly, the target power consumption capping value = power consumption provision information × maximum power consumption adjustment factor × first preset safety margin. Taking Table 1 as an example, the first preset safety margin is 0.97, and the target power consumption capping value = 1.2 × total rated power consumption that the power supply units can provide × 0.97.

[0054] The above embodiments do not limit how to determine the peak value of transitional power consumption. Based on the above embodiments, the present invention also provides an exemplary implementation method, which may include the following:

[0055] For scenarios that support user-defined power cap configuration via a user interface, the current power supply value (i.e., the new power supply value) is determined based on the number of power supply units in place and operating normally, and their corresponding rated power. A preset power safety adjustment factor is determined based on the operating current safety range, and the power cap adjustment value is determined based on the new power supply value and the preset power safety adjustment factor. When a user-defined power adjustment request is received, if the user-inputted power cap value is less than or equal to the power cap adjustment value, the user-inputted power cap value is used as the new power cap value; if the user-inputted power cap value is greater than the power cap adjustment value, the power cap adjustment value is used as the transitional power cap value.

[0056] For scenarios that do not support the configuration of user-defined power consumption caps or have not received user-defined power consumption caps, the new power consumption provision value can be determined based on the number of power supply units in the in-situ and normal operating state and their corresponding rated power; the power consumption safety adjustment factor can be determined based on the operating current safety range; and the transitional power consumption cap value can be determined based on the new power consumption provision value, the preset power consumption safety adjustment factor, and the second preset safety margin value.

[0057] The second preset safety margin value is used to establish a safe, stable, and reliable buffer between hardware limits and software control objectives to cope with control delays, measurement errors, and instantaneous peak values, ensuring smooth and effective capping operations without triggering hardware protection mechanisms that could cause system crashes. This value can be, for example, 0.97, 0.95, or 0.98, and those skilled in the art can choose according to the actual situation. The second preset safety margin value can be the same as or different from the first preset safety margin value; this does not affect the implementation of the invention. The new power consumption provision value is the sum of the currently provided power consumption. For example, if there are m power supply units in the in-situ and normal operating state (power supply unit 1, power supply unit 2, ..., power supply unit m), the new power consumption provision value = power supply unit 1 × rated power 1 + power supply unit 2 × rated power 2 + ... + power supply unit m × rated power m. The power consumption cap adjustment value = power consumption safety adjustment factor × new power consumption provided value; the transition power consumption cap value = new power consumption provided value × power consumption safety adjustment factor × second preset safety margin value, with the second preset safety margin being 0.97 and the power consumption safety adjustment factor being 0.8; the transition power consumption cap value = 0.8 × the sum of the rated power consumption that the power supply unit can provide × 0.97.

[0058] As can be seen from the above, this embodiment supports the calculation of power consumption cap values ​​under different scenarios, which not only enables dynamic adjustment of server power consumption limits according to power supply status, but also maximizes power supply performance.

[0059] Based on the above embodiments, after the logic device triggers a frequency reduction and throttling operation, this embodiment also provides a power callback implementation method, such as... Figure 4 As shown, it may include the following:

[0060] When a throttling notification signal is received, an alarm signal is generated indicating that the status information of each power supply unit does not match the power configuration parameter information, and the periodic polling operation is stopped; at the same time, the status information of each power supply unit is continuously acquired until the status information of the power supply units at adjacent times meets the same preset similar conditions.

[0061] The management controller acquires the status information of each power supply unit according to a preset status query frequency. When it receives a throttling notification signal, indicating that at least one target power supply unit is unable to supply power normally after the power limit policy updated according to the target power cap value takes effect, an alarm operation is initiated. Simultaneously, the current periodic polling of the power supply units is stopped, and multiple attempts are immediately made to read the status of the power supply units at the current moment. After the status of the power supply units stabilizes, i.e., the status information of the power supply units at adjacent moments meets the preset similarity conditions, the acquired power supply unit status information after stabilization is used to recalculate a new power cap value, i.e., a transitional power cap value. When the power limit policy updated according to the transitional power cap value takes effect, a throttling interrupt notification signal is sent. Alternatively, a few seconds, such as 3 seconds, can be waited before sending the interrupt notification signal, and the periodic polling operation is resumed. Continuous monitoring is performed to determine if the conditions for lifting the frequency reduction and throttling operation are met: If, based on the latest power supply unit status information, the number of power supply units in normal working condition is greater than or equal to the required number of power supply units, an alarm cancellation signal is generated, and the server's power consumption cap is restored to the target power consumption cap. The latest power limit policy is then updated based on the target power consumption cap. For example, if the transitional power consumption cap is 0.8 times the total power consumption that a currently normal PSU can provide, or a user-preset value, when periodic polling detects that a power supply unit has been replaced (i.e., the number of currently normal power supply units is greater than or equal to the power demand), the alarm is lifted, and the power consumption policy that sets the total power consumption that a currently normal power supply unit can provide multiplied by 0.8 to the power limit policy for overall power consumption capping is deleted. Power consumption capping is then adjusted using the total power consumption that a currently normal power supply unit can provide multiplied by 1.2 multiplied by 0.97.

