Power consumption control method, power consumption control module and power consumption control system

Abstract

The invention discloses a power consumption control method, a power consumption control module and a power consumption control system. Relates to the technical field of computers. The method is applied to a power consumption control module and comprises the steps that power consumption monitoring data of a power consumption monitoring module on an SOC and a display card are obtained, and the power consumption monitoring data of the display card are obtained by the power consumption monitoring module through monitoring of a display card power consumption detection module; and predicting a load trend according to the power consumption monitoring data, and controlling the power consumption of the SOC and the video card according to a load trend prediction result. According to the scheme, the power consumption monitoring module and the display card power consumption detection module are utilized in the system, real-time monitoring of SOC power consumption and display card power consumption is achieved, the power consumption control module can achieve cooperative management of the SOC power consumption and the display card power consumption on the basis of predicting the load trend, a display card drive or a third-party application is not needed, and the cost is reduced. The power consumption control efficiency and the power consumption coordination of the two are improved, resource conflicts and power consumption waste are reduced, and the resource utilization rate and the system performance are improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a power consumption control method, a power consumption control module, and a power consumption control system. Background Technology

[0002] With the development of Artificial Intelligence Personal Computers (AIPCs), the local heterogeneous computing power of System-on-a-Chip (SoC) offers excellent low-latency processing capabilities. Simultaneously, the demand for high computing power and high power consumption from graphics cards using the Peripheral Component Interconnect Express (PCIE) standard has increased significantly. The manufacturers, power supply schemes, and rated power consumption of the expansion lines in PCIE slots vary. For example, some graphics cards have a rated power consumption of less than 75W, some have a rated power consumption of 75W (PCIE x16 slot) and no external power supply interface, while others have a rated power consumption greater than 75W (PCIE x16 slot) and an external power supply interface.

[0003] Currently, both SoCs and PCIe graphics cards employ independent power management mechanisms, managing system power consumption for local heterogeneous computing power and PCIe graphics card power consumption independently. During the execution of Artificial Intelligence (AI) models, the graphics card load increases instantaneously, and some graphics cards may be overclocked to briefly exceed their Thermal Design Power (TDP). After execution, the graphics card returns to standby. During this process, it's necessary to consider that different AIPCs may use graphics cards from different manufacturers or models, requiring targeted power management for the graphics cards. Resource conflicts may exist between the power management of SoCs and PCIe graphics cards. For example, when both the SoC and the graphics card are under high load, frequency throttling may be triggered, preventing either from performing at its full potential. The inability to effectively manage them collaboratively can also lead to power waste and performance bottlenecks. For instance, when one's load is low, the power saved cannot be used by the other; or, due to slot power limitations, the PCIe graphics card cannot fully utilize the resources released by the SoC. Furthermore, the real-time power consumption of PCIe graphics cards relies on graphics card drivers or third-party applications, making it difficult to achieve compatibility across different manufacturers, further complicating power management for both.

[0004] In summary, how to efficiently manage the power consumption of the SoC system and the graphics card without relying on graphics card drivers or third-party applications is an urgent problem to be solved. Summary of the Invention

[0005] This application provides a power consumption control method, a power consumption control module, and a power consumption control system to achieve efficient collaborative management of the power consumption of the SoC system and the graphics card.

[0006] In a first aspect, embodiments of this application provide a power consumption control method applied to a power consumption control module in a power consumption control system. The power consumption control system further includes a system-on-a-chip (SOC), a graphics card, a power consumption monitoring module, and a graphics card power consumption detection module for monitoring. The graphics card power consumption detection module is connected to the power supply interface of the graphics card to supply power to the graphics card. The power consumption control method includes: The power consumption monitoring module acquires power consumption monitoring data for the SOC and the graphics card, wherein the power consumption monitoring data for the graphics card is obtained by the power consumption monitoring module through the graphics card power consumption detection module. The load trend is predicted based on the power consumption monitoring data, and the power consumption of the SOC and the graphics card is controlled based on the load trend prediction results.

[0007] Secondly, embodiments of this application provide a power consumption control module, including: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the power consumption control method as described in the first aspect.

[0008] Thirdly, embodiments of this application also provide a power consumption control system, including: a power consumption monitoring module, a SOC, a graphics card, a graphics card power consumption detection module, and a power consumption control module as described in the second aspect; the power consumption monitoring module is connected to the SOC, the graphics card power consumption detection module, and the power consumption control module respectively; the graphics card power consumption detection module is connected to the power supply interface of the graphics card to supply power to the graphics card; The power consumption monitoring module is used to monitor the power consumption of the SOC and the power consumption of the graphics card through the graphics card power consumption detection module to obtain power consumption monitoring data.

