A method, apparatus and electronic device for determining operating mode
By collecting and fusing power consumption, utilization, and temperature information of hardware units, and combining load change trends and real-time status, the operating mode is precisely scheduled, solving the problems of lag and misjudgment in temperature threshold control in existing technologies, and improving the performance and energy efficiency of computing devices.
Patent Information
- Application Number
- CN202610983568.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-02
- Publication Date
- 2026-11-10
AI Technical Summary
In existing technologies, relying on temperature thresholds to regulate hardware power consumption has problems of lag and misjudgment, resulting in wasted performance and energy efficiency of computing devices such as laptops under high load scenarios, and poor user experience.
By collecting power consumption, utilization, and temperature information of hardware units, a multi-dimensional fusion evaluation is performed. Combined with load change trends and real-time status, the load status is accurately identified and the operating mode is scheduled in advance to avoid mode switching lag.
It achieves stable operation of hardware units under different load scenarios, balances performance release and energy consumption control, improves system response speed and overall operating efficiency, and reduces mode switching lag.
Smart Images

Figure CN123067672A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of data processing technology, and in particular to a method, apparatus and electronic device for determining operating modes. Background Technology
[0002] In computing devices such as laptops, the built-in cooling modules have a fixed physical limit to heat dissipation. The higher the power consumption of the internal hardware units, the greater the heat generated during operation. Constrained by this cooling limit, each hardware component cannot simultaneously maintain optimal performance. Therefore, power allocation among the various hardware units is particularly important. Summary of the Invention
[0003] This disclosure provides a method, apparatus, and electronic device for determining operating modes, in order to at least solve the above-mentioned technical problems existing in the prior art.
[0004] A first aspect of this disclosure provides a method for determining an operating mode, the method comprising: Collect power consumption, utilization, and temperature information of the target hardware unit; Based on the power consumption information, utilization information, and temperature information, the real-time load status of the target hardware unit is determined; Based on the power consumption information and the utilization information, the load change trend of the target hardware unit is determined; Obtain a judgment result on whether the load change trend meets the preset control conditions; Based on the judgment result, and one of the real-time load status and the load change trend, the operating mode of the target hardware unit is determined; wherein the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
[0005] In one possible implementation, determining the operating mode of the target hardware unit based on the judgment result, the real-time load status, and the load change trend includes: If the judgment result indicates that the load change trend meets the preset control conditions, the operating mode of the target hardware unit is determined based on the load change trend; If the judgment result indicates that the load change trend does not meet the preset control conditions, the operating mode of the target hardware unit is determined based on the real-time load status.
[0006] In one possible implementation, determining the real-time load state of the target hardware unit based on the power consumption information, utilization information, and temperature information includes: Based on the power consumption information, the utilization information, and the temperature information, the high load factor and the low load factor are determined. In response to the high load factor satisfying the preset high load condition, the real-time load state of the target hardware unit is determined to be a high load state; In response to the low load factor satisfying the preset low load condition, the real-time load state of the target hardware unit is determined to be a low load state.
[0007] In one possible implementation, determining the load change trend of the target hardware unit based on the power consumption information and the utilization information includes: Based on the power consumption information, determine the initial power consumption change rate; Based on the utilization information, determine the initial utilization rate change rate; The initial power consumption change rate and the initial utilization rate change rate are respectively subjected to smoothing filtering to obtain the smoothed power consumption change rate and the smoothed utilization rate change rate; Determine the current steady-state load bias term of the target hardware unit; Based on the smoothed power consumption change rate, the smoothed utilization change rate, and the steady-state load bias term, the load change trend is obtained.
[0008] In one possible implementation, determining a low load factor based on the power consumption information, the utilization information, and the temperature information includes: Obtain the temperature threshold; Based on the temperature information and the temperature threshold, a temperature constraint penalty term is determined; Based on the power consumption information, utilization information, temperature information, and the temperature constraint penalty term, a low load factor is determined.
[0009] In one possible implementation, determining the current steady-state load bias term of the target hardware unit includes: Obtain the real-time load status of the target hardware unit and its corresponding preset bias coefficient; Based on the real-time load status and its corresponding preset bias coefficient, the current steady-state load bias term of the target hardware unit is determined.
[0010] In one possible implementation, the preset control conditions include a positive prediction threshold and a negative prediction threshold; obtaining the judgment result of whether the load change trend meets the preset control conditions includes: The load change trend is compared with the positive prediction threshold and the negative prediction threshold respectively to obtain the comparison results; If the comparison result is that the load change trend is greater than or equal to the positive prediction threshold, the judgment result is determined to be that the load change trend meets the preset control conditions; If the comparison result is that the load change trend is less than or equal to the negative prediction threshold, the judgment result is determined to be that the load change trend does not meet the preset control conditions.
[0011] In one possible implementation, the method further includes: Based on the thermal management scheduling mode of the electronic device in which the target hardware unit is located, the preset high load condition and the preset low load condition are determined.
