Graphics processor and performance control method thereof, storage medium and terminal equipment

Abstract

The invention provides a graphics processor and a performance control method thereof, a storage medium and terminal equipment. The performance control method of the graphics processor comprises the following steps: acquiring a plurality of operation parameters of the graphics processor and the utilization rate of each functional module in the graphics processor; predicting the power consumption of the graphics processor by using the plurality of operation parameters and the utilization rate of each functional module; adjusting the frequency of the graphics processor according to the power consumption of the graphics processor. According to the method, the accuracy of power consumption prediction of the graphics processor and the performance of the graphics processor are improved by using the running parameters which more comprehensively and more truly reflect the running condition of the graphics processor.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a graphics processor and its performance control method, storage medium, and terminal device. Background Technology

[0002] With the increasing power consumption of modern Graphics Processing Units (GPUs), in many cases, if left unchecked, the chip may exceed the system's thermal design power, leading to system failure. To balance power consumption and performance, systems and drivers need to acquire GPU power consumption data in real time to adjust its frequency and supply voltage as needed to meet different usage scenarios.

[0003] Existing methods for obtaining the real-time power consumption of a GPU fall into two categories: those using external circuitry and those using internal instructions to estimate power consumption. Generally, the first method is more accurate than the second. The first method requires obtaining the real-time current and supply voltage of the graphics processor's power supply and then calculating the power consumption. The second method typically estimates power consumption by modeling using internal chip instructions.

[0004] However, existing solutions for obtaining real-time GPU power consumption using external circuits require extensive specialized equipment for accurate real-time measurement, making them impractical for mass-produced chips. Current mass-production solutions also suffer from inaccuracies or inability to simultaneously adapt to different scenarios, such as high and low power consumption. While existing solutions for predicting GPU power consumption using internal instructions have established power estimation models, these models are not sufficiently accurate. Summary of the Invention

[0005] This application provides a power consumption prediction scheme for graphics processing, which can improve the accuracy of power consumption prediction and thus ensure the working performance of the graphics processor.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] In a first aspect, this application provides a performance control method for a graphics processing unit (GPU), comprising: acquiring multiple operating parameters of the GPU and the utilization rate of each functional module within the GPU; predicting the power consumption of the GPU using the multiple operating parameters and the utilization rate of each functional module; and adjusting the frequency of the GPU based on the power consumption of the GPU.

[0008] Optionally, the utilization rate of each functional module is the ratio of the working time and idle time of each functional module within a preset time interval.

[0009] Optionally, determining the power consumption of the graphics processor using the plurality of operating parameters and the utilization rate of each functional module includes: predicting the weight corresponding to the utilization rate of each functional module using the plurality of operating parameters, and determining the total utilization rate of the graphics processor based on the predicted weights and the utilization rate of each functional module; and determining the power consumption of the graphics processor based on the total utilization rate of the graphics processor and the plurality of operating parameters.

[0010] Optionally, the process data includes leakage current, and the acquisition of multiple operating parameters of the graphics processor includes: acquiring the temperature; determining the leakage current based on the temperature and a first mapping relationship, wherein the first mapping relationship includes each temperature and its corresponding leakage current.

[0011] Optionally, the plurality of operating parameters may also include power supply voltage, temperature, clock frequency, and process data.

[0012] Optionally, the process data includes leakage current, and the power consumption of the graphics processor is calculated using the following formula: ,in, This indicates the power consumption of the graphics processor. This represents the total utilization rate. Indicates the load capacitance. This indicates the power supply voltage. Indicates clock frequency. Indicates the short-circuit time. Indicates short-circuit current. This indicates leakage current.

[0013] Optionally, before acquiring the multiple operating parameters of the graphics processor, the method further includes: locking the multiple operating parameters in response to the count reaching a target value; storing the multiple operating parameters in a memory; and recounting.