[0009] This application provides a power consumption control method, a power consumption control module, and a power consumption control system. The power consumption control method is applied to the power consumption control module within the power consumption control system. The power consumption control system also includes a system-on-a-chip (SOC), a graphics card, a power consumption monitoring module, and a graphics card power consumption detection module for monitoring. The graphics card power consumption detection module is connected to the power supply interface of the graphics card to supply power to the graphics card. The power consumption control method includes: acquiring power consumption monitoring data of the SOC and the graphics card from the power consumption monitoring module, wherein the power consumption monitoring data of the graphics card is obtained by the power consumption monitoring module through the graphics card power consumption detection module; predicting load trends based on the power consumption monitoring data; and controlling the power consumption of the SOC and the graphics card based on the load trend prediction results. This technical solution achieves real-time monitoring of SOC and graphics card power consumption by utilizing the power consumption monitoring module and the graphics card power consumption detection module within the system. Furthermore, the power consumption control module can achieve coordinated management of SOC and graphics card power consumption based on predicted load trends, without relying on graphics card drivers or third-party applications. This improves the efficiency of power consumption control and the coordination of power consumption between the two components, thereby reducing resource conflicts and power waste, and improving resource utilization and system performance. Attached Figure Description

[0010] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0011] Figure 1 A flowchart illustrating a power consumption control method provided in an embodiment of this application; Figure 2 A flowchart illustrating a power consumption control process provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the temperature, real-time power consumption, load trend, and power consumption limit value of a SOC during a power consumption control process, as provided in one embodiment. Figure 4 This is a schematic diagram of the structure of a power consumption control module provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a power consumption control system provided in an embodiment of this application; Figure 6 This is a schematic diagram illustrating the implementation of a power consumption control system according to one embodiment. Detailed Implementation

[0012] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present application, not the entire structure.

[0013] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or power control methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. The process can be terminated when its operation is complete, but may also have additional steps not included in the figures. The process may correspond to a power control method, function, procedure, subroutine, subroutine, etc.

[0014] It should be noted that the concepts of "first" and "second" mentioned in the embodiments of this application are only used to distinguish different devices, modules, units or other objects, and are not used to limit the order of functions performed by these devices, modules, units or other objects or their interdependencies.

[0015] Furthermore, the embodiments and features described in this application may be combined with each other, unless otherwise specified.

[0016] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0017] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the relevant content of the solution.

[0018] Figure 1 This is a flowchart illustrating a power consumption control method provided in an embodiment of this application. This embodiment is applicable to situations where SOC power consumption and graphics card power consumption are managed uniformly. Specifically, this power consumption control method can be executed by a device, which can be implemented through software and / or hardware and integrated into the power consumption control module of a power consumption control system.

[0019] The power control system can be a system integrating a System-on-a-Chip (SoC) and a graphics card, such as an AIPC. The SoC can integrate multiple processors to achieve local heterogeneous computing power. For example, the SoC may include a Graphics Processing Unit (GPU), a Neural Network Processing Unit (NPU), and / or a Central Processing Unit (CPU). The power control system includes a power control module, as well as the SoC, graphics card, a power management module, and a graphics card power detection module for monitoring. The graphics card power detection module is connected to the graphics card's power supply interface to supply power to the graphics card and provides the power management module with monitoring functions for the graphics card's power consumption. The power control in this application embodiment mainly describes the process of monitoring and managing the SoC (including GPU, NPU, and / or CPU) and graphics card.

[0020] like Figure 1 As shown, the power consumption control method specifically includes the following steps: S110. Obtain the power consumption monitoring data of the SOC and the graphics card by the power consumption monitoring module, wherein the power consumption monitoring data of the graphics card is obtained by the power consumption monitoring module through the graphics card power consumption detection module. For example, the power monitoring module is connected to the SOC, and the power control module can obtain the SOC's power monitoring data in real time through the power monitoring module. In addition, the power monitoring module and the graphics card power detection module are connected to the graphics card in sequence. The graphics card power detection module can be connected to the graphics card's external power supply interface through the graphics card adapter cable to supply power to the graphics card. During the power supply process, the power control module can obtain the graphics card's power monitoring data in real time through the power monitoring module and the graphics card power detection module.

[0021] Power consumption monitoring data can be acquired at certain periods or frequencies. Taking a System-on-a-Chip (SoC) as an example, power consumption monitoring data can be a sequence of power consumption values ​​obtained from the SoC, arranged in chronological order.

[0022] Optionally, the SOC may include at least two types of processors, such as CPU, GPU and NPU. The SOC power consumption monitoring data may include total power consumption monitoring data of all processors in the SOC, or power consumption monitoring data of each processor.

[0023] It is understandable that in an AIPC, in addition to the SOC and graphics card, other system devices may also be included, such as onboard devices and external I / O devices. The total power consumption of the system (i.e., the total power consumption of the PCB) may include the power consumption of the SOC, the power consumption of the graphics card, and the power consumption of other system devices.

[0024] S120. Predict the load trend based on the power consumption monitoring data, and control the power consumption of the SOC and the graphics card based on the load trend prediction results.