[0062] If the management controller restarts, during the restart process, it reads the power consumption limit policy and the frequency reduction and throttling data of the logic devices stored before the power failure; based on the frequency reduction and throttling data, the power consumption limit policy and the latest power supply unit status information, it adjusts the power consumption cap value of the power consumption limit policy in the manner described in the above embodiments.

[0063] As shown above, this embodiment directly monitors the presence status and alarm signals of the power supply unit through logic devices, and directly triggers component frequency reduction and current throttling through hardware links, achieving a fast hardware-level response with a response time ≤100ns. The logic devices jointly determine the power supply unit status based on power configuration parameters and power supply unit status information, initiating frequency reduction and current throttling operations when a power failure occurs. Simultaneously, a current limiting signal is sent to the management controller. The management controller judges the power supply unit status after the system stabilizes, makes a decision based on the current power supply unit status and user settings, and regulates power consumption through a power limit strategy. After the strategy takes effect, the frequency reduction and current throttling operation is lifted, avoiding excessive loss of server performance.

[0064] This invention also provides a flowchart of another server power management method. This embodiment is applicable to any logic device in a server, such as a CPLD. Figure 5 As shown, it may include the following:

[0065] S501: When power configuration parameter information is received, the power supply unit is polled according to the power configuration parameter information to obtain the status information of each power supply unit.

[0066] The power configuration parameters include the number of power supply units required to maintain normal server operation and the number of redundant power supply units.

[0067] S502: If the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit status information includes an alarm signal, then a notification message is sent to indicate that the power consumption configuration mode corresponding to the power supply state is the second power consumption configuration mode.

[0068] In this embodiment, after the logic device obtains the status information of each power supply unit at the current moment through S501, it determines that the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit has sent an alarm signal, indicating that the server power supply may fail. That is, the power consumption configuration mode matching the current power supply state is the second power consumption configuration mode. Then, it sends a notification message to the management controller indicating that the power consumption configuration mode matching the current power supply state is the second power consumption configuration mode. When the management controller receives the notification message, if the current power consumption configuration mode is the first power consumption configuration mode, the management controller determines the target power consumption cap based on the rated power consumption of each power supply unit and updates the power consumption limit strategy based on the target power consumption cap, based on the overcurrent warning protection corresponding to the safe operating current range and without triggering the overcurrent protection corresponding to the safe operating current range. Then, it switches to the second power consumption configuration mode. In this mode, the transition power consumption cap is determined based on the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, and the power consumption limit strategy is updated based on the transition power consumption cap. The preset power consumption safety adjustment factor is less than 1.

[0069] For example, the logic device can also generate a frequency reduction and throttling trigger signal and send the frequency reduction and throttling trigger signal directly to the power consumption hardware whose operating frequency needs to be reduced; generate a throttling processing notification signal and send the throttling processing notification signal; the throttling processing notification signal indicates that the status information of each power supply unit does not match the power configuration parameter information.

[0070] For example, the logic device acquires the status information of each power supply unit in real time. If a throttling interruption notification signal is received during the acquisition of the status information of each power supply unit, the acquisition of the status information of each power supply unit is stopped; and the operating frequency of the corresponding power consumption hardware is restored according to the throttling interruption notification signal.

[0071] Finally, to enable those skilled in the art to more clearly understand the technical solution of the present invention, the present invention also provides an exemplary implementation. This embodiment takes a BMC as the management controller, a CPLD as the logic device, and a PSU as the power supply unit as an example. The process of dynamically regulating the power consumption of the server may include the following:

[0072] A1: BMC provides the power configuration parameters for the server via the Redfish interface: the number of PSUs N and the number of redundant PSUs M required for normal server operation.

[0073] A2: During BMC operation, the number of PSUs in place and their rated power consumption are read in real time. If the number of PSUs is less than N+M, a redundancy loss alarm is triggered. After the server is powered on, the current maximum system power consumption is determined based on the machine type and asset information. If there is no PSU redundancy loss alarm, it is determined whether the current maximum system power consumption is greater than N * 1.4 times the rated power consumption of a single PSU. If it is greater, proceed to A3; otherwise, proceed to A4.