[0012] A second aspect of this disclosure provides an operating mode determining apparatus, the apparatus comprising: The information acquisition module is used to collect power consumption information, utilization information and temperature information of the target hardware unit; The status determination module is used to determine the real-time load status of the target hardware unit based on the power consumption information, utilization information and temperature information. The state determination module is further configured to determine the load change trend of the target hardware unit based on the power consumption information and the utilization information. The operating mode determination module is used to obtain a judgment result on whether the load change trend meets the preset control conditions; based on the judgment result, and one of the real-time load status and the load change trend, the operating mode of the target hardware unit is determined; wherein, the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
[0013] A third aspect of this disclosure provides an electronic device comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods described in this disclosure.
[0014] A fourth aspect of this disclosure provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the methods described in this disclosure.
[0015] The operating mode determination method, apparatus, and electronic equipment disclosed herein accurately identify the real-time load status of the target hardware unit by simultaneously collecting three types of information: power consumption, utilization rate, and temperature. Simultaneously, it predicts the load change trend of the target hardware unit based on power consumption and utilization rate information. Then, based on whether the predicted load change trend meets preset control conditions, it determines the operating mode by combining either the load change trend or the real-time load status. This achieves precise and proactive scheduling of the operating mode, ensuring stable operation of the target hardware unit under different load scenarios. Furthermore, by adapting to first and second modes with different power consumption, it balances performance release and energy consumption control, reduces mode switching lag, and improves system response speed and overall operating efficiency.
[0016] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description
[0017] The above and other objects, features, and advantages of this disclosure will become readily apparent from the following detailed description of exemplary embodiments, taken in conjunction with the accompanying drawings. Several embodiments of this disclosure are illustrated in the drawings by way of example and not limitation, in which: In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.
[0018] Figure 1 A schematic diagram of the composition structure of an electronic device according to an embodiment of the present disclosure is shown; Figure 2 A schematic diagram illustrating the implementation flow of a method for determining an operating mode according to an embodiment of this disclosure is shown; Figure 3 A schematic diagram of the composition structure of an operating mode determination device according to an embodiment of the present disclosure is shown; Figure 4 A schematic diagram of the composition structure of another electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation
[0019] To make the objectives, features, and advantages of this disclosure more apparent and understandable, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.
[0020] For electronic devices such as laptops and desktop computers, the higher the performance of their core hardware units, such as the CPU (Central Processing Unit) and GPU (Graphics Processing Unit), the higher the power consumption and the more heat they generate. Due to the heat dissipation limit of the cooling module, electronic devices cannot support all hardware units to maintain optimal performance at full load simultaneously. Therefore, in high-load and high-heat-pressure usage scenarios, it is necessary to limit the power consumption and performance of low-priority hardware units, and prioritize ensuring that high-priority hardware units achieve optimal performance.
[0021] The relevant technical solutions all rely on the device manufacturers to preset the control logic in the BIOS or EC firmware, using the built-in temperature sensor to collect temperature data of each hardware unit in real time. When the monitored temperature exceeds the preset threshold, the system triggers an adjustment to the overall power consumption allocation strategy. Taking the coordinated control of CPU and GPU as an example, after the temperature exceeds the limit, the device will often switch from dual-stress mode to single-stress mode, that is, stop the state of simultaneous high-load operation of CPU and GPU, and prioritize the stable output performance of CPU by limiting GPU power consumption, and vice versa.
[0022] However, this type of control method, which relies solely on temperature thresholds, has significant drawbacks. When the GPU suddenly takes on a high-load task, such as launching a large game or starting a high-definition rendering job, hardware power consumption and utilization will spike rapidly in a short period. Temperature changes, on the other hand, have an inherent lag and cannot reflect the real-time hardware load status in sync. Relying solely on temperature for judgment makes it difficult for the system to quickly respond to the GPU's actual performance needs, easily leading to issues such as screen stuttering and frame drops. Conversely, when the hardware load suddenly drops, the chassis temperature falls back down slowly, and the power limiting strategy cannot be lifted in time, causing the performance of hardware such as the CPU to fail to recover quickly, resulting in wasted performance and energy efficiency.
[0023] Meanwhile, if the monitored temperature fluctuates slightly around the preset threshold, relying solely on temperature judgment will frequently trigger mode switching, causing the overall system control strategy to oscillate and switch repeatedly, severely impacting the user experience. Long-term, high-frequency switching can also shorten the lifespan of the hardware. In addition, external factors such as blocked heat dissipation vents, increased ambient temperature, and dust accumulation on the chassis can all cause abnormally high temperature sensor readings, triggering incorrect power consumption control logic and further exacerbating the system's malfunctions.
[0024] In view of the technical deficiencies existing in the aforementioned related technologies, this disclosure provides a method, apparatus, and electronic device for determining operating modes. A first aspect of this disclosure provides a method for determining operating modes, which can be applied to various electronic devices such as laptops, desktop computers, servers, and smart terminals. Figure 1This illustrates a specific architecture of an electronic device, which includes a system service module, a driver module, and a Basic Input / Output System (BIOS) scheduling module. Figure 2 As shown, the method includes the following steps: Step 101: Collect power consumption information, utilization information, and temperature information of the target hardware unit.