[0014] Optionally, predicting the power consumption of the graphics processor using the plurality of operating parameters includes: inputting the plurality of operating parameters into a power consumption prediction model, wherein the power consumption prediction model is used to output the power consumption of the graphics processor.

[0015] Secondly, this application also discloses a graphics processor, which includes: at least one functional module; a utilization counter for generating the utilization rate of each functional module; a processing module for acquiring multiple operating parameters of the graphics processor and predicting the power consumption of the graphics processor using the multiple operating parameters and the utilization rate of each functional module; and a control module for adjusting the frequency of the graphics processor according to the power consumption of the graphics processor.

[0016] Thirdly, this application provides a terminal device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the performance control method of the graphics processor, or includes the graphics processor.

[0017] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to perform the steps of the performance control method of the graphics processor.

[0018] Fifthly, a computer program product is provided, on which a computer program is stored, the computer program being executed by a processor to perform the steps of the method provided in the first aspect.

[0019] Sixthly, embodiments of this application also provide a chip that stores a computer program, which, when executed by the chip, implements the steps of the above-described method.

[0020] In a seventh aspect, embodiments of this application also provide a system chip for use in a terminal. The chip system includes at least one processor and an interface circuit. The interface circuit and the at least one processor are interconnected via a line. The at least one processor is used to execute instructions to perform the steps of the above-described method.

[0021] Compared with the prior art, the technical solution of this application has the following beneficial effects:

[0022] In this application's technical solution, multiple operating parameters of the graphics processing unit (GPU) are obtained, including the utilization rate of each functional module within the GPU. These parameters are then used to predict the GPU's power consumption. The GPU's frequency is adjusted based on its power consumption. This solution improves the accuracy of power consumption prediction by obtaining the utilization rate of each functional module, reflecting the impact of key functional modules on power consumption, and combining this utilization rate with other operating parameters. In other words, it improves the accuracy of GPU power consumption prediction by using more comprehensive and accurate operating parameters that reflect the GPU's operating status. Furthermore, adjusting the GPU's frequency in conjunction with its power consumption ensures that the GPU's frequency matches its power consumption, thereby guaranteeing the GPU's performance.

[0023] Furthermore, in the technical solution of this application, multiple operating parameters also include power supply voltage, temperature, clock frequency, and process data. These multiple operating parameters, along with the utilization rate of each functional module, are input into a power consumption prediction model, which outputs the power consumption of the graphics processor. Based on the utilization rate of each functional module, clock frequency, power supply voltage, and process data, this application obtains a power consumption prediction model through supervised learning training (including but not limited to neural networks, linear regression, etc.). This model can estimate the power consumption of the graphics processor with high accuracy and adaptability through internal logic without requiring external devices.

[0024] Furthermore, under different operating environments, the impact of each functional module on the total utilization of the graphics processor varies. Since operating parameters can reflect the operating environment, this application uses multiple operating parameters to predict the weight corresponding to the utilization of each functional module, and determines the total utilization of the graphics processor based on the predicted weights and the utilization of each functional module. This ensures the accuracy of the total utilization of the graphics processor, conforms to the actual situation, and thus ensures the accuracy of the power consumption prediction of the graphics processor.

[0025] Furthermore, in response to the count reaching the target value, the plurality of operating parameters are locked; the plurality of operating parameters are stored in the memory, and the count is restarted. This application's technical solution, by locking operating parameters through counting and uniformly storing them in memory, ensures that the various operating parameters obtained during subsequent power consumption prediction are sampled at the same time and at specified intervals. Each operating parameter can more accurately reflect the actual operating status of the graphics processor, thereby ensuring the accuracy of subsequent power consumption prediction. Attached Figure Description

[0026] Figure 1 This is a flowchart of a performance control method for a graphics processor provided in an embodiment of this application;

[0027] Figure 2 This is a schematic diagram of an application scenario provided by an embodiment of this application;

[0028] Figure 3 This is a flowchart illustrating the construction of a power consumption prediction model provided in an embodiment of this application;

[0029] Figure 4 This is a structural diagram of a graphics processor provided in an embodiment of this application. Detailed Implementation

[0030] As described in the background section, existing solutions for obtaining real-time GPU power consumption using external circuits require a large number of specialized instruments to achieve accurate real-time measurement, making them impractical for mass-produced chips. Current mass-production solutions also suffer from insufficient accuracy or inability to simultaneously adapt to different scenarios such as high and low power consumption. Existing solutions for predicting GPU power consumption using internal instructions, while establishing power estimation models, suffer from inaccuracies.