[0074] A3: BMC will log alerts.

[0075] A4: The BMC sends the power configuration parameter information to the CPLD.

[0076] A5: The BMC periodically polls the PSU status, calculates the total power consumption that a PSU in a normal state can provide based on the status of each PSU, multiplies this value by 1.2 * 0.97 to obtain the target power consumption cap, and sets it to the power limit policy of the power execution manager.

[0077] In this step, if the PSU is normal, the risk of power consumption exceeding the limit should be avoided. However, if a PSU fails or is accidentally removed during the polling interval, an overcurrent alarm may occur because the remaining PSUs cannot support the current power consumption. A severe overcurrent alarm will cause the PSU to shut down its output, which will then cause the system to shut down. At this time, the CPLD needs to work together to adjust the system power consumption in the shortest possible time. That is, the processor is triggered to reduce its frequency through the PROHOT pin connected to the CPLD and the processor to prevent the remaining PSUs from triggering overcurrent.

[0078] A6: The CPLD polls the PSU status in real time based on the power configuration parameters issued by the BMC. If it receives a throttling interruption notification from the BMC, it processes the notification with the highest priority. If there is no throttling interruption, it continues to acquire PSU status information. The CPLD determines whether to trigger frequency reduction and throttling by combining the PSU presence status (PRESENT#) and ALERT# alarm signals: if the number of currently present PSUs is less than or equal to the set required number of PSUs, and an ALERT# alarm is triggered for a PSU, then frequency reduction and throttling operation is triggered, and a throttling processing notification signal is sent to the BMC to inform the BMC that the CPLD has triggered frequency reduction and throttling operation.

[0079] A7: Upon receiving the throttling notification signal from the CPLD, the BMC initiates an alarm operation and simultaneously stops periodic polling of the PSU. It immediately attempts to read the current PSU status multiple times. After the PSU status stabilizes, based on the overall PSU status assessment results, it multiplies the total power consumption provided by a normally functioning PSU by 0.8 or 0.97, or a user-preset value, to determine the transitional power cap. This cap is then set again by the power management system for overall system power consumption. Once this power consumption strategy takes effect, a command is sent to the CPLD to cancel the current frequency reduction and throttling operation and resume periodic polling of the PSU status.

[0080] The user-preset value should not be greater than 0.8 times the total power consumption value that a PSU in normal condition can provide. If it is greater, use 0.8 times the total power consumption value that a PSU in normal condition can provide.

[0081] A8: When the BMC detects that the PSU has been replaced during the periodic polling of the PSU, it will clear the alarm and delete the power limiting policy set by the transition power cap value. It will continue to control the power cap by using the total power consumption value provided by the PSU in the current normal state * 1.2 * 0.97.

[0082] A9: If the BMC restarts, the relevant power management control has been written to the CPU register, and the CPU continues to control it. If the system loses power, the CPLD frequency reduction and throttling data and power limit strategy will be stored before the power loss. After the BMC is powered on and started, it will dynamically control it according to the real-time status.

[0083] Of course, the management controller BMC or CPLD can also collect the CPU and memory usage of the server system, train it using any large language model, and issue PSU status warnings in advance before business peaks.

[0084] As shown above, this embodiment achieves dynamic power consumption management of the server through dynamic redundancy management and dual security mechanisms. The BMC provides two working modes: stable state and abnormal state. Based on the PSU's presence and status word, combined with the N+M power configuration, if the currently normal PSU meets the N+M requirement, it operates in the stable state, where the power consumption limit is 1.2 * the total rated power available to the PSU * 0.97. If a CPLD frequency reduction / throttling signal is triggered, or if a PSU anomaly is detected, it operates in the abnormal state, where the power consumption limit is 0.8 * the total rated power available to the PSU * 0.97 or a user-defined value that meets the requirements. The CPLD uses power configuration parameters, PSU presence status, and alarm information to jointly determine whether to initiate frequency reduction / throttling in the event of a power failure. Simultaneously, it sends a current limiting signal to the BMC. The BMC determines the PSU status after system stabilization, makes a decision based on the current PSU status and user settings, and adjusts the power consumption using a power limit strategy. After the strategy takes effect, the frequency reduction / throttling operation is lifted, preventing excessive loss of server performance. The system directly monitors the presence status and alarm signals of the power supply unit through logic devices, and directly triggers the component to reduce frequency and throttle through hardware links, achieving a fast hardware-level response with a response time of ≤100ns.