[0025] The target hardware unit refers to a functional module within an electronic device that requires dynamic adjustment of its operating mode. Specifically, it can be a functional module with a lower computational priority than the CPU, such as a GPU or Neural Processing Unit (NPU). The system service module can periodically collect power consumption, utilization, and temperature information of the target hardware unit through a preset timer cycle (e.g., every 200ms). The power consumption information characterizes the real-time load power consumption of the target hardware unit, the utilization information characterizes the current proportion of computing resources occupied by the target hardware unit, and the temperature information characterizes the real-time operating temperature of the target hardware unit.
[0026] Step 102: Determine the real-time load status of the target hardware unit based on power consumption information, utilization information, and temperature information.
[0027] The system service module at the application layer of the electronic device performs multi-dimensional fusion evaluation of the collected power consumption, utilization, and temperature information to calculate a quantified real-time load coefficient. Based on the magnitude of the real-time load coefficient, it determines the real-time load state of the target hardware unit. This real-time load state characterizes the overall operating load level of the target hardware unit and can be categorized into high-load, low-load, and other types. Specifically, power consumption, utilization, and temperature information are normalized to eliminate differences in dimensions and values between different indicators. Then, a weighted fusion calculation is performed using the corresponding preset weights to obtain the real-time load coefficient. The real-time load coefficient is compared with the judgment thresholds corresponding to various load states. After matching the corresponding threshold range, the real-time load state of the target hardware unit is determined and output.
[0028] Step 103: Based on power consumption information and utilization information, determine the load change trend of the target hardware unit.
[0029] The system service module at the application layer of the electronic device determines the rate and trend of power consumption change for the target hardware unit based on power consumption information. It also determines the rate and trend of utilization change for the target hardware unit based on utilization information. Subsequently, exponential moving average smoothing filtering and weighted fusion calculations are performed on both power consumption and utilization data to obtain the load change trend. The load change trend can be expressed by a trend quantification value. The sign and magnitude of this quantification value characterize the direction and strength of load changes in the target hardware unit in the short term. For example, a large positive value indicates a rapid increase in load, a large negative value indicates a rapid decrease in load, and a value close to zero indicates a stable load.
[0030] Step 104: Obtain the judgment result of whether the load change trend meets the preset control conditions.
[0031] The preset control conditions are pre-defined criteria used to trigger operating mode switching. They are used to identify whether a significant change in load is imminent, and thus determine whether a switch in operating mode is necessary. The system service module at the electronic device application layer determines whether the load change trend meets the preset control conditions if it indicates a significant change in load in the near future (including an increase or decrease). If the load change trend indicates a relatively stable load in the near future, it determines whether the load change trend does not meet the preset control conditions. After completing the comparison, the system service module sends the determination result to the BIOS scheduling module through the driver module.
[0032] Step 105: Based on the judgment result and one of the real-time load status and load change trend, determine the operating mode of the target hardware unit; wherein the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
[0033] After receiving the judgment result, if the BIOS scheduling module determines that the load change trend meets the preset control conditions, it indicates that the load is about to fluctuate significantly, and the target hardware unit needs to be switched to the operating mode that matches the predicted trend. Specifically, if the predicted load change trend indicates that the load is about to rise rapidly, the operating mode is switched to the first mode; if the predicted load change trend indicates that the load is about to fall rapidly, the mode is switched to the second mode, thereby achieving advance and precise scheduling of modes.
[0034] If the load change trend does not meet the preset control conditions, it indicates that the load will not fluctuate significantly. In this case, the corresponding operating mode is matched according to the current real-time load status. When the real-time load status is high, the first mode is used; when the real-time load status is low, the second mode is used. During the load stabilization phase, the power consumption-adaptive operating mode is matched according to the current actual load, taking into account both hardware performance and overall power consumption control.
[0035] In one embodiment, the overall power consumption of the system is higher when the target hardware unit is running in the first mode than in the second mode. The first mode is a dual-stress mode, where both the CPU and the target hardware unit run at full load continuously, resulting in high overall power consumption and heat generation. The second mode is a single-stress mode, where only the CPU runs at full load, while the target hardware unit remains in low-load standby mode, resulting in relatively lower overall power consumption and heat generation.
[0036] The solution in this embodiment accurately identifies the real-time load status of the target hardware unit by synchronously collecting three key information types: power consumption, utilization rate, and temperature. At the same time, it predicts the load change trend of the target hardware unit by combining power consumption and utilization rate information. Then, based on whether the predicted load change trend meets the preset control conditions, it determines the operating mode by combining the load change trend and the real-time load status. This achieves accurate and advance scheduling of the operating mode, which not only ensures the stable operation of the target hardware unit under different load scenarios, but also balances performance release and energy consumption control by adapting to the first and second modes with different power consumption, reducing mode switching lag, and improving system response speed and overall operating efficiency.
[0037] In another embodiment of this disclosure, the operating mode of the target hardware unit is determined based on the judgment result, the real-time load status, and the load change trend, including: if the judgment result indicates that the load change trend meets the preset control conditions, the operating mode of the target hardware unit is determined based on the load change trend; if the judgment result indicates that the load change trend does not meet the preset control conditions, the operating mode of the target hardware unit is determined based on the real-time load status.