[0031] This application's technical solution reflects the impact of key functional modules on power consumption by obtaining the utilization rate of each functional module. This utilization rate, combined with other operating parameters, contributes to the prediction of graphics processor power consumption. In other words, it improves the accuracy of graphics processor power consumption prediction by using more comprehensive and realistic operating parameters that reflect the graphics processor's operating status. Furthermore, the graphics processor's frequency is adjusted based on its power consumption to match the processor's power consumption, thereby ensuring the graphics processor's performance.

[0032] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0033] See Figure 1 The performance control method for the graphics processor is executed by the terminal device, which can be the terminal device itself, or a chip or chip module inside the terminal device. It is understood that the performance control method for the graphics processor can also be executed by any other suitable entity.

[0034] Specifically, the performance control method for a graphics processor may include the following steps:

[0035] Step 101: Obtain multiple operating parameters of the graphics processor and the utilization rate of each functional module within the graphics processor.

[0036] Step 102: Predict the power consumption of the graphics processor using multiple operating parameters and the utilization rate of each functional module.

[0037] Step 103: Adjust the frequency of the graphics processor according to its power consumption.

[0038] It is understood that, in practical implementation, each step of the above method can be implemented using software programs, which run in the processor integrated within the chip or chip module. This method can also be implemented using a combination of software and hardware; this application does not impose any restrictions.

[0039] In practice, a graphics processing unit (GPU) operates with multiple parameters, such as its supply voltage, temperature, clock frequency, load, electrical characteristics, and manufacturing process data. The manufacturing process data primarily refers to the static leakage current (SIDD) measured during the post-packaging testing of the GPU chip. Electrical characteristics can include short-circuit current and short-circuit time. The load can specifically be the load capacitance.

[0040] Furthermore, the various functional modules within a graphics processor may be in different operating states; for example, some modules may be running while others are idle. The operating state of these modules affects the power consumption of the graphics processor. Therefore, in this embodiment, the utilization rate of each functional module within the graphics processor is also obtained to participate in power consumption prediction.

[0041] Furthermore, the utilization rate of each functional module is the ratio of the working time and idle time of that functional module within a preset time interval.

[0042] It should be noted that the specific value of the preset time interval can be adaptively set according to the actual application scenario, and this application does not impose any restrictions on it.

[0043] In the specific implementation of step 101, multiple operating parameters of the graphics processor are obtained. Specifically, if it is necessary to predict the power consumption of a certain functional module within the graphics processor, multiple operating parameters of that functional module can be obtained.

[0044] For details, please refer to Figure 2 The utilization counter 201 is an internal device of the graphics processing unit (GPU), and it corresponds one-to-one with a functional module within the GPU. The utilization counter 201 can count the utilization rate of the corresponding functional module. Therefore, the utilization rate of each functional module can be obtained from the utilization counter 201.

[0045] Other operating parameters, such as the supply voltage (or core voltage), can be acquired through the integrated voltage sensing module in the graphics processor (GPU); temperature can be detected by the temperature sensor in the GPU; and clock frequency can be acquired through the frequency acquisition module in the GPU. Electrical characteristic parameters, such as short-circuit current and short-circuit time, as well as the load, can be obtained from the GPU's design parameters.