[0085] It should be noted that there is no strict order of execution between the steps in this invention. As long as they conform to the logical order, these steps can be executed simultaneously or in a certain preset order. Figures 2-5 This is just an illustrative example and does not mean that this is the only possible execution order.

[0086] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0087] This invention also provides a corresponding device for server power management, further enhancing the practicality of the method. The device can be described from both a functional module perspective and a hardware perspective. The following describes the server power management device provided by this invention, which is used to implement the server power management method provided by this invention. In this embodiment, the server power management device may include or be divided into one or more program modules. These program modules are stored in a storage medium and executed by one or more processors to complete the server power management method disclosed in Embodiment 1. The program module referred to in this embodiment is a series of computer program instruction segments capable of performing a specific function, which is more suitable than the program itself for describing the execution process of the server power management device in the storage medium. The following description will specifically introduce the functions of each program module in this embodiment. The server power management device described below corresponds to the server power management method described above.

[0088] From the perspective of functional modules, see Figure 6 , Figure 6 This is a structural diagram of a server power management device provided in this embodiment under a specific implementation. The device is applied to a management controller and may include:

[0089] The data acquisition module 601 is used to acquire power supply configuration parameter information and the safe operating current range of the power supply unit.

[0090] The dynamic control module 602 is used to switch power consumption configuration modes based on power configuration parameter information and the status information of each power supply unit during operation. In the first power consumption configuration mode, based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range, the target power consumption cap value is determined according to the rated power consumption of each power supply unit, and the power consumption limit strategy is updated according to the target power consumption cap value. In the second power consumption configuration mode, based on the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, the transition power consumption cap value is determined, and the power consumption limit strategy is updated according to the transition power consumption cap value. The preset power consumption safety adjustment factor is less than 1.

[0091] For example, in some embodiments of this embodiment, the above-mentioned device may further include a function support identification module for obtaining the number of power supply units in place and their rated power consumption; if the maximum system power consumption of the server is less than or equal to the power supply demand power consumption under overload protection corresponding to the safe range of the trigger operating current, the power configuration parameter information is sent to the logic device, and the dynamic power consumption capping value configuration function is started.

[0092] As an exemplary implementation of the above embodiments, the above-mentioned function support identification module can be further used to: determine the total power consumption of the power supply unit based on the number of power supply units required and the rated power consumption of the power supply unit; if the fault delay time of the overload protection corresponding to the safe range of the triggering operating current is less than the first preset delay threshold, and the maximum fault delay time of the overcurrent protection is greater than the second preset delay threshold, then determine the maximum current triggered by the overcurrent protection corresponding to the maximum fault delay time, and determine the power consumption critical adjustment factor based on the maximum current and the rated current; determine the power supply demand power consumption based on the total power consumption of the power supply unit and the power consumption critical adjustment factor.

[0093] As another exemplary implementation of the above embodiments, the above-mentioned functional support identification module can be further used to: determine each power consumption hardware and its corresponding configuration information based on server asset information, and determine the rated power consumption of the corresponding power consumption hardware from the configuration information of each power consumption hardware; determine the basic power consumption of the server based on server model information and heat dissipation layout parameters; and determine the maximum system power consumption based on each power consumption hardware and its corresponding rated power consumption, the basic power consumption of the server, and the margin coefficient.

[0094] As another exemplary implementation of the above embodiments, the above-mentioned function support identification module can also be further used to: if the number of power supply units in place does not meet the power configuration parameter information, then perform a redundancy loss alarm; if the maximum system power consumption of the server is greater than the power supply demand power consumption, then generate a prompt message that dynamic power consumption cap configuration is not supported.

[0095] For example, in some other embodiments of this embodiment, the dynamic control module 602 can also be used to: determine power supply information based on the status information of each power supply unit; determine the maximum warning current that triggers the overcurrent warning based on the safe range of the operating current, and determine the maximum power supply adjustment factor based on the maximum warning current and the rated current; and determine the target power supply capping value based on the power supply information, the maximum power supply adjustment factor and the first preset safety margin value.

[0096] As an exemplary implementation of the above embodiments, the dynamic control module 602 can be further used to: determine the target power supply unit in normal working state among the power supply units in the in-situ state according to the status information of each power supply unit; and determine the power supply value according to the rated power consumption value of each target power supply unit and the total number of target power supply units.