[0038] If the judgment result indicates that the load change trend meets the preset control conditions, it means that the load of the target hardware unit is about to change significantly. In this case, the operating mode is determined based on the load change trend. By switching to the first or second mode that matches the predicted trend in advance, response lag caused by sudden load changes is avoided, ensuring that the hardware operation is accurately adapted to the load change. If the judgment result indicates that the load change trend does not meet the preset control conditions, it means that the load is currently in a stable state with no obvious change trend. In this case, the operating mode is determined based on the real-time load status. The corresponding operating mode is determined based on the real-time collected power consumption, utilization rate, and temperature information reflecting the actual load situation, ensuring the stability and rationality of hardware operation.
[0039] The solution in this embodiment not only fully leverages the proactive nature of load forecasting to achieve advance scheduling of modes, but also avoids misscheduling caused by forecast deviations by using real-time load status as a safety net, thus balancing the timeliness and reliability of operating mode adjustments and further improving the operating efficiency and energy consumption control of the target hardware unit.
[0040] In another embodiment of this disclosure, the real-time load state of the target hardware unit is determined based on power consumption information, utilization information, and temperature information. Specifically, this can be achieved through the following technical means: determining a high load factor and a low load factor based on power consumption information, utilization information, and temperature information; determining the real-time load state of the target hardware unit as a high load state in response to the high load factor meeting a preset high load condition; and determining the real-time load state of the target hardware unit as a low load state in response to the low load factor meeting a preset low load condition.
[0041] First, the collected raw power consumption, utilization, and temperature data undergo unified data cleaning and preprocessing. For the raw sampled data, a moving average process using a sliding window is employed, for example, using five sampling points within one second as a window for smoothing calculations. This filters out instantaneous spikes and random noise, ensuring the stability of the data used in subsequent multi-dimensional fusion analysis. The window size can be adaptively adjusted according to the degree of load change; the window is shortened to improve response speed when the load fluctuates drastically, and expanded to enhance data smoothness when the operation is stable. Simultaneously, combining statistical methods and historical data distribution characteristics, the system automatically identifies and removes anomalous sampling points such as sudden changes and distortions, avoiding the accidental deletion of valid key data. Furthermore, the system introduces a real-time anomaly marking mechanism. Identified anomalous data is not directly discarded but marked and its weight in the calculation is dynamically reduced to prevent distortion of subsequent prediction results due to direct data loss.
[0042] After the above cleaning and pretreatment process, stable and reliable power consumption, utilization rate, and temperature data are used to calculate the corresponding high load factor and low load factor, respectively. The specific implementation method is as follows: When calculating the high load factor, power consumption, utilization, and temperature information are divided by the corresponding reference upper limit value to perform normalization processing, converting each indicator into a standardized score between 0 and 1, eliminating the influence of dimensional differences and numerical ranges between different indicators. Each normalized score is then multiplied by a preset weighting coefficient, and the results are summed to obtain the final high load factor. The weighting coefficients can be flexibly configured according to the actual application scenario to highlight the importance of different indicators in high load determination. For example, in gaming scenarios, utilization best reflects the real-time hardware load pressure and can be given higher weight; in rendering scenarios, power consumption is more strongly correlated with load intensity and can be given higher weight, ensuring that the determination logic adapts to different usage needs.
[0043] The calculated high load coefficient is compared with the preset high load conditions. The preset high load conditions can take the form of a single judgment threshold (high load judgment threshold), a numerical range, or a combination of multiple conditions to adapt to the judgment accuracy requirements in different scenarios. The high load coefficient is used to comprehensively characterize the current overall load intensity and heat pressure of the target hardware unit. If the high load coefficient meets the preset high load conditions, it indicates that the current power consumption, utilization rate, and temperature of the target hardware unit are all at a high level, and the overall load pressure reaches the high load standard, then the target hardware unit is judged to be in a high load state.
[0044] Specifically, this can be achieved through the following formulas (1)-(2): (1) (2) in, Characterized by a high load factor; Characterizes power consumption information, Characterization utilization information Characterize temperature information, , , These are the reference upper limits for power consumption, utilization rate, and temperature, respectively. , , These are the weighting coefficients for the corresponding indicators; The threshold for determining high load is typically between 0.8 and 1; when Greater than or equal to The time signature meets the preset high load judgment threshold. A value of 1 indicates that the current real-time load status is high; otherwise... A value of 0 indicates that the high load state has not been reached.
[0045] When calculating the low load coefficient, the collected power consumption, utilization, and temperature information must first be normalized. Then, through reverse conversion, each indicator is converted into a corresponding idle degree. This is a numerical representation of the hardware's idle state; the higher the idle degree, the more pronounced the hardware's idle state. Subsequently, the idle degree corresponding to each indicator is multiplied by its preset weighting coefficient, and all weighted idle degrees are summed to obtain the final low load coefficient. Consistent with the weighting configuration logic of the high load coefficient, the weighting coefficients of each indicator in the low load determination process can be flexibly adjusted according to the needs of the actual usage scenario. For example, the weighting coefficient corresponding to utilization can be appropriately increased to ensure that as long as the utilization is at a sufficiently low level, the overall idle degree score can be significantly improved, thereby strongly driving the system to determine that the hardware is in a low load state. This effectively avoids the low load determination delay caused by temperature changes, ensuring the timeliness and accuracy of low load state determination, while also considering the stability of system operation.