[0025] For example, based on power consumption monitoring data of the SOC and graphics card, the load trends of the SOC and graphics card in the future can be predicted, and dynamic power consumption management can be performed accordingly. When predicting load trends, factors such as temperature, currently running applications or tasks, and user settings or usage habits can also be considered for comprehensive prediction. For instance, if the SOC temperature continues to rise, it indicates that the SOC load will increase. Similarly, different applications or tasks affect the load trends of the SOC and graphics card; for example, the graphics card load will increase when running computer games, the GPU load will increase when processing images or videos, the graphics card load will decrease when listening to music, and the CPU load will increase when searching for files. Furthermore, if the user has set a power-saving mode, is seeking long battery life, or is currently using a backup battery, the load of the SOC and / or graphics card will decrease; conversely, if the user has not set a power-saving mode, is seeking high performance, or is currently receiving continuous external power, the load of the SOC and / or graphics card will increase.

[0026] The primary means of power consumption control can be adjusting the power limits for the System-on-a-Chip (SoC) and / or the graphics card. For example, if the SoC's load is predicted to increase, the SoC's power limit can be increased, such as from 10W to 15W, while the graphics card's power limit can be appropriately decreased. Conversely, if the SoC's load is predicted to decrease, the SoC's power limit can be decreased, such as from 15W to 10W, and the saved power (5W) can be allocated to the graphics card, i.e., the graphics card's power limit can be increased by 5W. Furthermore, for the SoC, power consumption control can also be achieved by adjusting the frequency or frequency limit.

[0027] Optionally, a System-on-a-Chip (SoC) may include at least two types of processors. When predicting the load trend of the SoC, the overall load trend of the SoC can be predicted, or the load trends of each processor can be predicted separately. Controlling the power consumption of the SoC can be achieved by increasing or decreasing the overall power consumption limit of the SoC, or by increasing or decreasing the power consumption limit of any one or more processors within the SoC. For example, if it is predicted that the load of the SoC will increase, specifically the GPU load, while the CPU and NPU loads will decrease or remain unchanged, then the overall power consumption limit of the SoC and the power consumption limit of the GPU can be increased, and the power consumption limits of one or more processors other than the GPU in the SoC can be appropriately decreased, and / or the power consumption limit of the graphics card can be decreased. Based on this, coordinated power consumption management of the SoC (including its processors) and the graphics card can be achieved.

[0028] Optionally, when performing power consumption control, priorities can be assigned to each processor in the SoC. When power consumption limits need to be adjusted, the needs of high-priority processors can be prioritized. For example, if the CPU has the highest priority, its power consumption limit can be increased if the SoC load is predicted to increase; conversely, if the SoC load is predicted to decrease, the CPU's power consumption limit can be decreased. If this still doesn't meet the demand, then the power consumption limits of the NPU and / or GPU can be reduced. Furthermore, if reducing the GPU's power consumption limit can save power, the CPU's power consumption limit can be increased first. If the CPU's power consumption limit is already high enough, the saved power can be added to the NPU and / or GPU's power consumption limits, and so on.

[0029] The power consumption control method in this embodiment utilizes a power consumption monitoring module and a graphics card power consumption detection module within the system to achieve real-time monitoring of SOC power consumption and graphics card power consumption. Furthermore, the power consumption control module can achieve coordinated management of SOC power consumption and graphics card power consumption based on predicted load trends, without relying on graphics card drivers or third-party applications. This improves the efficiency of power consumption control and the coordination of power consumption between the two, thereby reducing resource conflicts and power waste, and improving resource utilization and system performance.

[0030] In one embodiment, the step of predicting load trends based on the power consumption monitoring data and controlling the power consumption of the SOC and the graphics card based on the load trend prediction results includes: S1210, Obtain auxiliary data; For example, auxiliary data can refer to data other than power consumption monitoring data that can provide reference and assistance for predicting load trends, such as temperature data, usage preference information and / or task type.

[0031] In this embodiment, the process of acquiring auxiliary data may include at least one of S1-S3.

[0032] S1. Acquire temperature data collected by the temperature sensor. The temperature data includes ambient temperature data, SOC temperature data, and / or graphics card temperature data.

[0033] For example, if the SOC temperature shows an upward or downward trend, the SOC load will increase or decrease accordingly; if the graphics card temperature shows an upward or downward trend, the graphics card load will increase or decrease accordingly; if the ambient temperature shows an upward or downward trend, the SOC and graphics card temperatures may increase or decrease accordingly to some extent. Based on this, the magnitude of the temperature change of the SOC and graphics card can be further analyzed to determine the degree of load change of the SOC and graphics card. Specifically, if the ambient temperature shows an upward trend, the SOC and graphics card temperatures may increase to a certain extent. If the increase in the SOC and / or graphics card temperature is large, it can be determined that the load of the SOC and / or graphics card has increased accordingly; otherwise, it can be considered a normal temperature change affected by the ambient temperature.

[0034] Optionally, temperature data can be acquired according to a certain period or frequency. Taking SOC as an example, temperature data can be a sequence of temperature values ​​composed of the collected SOC temperature values ​​arranged in chronological order.