[0097] For example, in some other embodiments of this embodiment, the dynamic control module 602 can also be used to: determine a new power consumption provision value based on the number of power supply units in the in-situ and normal working state and their corresponding rated power; determine a preset power consumption safety adjustment factor based on the safe range of the working current, and determine a power consumption cap adjustment value based on the new power consumption provision value and the preset power consumption safety adjustment factor; when a user-defined power consumption adjustment request is received, if the user-inputted power consumption cap custom value is less than or equal to the power consumption cap adjustment value, then the power consumption cap custom value is used as the new power consumption cap value; if the power consumption cap custom value is greater than the power consumption cap adjustment value, then the power consumption cap adjustment value is used as the transitional power consumption cap value.

[0098] For example, in some other embodiments of this embodiment, the dynamic control module 602 can also be used to: determine a new power consumption provision value based on the number of power supply units in the in-situ and normal working state and their corresponding rated power; determine a power consumption safety adjustment factor based on the safe range of the working current; and determine a transition power consumption cap value based on the new power consumption provision value, the preset power consumption safety adjustment factor, and the second preset safety margin value.

[0099] For example, in some other embodiments of this embodiment, the dynamic control module 602 can also be used to: acquire the status information of each power supply unit according to a preset status query frequency; when a throttling notification signal is received, generate an alarm signal indicating that the status information of each power supply unit does not match the power configuration parameter information, and stop the periodic polling operation; the throttling notification signal indicates that after the power limit strategy updated according to the target power consumption cap value takes effect, at least one target power supply unit cannot supply power normally; at the same time, continuously acquire the status information of each power supply unit until the status information of the power supply units at adjacent times meets the same preset similar conditions.

[0100] As an exemplary implementation of the above embodiments, the dynamic control module 602 can be further used to: send a throttling interruption notification signal and resume periodic polling operation after the power limit policy updated according to the transition power cap value takes effect; if it is determined according to the latest power supply unit status information that the number of power supply units in normal working condition among the in-situ power supply units is greater than or equal to the number of power supply units required, then generate an alarm cancellation signal, restore the server's power cap value to the target power cap value, and update the latest power limit policy according to the target power cap value.

[0101] From the perspective of functional modules, see Figure 7 , Figure 7 This is a structural diagram of the server power management device provided in this embodiment under another specific implementation. The device is applied to a logic device and may include:

[0102] The polling module 701 is used to poll each power supply unit according to the power configuration parameter information when it receives the power configuration parameter information, so as to obtain the status information of each power supply unit.

[0103] The anomaly identification module 702 is used to send a notification message indicating that the power consumption configuration mode corresponding to the power supply status is the second power consumption configuration mode if the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and the status information of at least one power supply unit includes an alarm signal. This is to determine the transitional power consumption cap value based on the rated power consumption of each normally powered power supply unit and a preset power consumption safety adjustment factor, and to update the power consumption limit strategy based on the transitional power consumption cap value. The preset power consumption safety adjustment factor is less than 1. The power consumption configuration mode corresponds to the first power consumption configuration mode, which is based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range. The target power consumption cap value is determined based on the rated power consumption of each power supply unit, and the power consumption limit strategy is updated based on the target power consumption cap value.

[0104] For example, in some embodiments of this embodiment, the above-mentioned anomaly identification module 702 may be further used to: generate a frequency reduction and throttling trigger signal, and send the frequency reduction and throttling trigger signal directly to the power consumption hardware whose operating frequency is to be reduced; generate a throttling processing notification signal, and send the throttling processing notification signal; the throttling processing notification signal indicates that the status information of each power supply unit does not match the power configuration parameter information.

[0105] For example, in some other embodiments of this embodiment, the above-mentioned device may further include an interrupt handling module, which is used to: when a throttling interrupt notification signal is received during the acquisition of status information of each power supply unit, stop acquiring status information of each power supply unit; and restore the operating frequency of the corresponding power consumption hardware according to the throttling interrupt notification signal.

[0106] For a description of the features in the embodiment of the server power management device, please refer to the relevant description of the embodiment of the server power management method, which will not be repeated here.

[0107] The server power management device mentioned above is described from the perspective of functional modules. Furthermore, the present invention also provides an electronic device, which is described from the perspective of hardware. Figure 8 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present invention. The electronic device includes a memory 801 and a processor 802. The memory 801 stores a computer program, and the processor 802 is configured to run the computer program to perform the steps in any of the above-described server power management method embodiments.

[0108] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above-described server power management method embodiments when running.

[0109] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0110] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described server power management method embodiments.

[0111] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps in any of the above-described server power management method embodiments.