[0046] The calculated low-load coefficient is compared with the preset low-load conditions. The preset low-load conditions can also take the form of a single judgment threshold (low-load judgment threshold), a numerical range, or a combination of multiple conditions. The low-load coefficient is used to comprehensively characterize the overall idle level of the target hardware unit. If the low-load coefficient meets the preset low-load conditions, it indicates that the target hardware unit's current power consumption and utilization are at a low level, its idle level is large, and its heat dissipation pressure has been effectively relieved. Therefore, the target hardware unit is determined to be in a low-load state.
[0047] The solution in this embodiment judges the load based on three key indicators: power consumption, utilization rate, and temperature. By using a multi-dimensional fusion scoring method, it avoids misjudgment caused by sudden changes in a single indicator. It not only ensures the accuracy of load judgment, but also flexibly adapts the weight configuration according to the scenario, and finally achieves accurate and reliable identification of the real-time load status of the target hardware unit.
[0048] In another embodiment of this disclosure, a low load factor is determined based on power consumption information, utilization information, and temperature information. Specifically, this can be achieved through the following technical means: obtaining a temperature threshold; determining a temperature constraint penalty term based on the temperature information and the temperature threshold; and determining a low load factor based on the power consumption information, utilization information, temperature information, and temperature constraint penalty term.
[0049] The temperature threshold is a pre-set critical temperature for heat dissipation mitigation, such as 60°C, used to determine whether the hardware still has high residual heat. When the temperature exceeds this threshold, the hardware is considered to be in a high-temperature state, the temperature constraint penalty takes effect, and the low-load coefficient is reduced to raise the threshold for low-load judgment and prevent the system from being mistakenly judged as low-load when the load decreases but the temperature has not yet fully dropped. When the temperature is below this threshold, the temperature constraint penalty does not take effect, and there is no longer a constraint on the low-load coefficient.
[0050] Specifically, this can be achieved through the following formulas (3)-(4): (3) (4) in, Indicates a low load factor; Characterized by a high load factor; Characterizes power consumption information, Characterizes utilization information; This represents the hardware's nominal typical power consumption. This is the maximum safe operating temperature for the hardware. , , The weighting coefficients for the corresponding indicators are usually set as follows: > > Because utilization rate is the most direct indicator of whether hardware is in an idle state; Temperature threshold; I is the temperature constraint penalty term; I is the indicator function, when > The value is 1 if the condition is met, and 0 otherwise. This is the threshold for determining low load. When L(t) is greater than or equal to... hour, A value of 1 indicates that the current state is low load; otherwise, a value of 0 indicates that the low load state has not been reached.
[0051] This embodiment introduces a temperature constraint penalty term when calculating the low load factor, appropriately raising the judgment threshold when the temperature is high to avoid false judgments. Simultaneously, by increasing the weighting of utilization and power consumption, it ensures that even if the temperature has not fully dropped, the low load judgment condition can still be met when the system is truly idle and both power consumption and utilization have significantly decreased. Therefore, while ensuring system stability, it avoids slow exit from high load mode due to slow temperature cooling, achieving timely and accurate determination of low load states.
[0052] In another embodiment of this disclosure, the method further includes: determining preset high load conditions and preset low load conditions based on the thermal management scheduling mode of the electronic device in which the target hardware unit is located.
[0053] Thermal management scheduling modes for electronic devices typically include performance-priority mode, balanced mode, temperature-priority mode, silent mode, and energy-saving mode. Different modes exhibit differences in load judgment sensitivity, temperature control constraints, and response strategies. This solution pre-configures corresponding preset high-load and low-load conditions for each thermal management scheduling mode based on historical operating data and scenario calibration, and stores these conditions in the form of mapping tables, configuration files, or function relationships. For example, a more lenient judgment threshold can be used in performance mode to quickly enter a high-load state; a stricter threshold is used in temperature-priority mode to prevent rapid temperature rise. After the electronic device starts running, the system first determines the currently active thermal management scheduling mode, and then matches and reads the corresponding preset high-load and low-load conditions from the pre-established configuration mapping relationship according to this mode, achieving dynamic adaptation of the load judgment rules.
[0054] The solution in this embodiment links the load judgment threshold with the thermal management scheduling mode, so that the judgment conditions for high load and low load can be adaptively adjusted according to the current performance of the equipment and the temperature control strategy. This ensures performance release while improving heat dissipation safety, and makes the mode switching more in line with the actual operating needs of the whole system.
[0055] In another embodiment of this disclosure, the load change trend of the target hardware unit is determined based on power consumption information and utilization information. Specifically, this can be achieved through the following technical means: determining the initial power consumption change rate based on power consumption information; determining the initial utilization change rate based on utilization information; performing smoothing filtering on the initial power consumption change rate and the initial utilization change rate to obtain the smoothed power consumption change rate and the smoothed utilization change rate; determining the current steady-state load bias term of the target hardware unit; and obtaining the load change trend based on the smoothed power consumption change rate, the smoothed utilization change rate, and the steady-state load bias term.