[0046] For process data, such as static leakage current, it can be obtained by applying different voltages to the graphics processing unit (GPU) chip under specific temperature and humidity conditions when it is not in operation, and testing the current under these conditions. This process data is recorded in the GPU and can be retrieved when it is necessary to predict the GPU's power consumption.

[0047] In one specific embodiment, the process data includes leakage current, which is temperature-dependent and may vary at different temperatures. Therefore, the leakage current can be determined based on the temperature after temperature acquisition, thereby ensuring the accuracy of the leakage current reading.

[0048] Specifically, the temperature is collected; the leakage current is determined based on the temperature and a first mapping relationship, which includes each temperature and its corresponding leakage current. This first mapping relationship is recorded in the graphics processor and is invoked when the leakage current needs to be determined.

[0049] In the specific implementation of step 102, the power consumption of the graphics processor is predicted by using multiple operating parameters and the utilization rate of each functional module.

[0050] Continue to refer to Figure 2 Multiple operating parameters and the utilization rate of each functional module are input into the power consumption prediction model 202, and the power consumption prediction model 202 outputs the power consumption of the graphics processor.

[0051] In this embodiment, the power consumption prediction model 202 can be obtained by fitting and training the training data through supervised learning methods such as neural networks or linear regression.

[0052] In a non-limiting embodiment, the impact of each functional module on the total utilization of the graphics processor varies under different operating environments. Since operating parameters can reflect the operating environment, this embodiment uses multiple operating parameters to predict the weight corresponding to the utilization of each functional module, and determines the total utilization of the graphics processor based on the predicted weights and the utilization of each functional module. This ensures the accuracy of the total utilization of the graphics processor, conforms to the actual situation, and thus ensures the accuracy of the power consumption prediction of the graphics processor.

[0053] In practice, the power consumption prediction model 202 uses multiple operating parameters to predict the weights corresponding to the utilization rates of each functional module, and determines the total utilization rate of the graphics processor based on the predicted weights and the utilization rates of each functional module.

[0054] For example, with Figure 2 Taking the scenario shown as an example, the utilization rates of the n functional modules counted by utilization counter 201 are as follows: , … The power consumption prediction model 202 uses multiple operating parameters to predict the utilization rate of each functional module, with corresponding weights w1, w2, ..., w... n So, the total utilization of the graphics processor =w1× +w2× +…+w n × .

[0055] Furthermore, the power consumption prediction model 202 determines the power consumption of the graphics processor based on the total utilization of the graphics processor and multiple operating parameters.

[0056] Specifically, the calculated relationship between the power consumption obtained from the power consumption prediction model 202 and various operating parameters is shown below:

[0057] ,

[0058] in, This indicates the power consumption of the graphics processor. This indicates the total utilization of the graphics processor. Indicates the load capacitance. Indicates the supply voltage. Indicates clock frequency. Indicates the short-circuit time. Indicates short-circuit current. This indicates leakage current.

[0059] Specifically, The average number of circuit flips inside a graphics processor chip can be calculated by the utilization rate of each functional module.

[0060] In addition, leakage current and short-circuit current It is related to the temperature in the environment and can reflect the temperature in the operating parameters.

[0061] Please refer to the details. Figure 3 , Figure 3 The training process of the power consumption prediction model is shown.

[0062] In step 301, a first combination of multiple power consumption state groups and the utilization rate of each functional module is established, as well as the power consumption corresponding to the first combination.

[0063] Specifically, each power state group may include supply voltage, temperature, clock frequency, and process data. The first group includes the power state group and the utilization rate of each functional module; that is, the first group includes multiple operating parameters (supply voltage, temperature, clock frequency, load, electrical characteristic parameters, process data, and utilization rate of each functional module). At least one of the multiple operating parameters in different first groups is different.

[0064] The power consumption corresponding to the first combination represents the power consumption of the graphics processor when operating using the various operating parameters and utilization rates of each functional module within the first combination. The specific power consumption corresponding to the first combination can be measured using an external power consumption measuring device.