[0112] Finally, the present invention also provides a server, please refer to [link to relevant documentation]. Figure 9 The server may include a logic device 901, a management controller 902, and a target processor 903, and the data interaction between the devices is as follows: Figure 10As shown. In this embodiment, the target processor 903 can be, for example, a central processing unit (CPU) or other processor that can directly reduce its operating frequency. The logic device 901 is connected to the management controller 902 via a first bus, such as I2C or SMbus (System Management Bus). The management controller 902 sends power limit policies, power configuration information, and throttling interrupt signals to the logic device 901 via this bus, and receives throttling processing notification signals from the logic device 901. When the management controller 902 executes a computer program, the logic device 901 executes the method steps in the server power management method described in any of the above embodiments. The logic device 901 includes at least a power supply status monitoring interface and a frequency reduction interaction interface. It is connected to each power supply unit via the power supply status monitoring interface and to the target processor 903 via the frequency reduction interaction interface, and sends a frequency reduction throttling trigger signal to the target processor 903 via the frequency reduction interaction interface. Each power supply unit connects at least the signal lines corresponding to PS_PRESENT# (present signal), PSU_ALERT_N (alarm signal), VIN Good (power supply voltage signal), and PSU_PWROK (power supply unit voltage signal) to the power supply status monitoring interface. The logic device 901 connects to the target processor 903, such as the PRC HOT input pin and MEM HOT input pin of the central processing unit, via a frequency reduction interaction interface. When the logic device 901 receives a frequency reduction throttling signal, it triggers the target processor 903 to perform frequency reduction throttling operations. If, after detecting a power consumption cap setting, at least one target power supply unit fails to supply power normally (faulty or disconnected), the logic device sends a frequency reduction message to the target processor 903 via the frequency reduction interaction interface. When the management controller 902 executes a computer program to implement the server power consumption management method as described in any of the above embodiments, the management controller 902 performs the method steps. The management controller 902 includes a power execution manager, which is connected to the target processor 903. The power execution manager updates the power limit policy according to the target power cap or the transition power cap and sends it to the target processor 903.

[0113] Furthermore, the logic device 901 also includes at least one direct-connect frequency reduction interface, which connects to power consumption hardware and sends frequency reduction and throttling trigger signals to the corresponding power consumption hardware. For example, the logic device 901 has reserved lines to connect to the B30 of a PCIe (Peripheral Component Interconnect Express) card with power brake function and the A70 of an OCP (Open Compute Project) 3.0 card. When it receives a frequency reduction and throttling signal from the CPLD, it can trigger the frequency reduction and throttling operation. Furthermore, the management controller 902 includes a management interface and a visualization page; the visualization page adds an option to enable / disable the dynamic capping value configuration function. When a parameter configuration request carrying authorized user information is received through the management interface, the required number of power supply units and the number of redundant power supply units to maintain the normal operation of the server are obtained from the parameter configuration request; power configuration parameter information is generated based on the required number of power supply units and the number of redundant power supply units.

[0114] The foregoing has provided a detailed description of a server and its power management method, electronic device, non-volatile storage medium, and computer program product provided by the present invention. The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Whether the units and algorithm steps of the various examples described in the disclosed embodiments are executed in electronic hardware or computer software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, and such implementations should not be considered beyond the scope of the present invention. Several improvements and modifications can be made to the present invention without departing from the principles of the invention, and these improvements and modifications also fall within the protection scope of the present invention.

Claims

1. A server power consumption management method, characterized in that, Applied to management controllers, including: Obtain power supply configuration parameter information and the safe operating current range of the power supply unit; Based on the power configuration parameter information and the status information of each power supply unit during operation, switch the power consumption configuration mode; In the first power consumption configuration mode, based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range, a target power consumption cap is determined according to the rated power consumption of each power supply unit, and the power consumption limit strategy is updated according to the target power consumption cap. In the second power consumption configuration mode, based on the rated power consumption of each normally powered power supply unit and a preset power consumption safety adjustment factor, a transition power consumption cap is determined, and the power consumption limit strategy is updated according to the transition power consumption cap, wherein the preset power consumption safety adjustment factor is less than 1. The process of switching power consumption configuration modes based on the power configuration parameter information and the status information of each power supply unit during operation includes: If the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit status information includes an alarm signal, then the power consumption configuration mode is in an abnormal state, and the power consumption configuration mode is switched to the second power consumption configuration mode; otherwise, the power consumption configuration mode is in a normal state, and the power consumption configuration mode is switched to the first power consumption configuration mode.