[0035] Optionally, the SOC may include at least two types of processors, and the SOC temperature data may include the overall temperature data of the SOC or the temperature data of each processor.

[0036] S2. Determine user preference information through artificial intelligence models; For example, an artificial intelligence (AI) model can infer user preference information by analyzing a user's historical usage data. This preference information may include the user's load trends for specific tasks (such as playing games, listening to music, or editing documents) and / or during specific time periods (such as day or night) and / or in specific scenarios (such as indoors or outdoors, or with or without external power supply), and / or the power consumption adjustment strategies adopted. The preference information may also include the priority of graphics cards, SoCs, and various processors within them. The preference information may also include the user's needs, such as whether they prioritize battery life, performance, or the performance of a particular processor.

[0037] For example, for various processors in a SoC, the usage preference information can be summarized into the following patterns: a. Performance mode prioritizes ensuring that the workload of the higher (or highest) priority processor is met, and prioritizes increasing the power consumption limit of the higher (or highest) priority processor. b. Balanced mode, which prioritizes ensuring that the load of the second-highest (or lowest) priority processor can be met, and prioritizes increasing the power consumption limit of the second-highest (or lowest) priority processor; c. Power saving mode: Prioritizes reducing the power consumption limit of the higher (or highest) priority processor to ensure it can maintain basic operation, and then considers reducing the power consumption limit of the second (or lowest) priority processor based on the actual situation.

[0038] d. User-defined mode, such as for specific tasks, time periods or scenarios, where it is necessary to maximize the performance of a certain processor, the power consumption limit of this processor is increased first, and other processors can be managed according to the default mode (such as balanced mode).

[0039] S3. Determine the type of task currently running.

[0040] For example, the type of task currently running affects the load trends of the SOC and graphics card. For instance, the graphics card load will increase when running computer games, the GPU load will increase when processing images or videos, the graphics card load will decrease when listening to music, and the CPU load will increase when searching for files.

[0041] S1220. Predict the load trend based on the power consumption monitoring data and auxiliary data, and control the power consumption of the SOC and the graphics card based on the load trend prediction result and the auxiliary data. For example, power consumption monitoring data and auxiliary data can be used together as inputs or variables to obtain load prediction results; alternatively, according to the application scenario and actual needs, corresponding weights can be set for power consumption monitoring data and auxiliary data respectively, so as to consider the degree of influence of different factors on load prediction and make more reliable predictions; or the load trend can be initially predicted based on power consumption monitoring data, and then the initial prediction results can be updated or corrected based on auxiliary data.

[0042] Based on this, load trends can be predicted more accurately and comprehensively, providing a more reliable basis for power consumption control.

[0043] It should be noted that this embodiment does not specifically limit the mapping relationship and prediction method between auxiliary data and load trends. For example, the mapping relationship can be solved or modeled based on linear or nonlinear functions, fitting methods and / or machine learning algorithms, and the load can be predicted based on the auxiliary data. Alternatively, the mapping relationship can be established based on experience or experiments, and the load can be predicted based on the auxiliary data.

[0044] In one embodiment, the SOC includes at least one processor; The method of controlling the power consumption of the SOC and the graphics card based on load trend prediction results includes: S1230, Determine the priority of each processor; S1240. Determine the target processor based on the load trend prediction results and the priority of each processor; S1250, Adjust the power consumption of the target processor and the power consumption of the graphics card.

[0045] In this context, the target processor can be understood as the processor whose power consumption limit needs to be adjusted. The priority of each processor can be a default value or set by the user. For different tasks, time periods, or scenarios, the priority of each processor can also be different, and the corresponding target processor can also be different. For example, when running computer games or between 7:00 PM and 9:00 PM, the GPU has a higher priority than the CPU, and the NPU has the lowest priority; when listening to music or without a stable external power supply, the CPU has a higher priority than the GPU, and the NPU has the lowest priority, and so on. Based on the load trends of the SOC and its various processors, the power consumption limits of some processors can be adjusted according to priority as needed. If this still does not meet the requirements, the power consumption limits of other processors can be adjusted according to priority until the requirements are met. It should be noted that this embodiment does not specifically limit the priorities of various processors; they can be flexibly set and dynamically adjusted according to actual needs. Based on this, the power consumption of the SOC can be flexibly controlled according to the priorities of different processors to meet the diverse practical needs of different application scenarios.

[0046] In one embodiment, the load trend prediction result is based on the weighted average power consumption at the current moment.

[0047] In this embodiment, the Exponentially Weighted Moving Average (EWMA) algorithm is used to predict the load trend. The load trend prediction result can be expressed as follows: ;in, The weighted average power consumption at the current moment. This is the current power consumption monitoring value. This is the power consumption monitoring value from the previous moment. The sampling period can be defined by software, such as 10ms or 100ms. These are predefined constants that can be defined based on AI learning, historical data, and / or experience. ,when At this time, EWMA1 = Pidle, and Pidle can be understood as the power consumption when the system is idle. Optionally, the graphics card, SOC, and various processors in the SOC can all be predicted based on the corresponding load trend results.