[0056] First, differential calculations are performed based on power consumption information at continuous time points to obtain the initial power consumption change rate, which characterizes the speed at which power consumption increases or decreases. Similarly, differential calculations are performed based on utilization information at continuous time points to obtain the initial utilization change rate, which characterizes the speed at which the computational load increases or decreases. Subsequently, the initial power consumption change rate and the initial utilization change rate are smoothed using filtering methods such as exponential moving average (EMA) to filter out instantaneous jitter and random interference, retaining the true and continuous load change trend, resulting in smoothed power consumption change rate and smoothed utilization change rate, avoiding prediction distortion caused by instantaneous data fluctuations. The current steady-state load bias term of the target hardware unit is used to adaptively correct the prediction results based on the current actual load conditions, making the predicted trend more consistent with the current system operating state and avoiding misjudgments against the trend.
[0057] Finally, the smoothed power consumption change rate, smoothed utilization change rate, and steady-state load bias term are weighted and fused to obtain the load change trend. This load change trend is used to comprehensively characterize the load's future short-term change trend and intensity. Positive values indicate that the load tends to increase, and the larger the value, the more drastic the increase; negative values indicate that the load tends to decrease, and the larger the value, the more drastic the decrease.
[0058] The solution in this embodiment extracts, smooths, and adaptively corrects the load change rate to achieve advanced prediction of the load change trend of the target hardware unit. It can complete the mode decision in advance before the actual load changes, effectively improve the system response speed, reduce control lag, and improve system performance and user experience.
[0059] In another embodiment of this disclosure, the current steady-state load bias term of the target hardware unit is determined by the following technical means: obtaining the real-time load state of the target hardware unit and its corresponding preset bias coefficient; and determining the current steady-state load bias term of the target hardware unit based on the real-time load state and its corresponding preset bias coefficient.
[0060] As described in the foregoing embodiments, the real-time load state includes a low load state and a non-low load state (high load state). For different real-time load states, corresponding preset bias coefficients are configured to apply adaptive directional corrections to the predicted load score. The preset bias coefficients are pre-set correction parameters whose function is to reasonably guide the trend prediction results based on the current steady-state condition of the system, avoiding misjudgments that contradict the current state.
[0061] Taking a low-load state as an example, if the current state is determined to be low-load, a corresponding bias coefficient is selected to make the steady-state load bias term output a negative value. This automatically lowers the predicted load score during prediction, making it easier to determine the overall trend as moving towards low load. If the current state is not low-load, a corresponding bias coefficient is selected to make the steady-state load bias term output a positive value, thereby raising the predicted load score and making it easier to determine the trend as moving towards high load. Through this bias correction method that is linked to the real-time load status, the prediction results can better match the current operating state of the system, improving the rationality and reliability of load change trend prediction.
[0062] The solution in this embodiment introduces a bias correction mechanism that is linked to the real-time load status, enabling the predicted load score to adapt to the current steady state of system operation, thereby further improving the rationality and reliability of trend prediction and avoiding prediction deviations under extreme operating conditions.
[0063] In another embodiment of this disclosure, the preset control conditions include a positive prediction threshold and a negative prediction threshold. Accordingly, the judgment result of whether the load change trend meets the preset control conditions can be obtained by the following technical means: comparing the load change trend with the positive prediction threshold and the negative prediction threshold respectively to obtain the comparison result; if the comparison result is that the load change trend is greater than or equal to the positive prediction threshold, the judgment result is determined to be that the load change trend meets the preset control conditions; if the comparison result is that the load change trend is less than or equal to the negative prediction threshold, the judgment result is determined to be that the load change trend does not meet the preset control conditions.
[0064] This embodiment configures independent positive and negative prediction thresholds to identify the strength of load change trends. Trend identification is achieved by comparing the trend quantization value representing the load change trend with the two types of thresholds: the positive prediction threshold is used to identify a continuously increasing load change trend. Only when the trend quantization value is greater than or equal to the threshold is the load change trend determined to meet the preset control conditions, indicating that the target hardware unit will enter a high-load condition in the short term; the negative prediction threshold is used to identify a continuously decreasing load change trend. Only when the trend quantization value is less than or equal to the threshold is the load change trend determined to meet the preset control conditions, indicating that the target hardware unit will enter a low-load condition in the short term.
[0065] This embodiment uses a two-way threshold decision mechanism to clearly distinguish between three types of operating conditions: load change trend is about to rise, about to fall, and stable with no obvious fluctuations. The prediction logic is clear and the decision results are stable, which can provide a reliable and clear basis for early switching of hardware operating modes.