2. The server power consumption management method according to claim 1, characterized in that, After obtaining the power supply configuration parameters and the safe operating current range of the power supply unit, and before determining the target power consumption cap based on the rated power consumption of each power supply unit, the following steps are also included: Obtain the number of power supply units in place and their rated power consumption; If the server's maximum system power consumption is less than or equal to the power demand under overload protection corresponding to the operating current safety range, the power configuration parameter information is sent to the logic device, and the dynamic power cap configuration function is activated.

3. The server power consumption management method according to claim 2, characterized in that, The power configuration parameter information includes the required number of power supply units and the number of redundant power supply units to maintain normal server operation. After obtaining the number of power supply units in place and their rated power consumption, it also includes: The total power consumption of the power supply units is determined based on the number of power supply units required and the rated power consumption of the power supply units. If the fault delay time that triggers the overload protection corresponding to the safe operating current range is less than the first preset delay threshold, and the maximum fault delay time that triggers the overcurrent protection is greater than the second preset delay threshold, then the maximum current that triggers the overcurrent protection corresponding to the maximum fault delay time is determined, and the power consumption critical adjustment factor is determined based on the maximum current and the rated current. The power supply demand is determined based on the total power consumption of the power supply unit and the critical power consumption adjustment factor.

4. The server power consumption management method according to claim 2, characterized in that, After obtaining the number of power supply units in place and their rated power consumption, the following is also included: Based on the server asset information, determine each power-consuming hardware and its corresponding configuration information, and determine the rated power consumption of the corresponding power-consuming hardware from the configuration information of each power-consuming hardware. Determine the server's base power consumption based on the server model information and thermal layout parameters; The maximum system power consumption is determined based on each power-consuming hardware component and its corresponding rated power consumption, the server's base power consumption, and the margin factor.

5. The server power consumption management method according to claim 2, characterized in that, After obtaining the number of power supply units in place and their rated power consumption, the following is also included: If the number of power supply units in place does not meet the power configuration parameter information, a redundancy loss alarm will be triggered. If the server's maximum system power consumption exceeds the power supply requirement, a message indicating that dynamic power cap configuration is not supported will be generated.

6. The server power consumption management method according to any one of claims 1 to 5, characterized in that, Based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range, the target power consumption cap is determined according to the rated power consumption of each power supply unit, including: The power consumption information is determined based on the status information of each power supply unit. The maximum warning current for triggering overcurrent warning is determined based on the operating current safety range, and the maximum power consumption adjustment factor is determined based on the maximum warning current and the rated current. The target power consumption cap is determined based on the power consumption information, the maximum power consumption adjustment factor, and the first preset safety margin value.

7. The server power consumption management method according to claim 6, characterized in that, The power consumption information is determined based on the status information of each power supply unit, including: Based on the status information of each power supply unit, determine the target power supply unit that is in normal working condition among the power supply units in the in-situ status; The power supply value is determined based on the rated power consumption of each target power supply unit and the total number of target power supply units.

8. The server power consumption management method according to any one of claims 1 to 5, characterized in that, Based on the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, the transition power consumption cap value is determined, including: The new power consumption provision value is determined based on the number of power supply units that are in place and operating normally, and their corresponding rated power. A preset power consumption safety adjustment factor is determined based on the operating current safety range, and a power consumption cap adjustment value is determined based on the new power consumption provision value and the preset power consumption safety adjustment factor. When a user-defined power consumption adjustment request is received, if the user-inputted power consumption cap value is less than or equal to the power consumption cap adjustment value, then the user-defined power consumption cap value will be used as the new power consumption cap value. If the custom power consumption cap value is greater than the power consumption cap adjustment value, then the power consumption cap adjustment value will be used as the transitional power consumption cap value.

9. The server power consumption management method according to any one of claims 1 to 5, characterized in that, Based on the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, the transition power consumption cap value is determined, including: The new power consumption provision value is determined based on the number of power supply units that are in place and operating normally, and their corresponding rated power. A preset power consumption safety adjustment factor is determined based on the operating current safety range; The transitional power consumption cap is determined based on the new power consumption provision value, the preset power consumption safety adjustment factor, and the second preset safety margin value.

10. The server power consumption management method according to any one of claims 1 to 5, characterized in that, The power configuration parameters include the required number of power supply units and the number of redundant power supply units to maintain normal server operation. The status information of each power supply unit is obtained according to a preset status query frequency, based on the rated power consumption of each normally powered power supply unit and a preset power consumption safety adjustment factor, including: When a throttling notification signal is received, an alarm signal is generated indicating that the status information of each power supply unit does not match the power configuration parameter information, and the periodic polling operation is stopped; the throttling notification signal indicates that after the power limit policy updated according to the target power cap value takes effect, at least one target power supply unit cannot supply power normally. At the same time, the status information of each power supply unit is continuously acquired until the status information of the power supply units at adjacent times meets the same preset similar conditions.