[0048] Based on this, the EWMA algorithm can assign exponentially decaying weights to historical load data, making recent data contribute more to the current average, thereby more accurately measuring load trends, making load forecasting more timely, responding to trend changes faster, reducing lag, providing a reliable basis for power consumption control, and also reducing storage and computational complexity, improving computational efficiency and flexibility.

[0049] In one embodiment, controlling the power consumption of the SOC and the graphics card based on load trend prediction results includes: If the difference between the power consumption limit of the target processor and the weighted average power consumption of the target processor is less than or equal to a set threshold, the power consumption limit of the target processor shall be increased; optionally, the power consumption limit of non-target processors and / or graphics cards may be appropriately reduced. If the difference between the power consumption limit of the target processor and the weighted average power consumption of the target processor is greater than a set threshold, the power consumption limit of the target processor shall be reduced; alternatively, the power consumption limit of non-target processors and / or graphics cards may be appropriately increased.

[0050] For example, the current available power budget can be calculated using load trend forecast results. The current available power budget can be the difference between the current power limit and the load trend forecast result: .

[0051] Taking the CPU in a System-on-a-Chip (SoC) as an example, if the CPU's current available power budget is greater than 0, it indicates that there is still some surplus power. If the CPU's current available power budget is relatively large (greater than a set threshold), it means that its power limit is significantly higher than the actual power consumption and can be appropriately reduced. The saved power can be allocated to other processors in the SoC or to the graphics card. If the CPU's current available power budget is less than or equal to 0, it means that the CPU's load has reached or exceeded its capacity. If the exceedance is significant (i.e., the current available power budget is less than or equal to the set threshold), it means that its power limit is significantly lower than the actual power consumption and its power limit can be appropriately increased. The power limits of other processors or graphics cards in the SoC can then be appropriately reduced. Based on this, accurate analysis of the available power budget can provide a basis for power control of each processor and graphics card in the SoC, improving the reliability, flexibility, and coordination of power control.

[0052] In one embodiment, the method further includes: S130. In the event of abnormal power consumption and / or abnormal temperature, perform frequency reduction or shutdown operation on the processor in the SOC.

[0053] For example, during power consumption control, in the event of abnormal power consumption and / or temperature, emergency actions can be taken, such as downclocking the processor in the SOC. If the power consumption and / or temperature remain abnormal after one or more downclocking attempts, the user can be notified of the current anomaly and an imminent shutdown, thereby preventing continued anomalies from affecting system functionality or causing damage to the system. Based on this, an anomaly handling mechanism is provided to effectively adjust the power consumption of the target processor and graphics card, promptly resolve anomalies, and improve system stability.

[0054] In one embodiment, the method further includes: Implement at least one of the following functions through an application (APP) and / or a graphical user interface (GUI): Display information on power consumption control, such as load trend prediction results, auxiliary data, power consumption limits, power consumption control process, and / or power consumption control results; Obtain user input regarding usage requirements, such as user-defined tasks, scenarios, modes, and / or priorities; Collect users' historical usage data, such as users' usage needs, preferences and / or power consumption control strategies and results for historical tasks, different scenarios and different time periods; The system displays recommended power control operations, which are determined based on the usage requirements and historical usage data, such as recommended modes, priorities, and / or power limit values.

[0055] Based on this, it can interact with users in a visual way, promptly grasp users' usage needs and habits, recommend or suggest suitable power control operations, and facilitate users to understand and follow up on the power control process in real time, control the effect of power control, and improve the user experience.

[0056] Figure 2 A flowchart illustrating a power consumption control process provided in an embodiment of this application; as shown Figure 2 As shown, the exemplary dynamic power consumption coordination management process mainly includes: The power consumption monitoring module and the graphics card power consumption detection module monitor the real-time power consumption of the SOC and graphics card to obtain power consumption monitoring data. The ambient temperature, SOC temperature and graphics card temperature are collected in real time by the temperature sensor to obtain temperature data. Ensure that the system is currently executing a task; Analyze user habits to obtain user preference information (set patterns); Based on power consumption monitoring data, temperature data, current tasks, and usage preference information, the system analyzes the current load status, such as which processor accounts for the majority of the SOC load, whether the load of each processor is balanced, or whether the load of the SOC and graphics card is appropriate, too high, or too low. It can also assess the current power bottleneck, such as which processor has an excessively high or low load, or whether there are abnormal power consumption or temperature. Based on this, the system predicts the load trend, and the current available power budget can also be determined based on the load trend prediction results. Determine the priority of power consumption modules. Power consumption modules can include different processors. Taking CPU and GPU as examples, GPU can be further divided into integrated graphics processing units (iGPU) and discrete graphics processing units (dGPU). The target SOC is determined based on the load trend forecast results and the priority of power consumption modules; Adjusting the power consumption of the target SOC can also appropriately adjust the power consumption of non-target SOCs; Based on the effects of power consumption adjustment, such as whether there are abnormal power consumption and / or abnormal temperature, new power consumption data and auxiliary data can be obtained. Based on this, the AI ​​model can be optimized and updated in conjunction with the power consumption control process, so as to better analyze user preferences and make more accurate load trend predictions in subsequent execution. If power consumption and / or temperature are abnormal, the processor in the SOC can be downclocked. If abnormal power consumption and / or temperature persist after frequency reduction, an emergency mechanism can be activated to reduce the frequency to the lowest possible level. If the abnormality still exists, the user can be notified and the device can be shut down.