[0066] Specifically, the above embodiments can be implemented using the following formulas (5)-(7): (5) (6) (7) in, The trend quantification value represents the load change trend and is used to comprehensively characterize the direction and intensity of load change of the target hardware unit in the near future. A positive value indicates that the load tends to increase, and a negative value indicates that the load tends to decrease. This represents the initial rate of change of power consumption, reflecting the amplitude and direction of power consumption fluctuations per unit time. This represents the rate of change of initial utilization, reflecting the amplitude and direction of fluctuations in hardware load per unit time. This indicates smoothing filtering processing; , These are the normalization coefficients for smoothing the rate of change in power consumption and the rate of change in smoothing the rate of change in utilization, respectively. , These are the weighting coefficients for smoothing the rate of change in power consumption and the rate of change in smoothing the rate of change in utilization, respectively. These are the weighting coefficients for the steady-state load bias term, used to control the strength of the adaptive correction; The low load factor at time t, with a value of 0 or 1; This is a steady-state load bias term, used to adaptively correct the trend quantization value based on the current real-time load status, so that the predicted trend fits the current operating conditions of the system. This is the positive prediction threshold, used to determine whether the upward trend in load is strong enough; H(t) is a negative prediction threshold used to determine whether the load decline trend is strong enough; H(t) is a high load prediction flag, which means that the trend quantification value exceeds the positive threshold, identifies the upward load change trend, and predicts that the equipment is about to enter a high load; otherwise, it is 0. L(t) is a low load prediction flag, which means that the trend quantification value is lower than the negative threshold, identifies the downward load change trend, and predicts that the equipment is about to enter a low load; otherwise, it is 0.
[0067] The following is an illustration through a specific application scenario, in which the target hardware unit is a GPU.
[0068] To verify the practical application effect of this solution, high-load, medium-high-load, and comprehensive graphics performance scenarios were selected as test benchmarks. The system performance was compared under three typical thermal management modes: Quiet, Balance, and Performance. During the testing process, multiple rounds of sampling verification were performed on the solution disclosed herein and related technologies, ensuring sufficient intervals between adjacent tests to eliminate the interference of temperature inertia on the test results.
[0069] Scenario 1: High load scenario (load increases sharply): When the system initiates a high-load task, GPU utilization and power consumption rise rapidly, with a large positive rate of change. After exponential moving average (EMA) smoothing filtering, the predicted trend quantization value Trend(t) continues to increase positively and exceeds the positive prediction threshold. At this point, even if the real-time load judgment conditions based on power consumption, utilization, and absolute temperature values are not yet met, the system has already identified the impending high load through the prediction mechanism and quickly triggered the operation mode switching preparation process. This proactive prediction mechanism effectively shortens the response latency, enabling performance configuration adjustments to be completed before the load peak arrives, avoiding operational stuttering due to mode switching lag.
[0070] Scenario 2: Sudden Load Drop (Rapid Load Decrease): After a high-load task ends, hardware utilization and power consumption drop rapidly, with the rate of change turning into a large negative value. The predicted trend quantification value Trend(t) continues to decrease negatively and reaches the negative prediction threshold. Based on this, the system determines that the load will remain at a low level and promptly switches to a low-power operating mode. This logic effectively avoids the problem of the system maintaining high power consumption for a long time due to the lag in temperature drop in traditional solutions, reducing unnecessary energy consumption while ensuring the overall temperature control safety of the system.
[0071] Actual performance results: In both silent and balanced modes, multiple sets of graphics performance benchmark tests were conducted on the relevant technical solutions and this solution. Each scenario was tested multiple times and the average value was taken. At the same time, sufficient test intervals were ensured to allow the hardware temperature to completely drop, ensuring that the test results are objective and reliable.
[0072] Test data shows that in silent mode, this solution outperforms related technologies in terms of graphics score, CPU score, and average frame rate across various scenarios. In one representative scenario, the average graphics score increased from approximately 6516 to 6734, and the frame rate increased from 30.17 FPS to 31.18 FPS. In another typical scenario, the graphics score increased from 2641 to 2761, and the frame rate increased from 26.41 FPS to 27.61 FPS. In comprehensive scenarios, the CPU score increased from 10804 to 11189, and the graphics score increased from 15709 to 15958, with overall performance improvements of approximately 3%–4% compared to existing solutions. In balanced mode, this solution also achieved comprehensive performance improvements: in one representative scenario, the graphics score increased from 8573 to 8812, and the frame rate increased from 39.69 FPS to 40.80 FPS; in another typical scenario, the graphics score increased from 3463 to 3548, and the frame rate increased from 34.63 FPS to 35.49 FPS; in comprehensive scenarios, the CPU score increased from 13928 to 14219, with overall performance improving by approximately 2% compared to existing solutions. These data indicate that this solution, by proactively predicting load changes and adjusting the operating mode in a timely manner, can stably improve system performance under different temperature control strategies. Furthermore, this test only used a simplified algorithm example; further optimization of parameter configuration and scheduling strategies could lead to further performance improvements.
[0073] A second aspect of this disclosure provides an operating mode determination apparatus, such as... Figure 3As shown, the device includes: an information acquisition module 401, used to acquire power consumption information, utilization information, and temperature information of the target hardware unit; a status determination module 402, used to determine the real-time load status of the target hardware unit based on the power consumption information, utilization information, and temperature information; the status determination module 402 is also used to determine the load change trend of the target hardware unit based on the power consumption information and utilization information; an operating mode determination module 403, used to obtain a judgment result on whether the load change trend meets preset control conditions; and to determine the operating mode of the target hardware unit based on the judgment result, and one of the real-time load status and the load change trend; wherein the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
[0074] The operating mode determination device in this embodiment and the aforementioned operating mode determination method are derived from the same technical concept. For the specific implementation method, please refer to the implementation method of the operating mode determination method, which will not be repeated here.
[0075] According to embodiments of this disclosure, this disclosure also provides an electronic device and a readable storage medium.