11. The server power consumption management method according to claim 10, characterized in that, After updating the power limit strategy based on the aforementioned transitional power cap value, the following is also included: Once the power limit policy updated according to the aforementioned transitional power cap value takes effect, a throttling interruption notification signal is sent, and the periodic polling operation is resumed. If, based on the latest power supply unit status information, it is determined that the number of power supply units in normal working condition among the available power supply units is greater than or equal to the required number of power supply units, then an alarm cancellation signal is generated, and the server's power consumption cap is restored to the target power consumption cap. The latest power consumption limit policy is then updated based on the target power consumption cap.

12. A server power consumption management method, characterized in that, Applied to logic devices, including: When power configuration parameter information is received, each power supply unit is polled according to the power configuration parameter information to obtain the status information of each power supply unit; If the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit status information includes an alarm signal, then a notification message is sent to indicate that the power consumption configuration mode corresponding to the power supply state is the second power consumption configuration mode, so as to determine the transition power consumption cap value according to the rated power consumption of each normally powered power supply unit and the preset power consumption safety adjustment factor, and update the power consumption limit strategy according to the transition power consumption cap value, wherein the preset power consumption safety adjustment factor is less than 1. Specifically, if the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit status information includes an alarm signal, then it is in an abnormal state; otherwise, it is in a normal state. The power consumption configuration mode corresponds to the first power consumption configuration mode, which is based on triggering the overcurrent warning protection corresponding to the safe operating current range without triggering the overcurrent protection corresponding to the safe operating current range. The target power consumption cap is determined according to the rated power consumption of each power supply unit, and the power consumption limit strategy is updated according to the target power consumption cap.

13. The server power consumption management method according to claim 12, characterized in that, If the number of power supply units in the in-situ state is less than or equal to the number of power supply units required, and at least one power supply unit status information includes, after the alarm signal, the following: Generate a frequency reduction and throttling trigger signal, and send the frequency reduction and throttling trigger signal directly to the power consumption hardware whose operating frequency needs to be reduced; A throttling notification signal is generated and sent; the throttling notification signal indicates that the status information of each power supply unit does not match the power configuration parameter information.

14. The server power consumption management method according to claim 12, characterized in that, Obtain the status information of each power supply unit, including: If a throttling interruption notification signal is received during the process of acquiring the status information of each power supply unit, the acquisition of the status information of each power supply unit shall be stopped. Based on the throttling interrupt notification signal, the operating frequency of the corresponding power-consuming hardware is restored.

15. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the server power management method as described in any one of claims 1 to 14 when executing the computer program.

16. A non-volatile storage medium, characterized in that, The non-volatile storage medium stores a computer program, which, when executed by a processor, implements the steps of the server power management method as described in any one of claims 1 to 14.

17. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the server power management method according to any one of claims 1 to 14.

18. A server, characterized in that, Includes logic devices, management controllers, and target processors; The logic device is connected to the management controller via a first bus, and is used to implement the steps of the server power management method as described in any one of claims 12 to 14 when executing a computer program; the logic device includes at least a power supply status monitoring interface and a frequency reduction interaction interface, is connected to each power supply unit through the power supply status monitoring interface, is connected to the target processor through the frequency reduction interaction interface, and sends a frequency reduction throttling trigger signal to the target processor through the frequency reduction interaction interface. The management controller is used to implement the steps of the server power management method as described in any one of claims 1 to 11 when executing a computer program; The management controller includes a power execution manager connected to the target processor. The power execution manager updates the power limit policy based on the target power cap or the transitional power cap and sends it to the target processor.

19. The server according to claim 18, characterized in that, The logic device also includes at least one direct-connect frequency reduction interface, which is connected to power consumption hardware and sends a frequency reduction and throttling trigger signal to the corresponding power consumption hardware through the direct-connect frequency reduction interface.

20. The server according to claim 18, characterized in that, The management controller also includes a management interface; When a parameter configuration request carrying authorized user information is received through the management interface, the required number of power supply units and the number of redundant power supply units to maintain the normal operation of the server are obtained from the parameter configuration request. Power configuration parameter information is generated based on the required number of power supply units and the number of redundant power supply units.

Images