[0057] Table 1 shows the changes in SOC power consumption and GPU power consumption during power control. As shown in Table 1, the exemplary total system power consumption includes SOC power consumption, GPU power consumption, and power consumption of other system devices. The SOC includes NPU, GPU, and other processors (SOC-other, such as CPU). The GPU can be divided into iGPU and dGPU, and the power consumption of iGPU is recorded here. GPU power consumption can also be divided into power consumption detected from the PCIe slot and power consumption detected from the external power supply; system device power consumption can be divided into onboard device power consumption and external device power consumption. The power consumption priority records the highest priority processor in the SOC. For example, "CPU" indicates that the CPU has the highest priority, "N / A" indicates no priority setting, and dynamic CPU / GPU / NPU indicates no fixed setting. The priority of each processor can be set according to the actual situation.

[0058] Table 1. Changes in SOC power consumption and GPU power consumption Figure 3 This is a schematic diagram illustrating the temperature, real-time power consumption, load trend, and power consumption limit value of a System-on-a-Chip (SOC) during a power consumption control process, as provided in an embodiment of this application. Figure 3 As shown, when the temperature triggers the maximum setpoint (SOC 88°), the power consumption begins to decrease to the maximum continuous power consumption of 35W at SOC.

[0059] As can be seen, by adopting the power consumption control method of this application, appropriate power consumption control strategies can be used for different scenarios. By predicting load trends and available power consumption budget, and adjusting the power consumption limit (PWR LIMIT), coordinated management of the total power consumption of the system and the power consumption of the SOC and graphics card can be achieved.

[0060] The power consumption control method of this application decouples the dependence on third-party software by connecting the external power supply path of the PCIe graphics card to the graphics card power consumption detection module on the system PCB board. It can deeply integrate with the software to adjust the power consumption of the SOC and graphics card in real time, greatly improving the versatility of the solution. When both the SOC and graphics card are under high load, real-time data (power consumption and temperature) can be used in conjunction with the current power consumption control strategy (preference information or mode) to solve the problem of performance not being maximized when resources conflict. When the SOC or graphics card load is not high, power consumption can be dynamically allocated through the real-time power consumption monitoring module and power consumption prediction mechanism to maximize power utilization and improve the response speed during power switching, thereby improving the user experience. When the PCIe graphics card is limited by the maximum power consumption limit of the slot, the system power consumption can be maximized by reducing the power consumption of the SOC or system or by designing the system. In addition, through visualization of GPI / APP, combined with AI, the system can continuously learn, upgrade and optimize the power allocation strategy, and provide customers with one-click configuration of the optimal power consumption strategy, thereby improving the user experience.

[0061] Figure 4 A schematic diagram of a power control module 10, which can be used to implement embodiments of this application, is shown. The power control module 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The power control module 10 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, user equipment, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0062] like Figure 4As shown, the power consumption control module 10 includes a controller 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, which is communicatively connected to the controller 11. The memory stores computer programs that can be executed by the controller. The controller 11 can perform various appropriate actions and processes based on the computer programs stored in the ROM 12 or loaded into the RAM 13 from the storage unit 18. The RAM 13 can also store various programs and data required for the operation of the power consumption control module 10. The controller 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0063] Multiple components in the power control module 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard or mouse; an output unit 17, such as various types of displays or speakers; a storage unit 18, such as a disk or optical disk; and a communication unit 19, such as a network card, modem, or wireless transceiver. The communication unit 19 allows the power control module 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks and wireless networks.

[0064] Controller 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of controller 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Controller 11 performs the various power control methods and processes described above.

[0065] In some embodiments, the power consumption control method described above can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the power consumption control module 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by controller 11, one or more steps of the power consumption control method described above can be performed. Alternatively, in other embodiments, controller 11 can be configured to perform the power consumption control method of any of the embodiments described above by any other suitable means (e.g., by means of firmware).

[0066] Figure 5 This is a schematic diagram of a power consumption control system provided in one embodiment. Figure 5As shown, the power consumption control system includes: a power consumption monitoring module 20, a SOC 30, a graphics card 40, a graphics card power consumption detection module 50, and a power consumption control module 10 as described in any of the above embodiments; the power consumption monitoring module 20 is connected to the SOC 30, the graphics card power consumption detection module 50, and the power consumption control module 10 respectively; the graphics card power consumption detection module 50 is connected to the power supply interface of the graphics card 40 to supply power to the graphics card 40; the power consumption monitoring module 20 is used to monitor the power consumption of the SOC 30 and monitor the power consumption of the graphics card 40 through the graphics card power consumption detection module 50 to obtain power consumption monitoring data.