[0076] Figure 4 A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device 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 electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, 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 present disclosure described and / or claimed herein.
[0077] like Figure 4 As shown, device 800 includes a computing unit 801, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 802 or a computer program loaded from storage unit 808 into random access memory (RAM) 803. RAM 803 may also store various programs and data required for the operation of device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via bus 804. Input / output (I / O) interface 805 is also connected to bus 804.
[0078] Multiple components in device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of monitors, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0079] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 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 computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as the operating mode determination method. For example, in some embodiments, the operating mode determination method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the operating mode determination method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the operating mode determination method by any other suitable means (e.g., by means of firmware).
[0080] Various embodiments of the 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), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments 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 transferring data and instructions to the storage system, the at least one input device, and the at least one output device.
[0081] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0082] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. 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.
[0083] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0084] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with 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 of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.
[0085] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.
[0086] 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 disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.
[0087] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means two or more, unless otherwise explicitly specified.
[0088] The above description is merely a specific embodiment of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.
Claims
1. A method for determining an operating mode, characterized in that, The method includes: Collect power consumption, utilization, and temperature information of the target hardware unit; Based on the power consumption information, utilization information, and temperature information, the real-time load status of the target hardware unit is determined; Based on the power consumption information and the utilization information, the load change trend of the target hardware unit is determined; Obtain a judgment result on whether the load change trend meets the preset control conditions; Based on the judgment result, and one of the real-time load status and the load change trend, the operating mode of the target hardware unit is determined; wherein the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
2. The method for determining the operating mode according to claim 1, characterized in that, The step of determining the operating mode of the target hardware unit based on the judgment result, the real-time load status, and the load change trend includes: If the judgment result indicates that the load change trend meets the preset control conditions, the operating mode of the target hardware unit is determined based on the load change trend; If the judgment result indicates that the load change trend does not meet the preset control conditions, the operating mode of the target hardware unit is determined based on the real-time load status.
3. The method for determining the operating mode according to claim 1, characterized in that, Determining the real-time load status of the target hardware unit based on the power consumption information, utilization information, and temperature information includes: Based on the power consumption information, the utilization information, and the temperature information, the high load factor and the low load factor are determined. In response to the high load factor satisfying the preset high load condition, the real-time load state of the target hardware unit is determined to be a high load state; In response to the low load factor satisfying the preset low load condition, the real-time load state of the target hardware unit is determined to be a low load state.
4. The method for determining the operating mode according to claim 1, characterized in that, Determining the load change trend of the target hardware unit based on the power consumption information and the utilization information includes: Based on the power consumption information, determine the initial power consumption change rate; Based on the utilization information, determine the initial utilization rate change rate; The initial power consumption change rate and the initial utilization rate change rate are respectively subjected to smoothing filtering to obtain the smoothed power consumption change rate and the smoothed utilization rate change rate; Determine the current steady-state load bias term of the target hardware unit; Based on the smoothed power consumption change rate, the smoothed utilization change rate, and the steady-state load bias term, the load change trend is obtained.
5. The method for determining the operating mode according to claim 3, characterized in that, Determining the low load factor based on the power consumption information, the utilization information, and the temperature information includes: Obtain the temperature threshold; Based on the temperature information and the temperature threshold, a temperature constraint penalty term is determined; Based on the power consumption information, utilization information, temperature information, and the temperature constraint penalty term, a low load factor is determined.
6. The method for determining the operating mode according to claim 4, characterized in that, Determining the current steady-state load bias term of the target hardware unit includes: Obtain the real-time load status of the target hardware unit and its corresponding preset bias coefficient; Based on the real-time load status and its corresponding preset bias coefficient, the current steady-state load bias term of the target hardware unit is determined.
7. The method for determining the operating mode according to claim 1, characterized in that, The preset control conditions include a positive prediction threshold and a negative prediction threshold; The determination result of whether the load change trend meets the preset control conditions includes: The load change trend is compared with the positive prediction threshold and the negative prediction threshold respectively to obtain the comparison results; If the comparison result is that the load change trend is greater than or equal to the positive prediction threshold, the judgment result is determined to be that the load change trend meets the preset control conditions; If the comparison result is that the load change trend is less than or equal to the negative prediction threshold, the judgment result is determined to be that the load change trend does not meet the preset control conditions.
8. The method for determining the operating mode according to claim 3, characterized in that, The method further includes: Based on the thermal management scheduling mode of the electronic device in which the target hardware unit is located, the preset high load condition and the preset low load condition are determined.
9. An operating mode determination device, characterized in that, The device includes: The information acquisition module is used to collect power consumption information, utilization information and temperature information of the target hardware unit; The status determination module is used to determine the real-time load status of the target hardware unit based on the power consumption information, utilization information and temperature information. The state determination module is further configured to determine the load change trend of the target hardware unit based on the power consumption information and the utilization information. The operating mode determination module is used to obtain a judgment result on whether the load change trend meets the preset control conditions; based on the judgment result, and one of the real-time load status and the load change trend, the operating mode of the target hardware unit is determined; wherein, the operating mode includes a first mode and a second mode, and the power consumption of the target hardware unit is different in the first mode and the second mode.
10. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.