[0067] In one embodiment, the system further includes: a temperature sensor for collecting temperature data, including ambient temperature data, SOC temperature data, and graphics card temperature data; a power consumption control module 10 specifically for predicting load trends based on power consumption monitoring data and temperature data, and controlling the power consumption of the SOC and the graphics card based on the load trend prediction results and temperature data; the power consumption control module 10 is also used to: perform frequency reduction or shutdown operations on the processor in the SOC in the event of abnormal power consumption and / or abnormal temperature.

[0068] In one embodiment, the graphics card power consumption detection module 50 includes a backup power supply module, which is used to supply power to the graphics card using the power supply of the SOC 30 when no external power supply is used.

[0069] For example, taking a 100W system motherboard as an example, it can connect to a 75W PCIe graphics card (without an external power supply interface). When overclocked to 90W for a short time, it can reduce the power consumption of the SOC system and support its stable operation. It can also connect to a 150W external power supply (with 2*3 external power supply interfaces). Through the design of the backup power supply (Oring) module, it can support the graphics card to perform at 75W~90W without the need for an external power supply. If the graphics card is powered by an external power supply through the motherboard + 2*3 PICE adapter cable, it can dynamically switch between the system motherboard power supply and the external power supply, thereby maximizing power consumption utilization.

[0070] Figure 6 This is a schematic diagram illustrating the implementation of a power consumption control system according to one embodiment. For example... Figure 6 As shown, module ① is a power consumption monitoring (PM) module, which can be used to monitor the power consumption of SOC (including CPU, GPU and NPU), PCIe slot, PCIe graphics card (PCIe GFX) external power supply, etc.

[0071] The PCIe graphics card external power supply power consumption detection module can be connected to the motherboard. It should be noted that in traditional PCIe external power supply solutions, the external power supply directly powers the graphics card through the graphics card power cable. However, in this embodiment, the external power supply is first connected to the motherboard (2*3 / 2*4 standard graphics card power connectors), then passes through the graphics card power consumption detection module, and finally connects to the PCIe graphics card external power supply connector via a graphics card adapter cable. This solves the problem of relying on graphics card drivers or third-party software for real-time graphics card power consumption monitoring, and is compatible with different graphics card power control schemes.

[0072] Module ② is the power consumption control module, mainly used for dynamic power consumption management. Utilizing the real-time power consumption of the system and graphics card obtained by Module ①, combined with data feedback from the temperature sensor, the software uses a power consumption coordination management mechanism to predict future short-term load trends based on the current application, dynamically adjusting the power consumption of the system and PCIe graphics card to improve response speed and maximize the performance of the SOC and PCIe graphics card. In addition, the temperature sensor can combine ambient temperature, CPU, GPU, NPU, and PCIe power supply temperature to comprehensively determine and predict load trends, achieving optimal control of temperature and performance.

[0073] Module ③ is the anomaly handling module, primarily responsible for anomaly handling mechanisms. Based on real-time data from the power consumption detection module and combined with AI algorithms, it determines or predicts potential power supply anomalies, temperature anomalies, or system instability, and promptly reduces the frequency or shuts down the device to protect it from damage to the greatest extent possible. It should be noted that the function of Module ③ can also be implemented by Module ②; that is, the anomaly handling module can be merged with the power consumption control module, included within the power consumption control module, or it can be a separate module.

[0074] Module ④ is the visualization module, which is mainly used to develop user-visible and controllable APP / GUI interfaces, allowing users to see real-time power allocation or allocate power consumption as needed. Under the condition that the user allows, it can collect customer usage data and combine AI to recommend the optimal power consumption configuration rate for the user.

[0075] Module ⑤ is the backup power supply (oring) module. Referring to the power supply path shown by the red line, this module supports external power supply, or in scenarios where the system PCB power consumption is sufficient to support the power supply of the graphics card, it can be directly powered from the system PCB (from the PCB power supply path triggered by the SOC) without the need for an additional external power supply to power the graphics card. This achieves power redundancy backup, rapid isolation in the event of power failure, and seamless switching between different power supplies, maximizing power utilization.

[0076] The power control system described in this application decouples the dependence on third-party software by connecting the external power supply path of the PCIe graphics card to the graphics card power consumption detection module on the system PCB board. This allows for deep integration with software to adjust the power consumption of the SOC and graphics card in real time, greatly improving the versatility of the solution. When both the SOC and graphics card are under high load, real-time data (power consumption and temperature) combined with current power control strategies (preference information or modes) can be used to resolve performance issues caused by resource conflicts. When the SOC or graphics card load is low, power consumption can be dynamically allocated through a real-time power monitoring module and power prediction mechanism, maximizing power utilization and improving response speed during power switching, thus enhancing the user experience. When the PCIe graphics card is limited by the maximum power consumption of the slot, the system power consumption can be maximized by reducing the SOC or system power consumption or by optimizing the design. Furthermore, through a visualized GPI / APP, combined with AI, the system can continuously learn, upgrade, and optimize power allocation strategies, and provide customers with one-click configuration of optimal power consumption strategies, further enhancing the user experience.

[0077] The power consumption control system provided in this application can be used to implement the power consumption control method provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0078] Various implementations of the power control systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0079] Computer programs used to implement the power consumption control method of this application can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on the machine, partially on the machine, or as a standalone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0080] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution power control system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor power control systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0081] The power control systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or middleware components (e.g., an application server), or frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with the implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., a communication network) of any form or medium. Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0082] This application also provides a computer program product, including a computer program and / or instructions, which, when executed by a processor, implement the power consumption control method as described in any of the above embodiments.

[0083] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

[0084] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A power consumption control method, characterized in that, A power control module is used in a power control system, which also includes a system-on-a-chip (SOC), a graphics card, a power monitoring module, and a graphics card power detection module for monitoring. The graphics card power detection module is connected to the power supply interface of the graphics card to supply power to the graphics card. The power consumption control method includes: The power consumption monitoring module acquires power consumption monitoring data for the SOC and the graphics card, wherein the power consumption monitoring data for the graphics card is obtained by the power consumption monitoring module through the graphics card power consumption detection module. The load trend is predicted based on the power consumption monitoring data, and the power consumption of the SOC and the graphics card is controlled based on the load trend prediction results.

2. The method according to claim 1, characterized in that, The step of predicting load trends based on the power consumption monitoring data and controlling the power consumption of the SOC and the graphics card based on the load trend prediction results includes: Acquire auxiliary data; The load trend is predicted based on the power consumption monitoring data and auxiliary data, and the power consumption of the SOC and the graphics card is controlled based on the load trend prediction results and the auxiliary data. The acquisition of auxiliary data includes at least one of the following: Acquire temperature data collected by a temperature sensor, including ambient temperature data, SOC temperature data, and / or graphics card temperature data; Determine user preferences using artificial intelligence models; Determine the type of task currently running.

3. The method according to claim 1, characterized in that, The SOC includes at least one processor; The method of controlling the power consumption of the SOC and the graphics card based on load trend prediction results includes: Determine the priority of each processor; In the event of abnormal power consumption and / or abnormal temperature, the target processor is determined based on the load trend prediction results and the priority of each processor. Adjust the power consumption of the target processor and the graphics card.

4. The method according to claim 1, characterized in that, The load trend prediction result is based on the weighted average power consumption at the current moment, and the load trend prediction result is expressed as: ; where is the weighted average power consumption at the current moment, is the power consumption at the current moment, is the power consumption at the previous moment, is the sampling period, and is a predefined constant.

5. The method according to claim 4, characterized in that, The method of controlling the power consumption of the SOC and the graphics card based on load trend prediction results includes: If the difference between the power consumption limit of the target processor and the weighted average power consumption of the target processor is less than or equal to a set threshold, the power consumption limit of the target processor shall be increased. If the difference between the power consumption limit of the target processor and the weighted average power consumption of the target processor is greater than a set threshold, the power consumption limit of the target processor shall be reduced.

6. The method according to claim 1, characterized in that, Its characteristic is that it further includes: In the event of abnormal power consumption and / or abnormal temperature, the processor in the SOC will be downclocked or shut down.

7. A power consumption control module, characterized in that, include: Controller; The memory is communicatively connected to the controller; wherein, The memory stores a computer program that can be executed by the controller, the computer program being executed by the controller to enable the controller to perform the power consumption control method as described in any one of claims 1-6.

8. A power consumption control system, characterized in that, include: The power consumption monitoring module, the system-on-a-chip (SOC), the graphics card, the graphics card power consumption detection module, and the power consumption control module as described in claim 7; The power consumption monitoring module is connected to the SOC, the graphics card power consumption detection module, and the power consumption control module, respectively; the graphics card power consumption detection module is connected to the power supply interface of the graphics card to supply power to the graphics card; The power consumption monitoring module is used to monitor the power consumption of the SOC and the power consumption of the graphics card through the graphics card power consumption detection module to obtain power consumption monitoring data.

9. The system according to claim 8, characterized in that, Also includes: A temperature sensor is used to collect temperature data, including ambient temperature data, SOC temperature data, and graphics card temperature data. The power consumption control module is specifically used to predict the load trend based on the power consumption monitoring data and the temperature data, and to control the power consumption of the SOC and the graphics card based on the load trend prediction result and the temperature data. The power consumption control module is also used to: perform frequency reduction or shutdown operations on the processor in the SOC in the event of abnormal power consumption and / or abnormal temperature.

10. The system according to claim 8, characterized in that, The graphics card power consumption detection module includes a backup power supply module, which is used to supply power to the graphics card using the power supply of the SOC when no external power supply is used.

Images