[0065] In one specific embodiment, a micro-test program library can be pre-established for each functional module (each functional module corresponds to a utilization counter), ensuring that each functional module has micro-test program coverage. Each micro-test program ensures that one or more functional modules are used to varying degrees during operation. The utilization counter value is read at fixed time intervals (e.g., 10ms), and the power consumption of the graphics processor is measured using an external power consumption device to obtain the power consumption corresponding to the first combination.

[0066] In step 302, a power prediction model is trained based on the first combination and its corresponding power consumption. The first combination and its corresponding power consumption are used as training data for the power prediction model.

[0067] Specifically, supervised learning algorithms (including but not limited to neural networks and linear regression) can be used to build power consumption prediction models. These models can then be used to predict the real-time power consumption of graphics processing units (GPUs) during online operation.

[0068] In a non-limiting embodiment, in order to ensure that multiple operating parameters collected by multiple utilization counters and sensors are acquired at the same time, a unified lock-shadow-reset mechanism can be designed inside the graphics processor.

[0069] Specifically, a counter is set within the graphics processor.

[0070] exist Figure 1 Before step 101, i.e., before acquiring multiple operating parameters, multiple operating parameters are locked in response to the counters reaching the target value. These multiple operating parameters are stored in memory (e.g., a shadow register) for use by the internal firmware. While saving the current state through the shadow register, each counter restarts its count, and multiple utilization counters and sensors are re-entered into the counting process for the next time slice.

[0071] The embodiments of this application ensure that the various operating parameters obtained by the firmware are sampled at the same time and at a specified interval through the above method. Each operating parameter can more accurately reflect the actual operating status of the graphics processor, thereby ensuring the accuracy of subsequent power consumption fitting.

[0072] To ensure the performance of the graphics processor, the frequency of the graphics processor can be adaptively adjusted when the power consumption of the graphics processor changes.

[0073] Continue to refer to Figure 1In step 103, the frequency of the graphics processor is adjusted based on the predicted power consumption of the graphics processor. Specifically, since the frequency of the graphics processor is matched with the power supply voltage of the graphics processor, the frequency and power supply voltage of the graphics processor can be adjusted according to the predicted power consumption of the graphics processor. There is a mapping relationship between the frequency and the power supply voltage, which can be preset, and this application does not limit this.

[0074] For example, a first threshold and a second threshold are set, with the second threshold being greater than the first threshold. When the predicted power consumption of the graphics processor is less than or equal to the first threshold, the frequency and supply voltage of the graphics processor are reduced; when the predicted power consumption of the graphics processor is greater than the first threshold but less than or equal to the second threshold, the frequency and supply voltage of the graphics processor are maintained; when the predicted power consumption of the graphics processor is greater than the second threshold, the frequency and supply voltage of the graphics processor are increased.

[0075] It should be noted that the specific values ​​of the first and second thresholds can be set according to the actual application scenario, and this application does not impose any restrictions on them.

[0076] In practical applications, when the power consumption of a graphics processing unit (GPU) increases, factors such as the potential for a rapid rise in power consumption within a short period, the complexity of frequency adjustment, and adjustment latency can prevent timely frequency adjustments, thus affecting GPU performance and consequently impacting user experience. Therefore, this application predicts the GPU's power consumption over a future period, enabling timely frequency adjustments to further ensure GPU performance and improve user experience.

[0077] The frequency of the graphics processor referred to in this application embodiment can be one or more of the following frequencies: core clock, memory clock, boost clock, and effective clock.

[0078] In an alternative embodiment, the predicted power consumption of the graphics processor can also be sent to the user for reference in adjusting specific services.

[0079] In one specific embodiment, please refer to Figure 4 , Figure 4 This illustrates a specific structure of a graphics processing unit (also known as a graphics processing unit).

[0080] In this embodiment, the graphics processor 40 includes multiple functional modules 401 and multiple utilization counters 201 corresponding to the multiple functional modules 401. The utilization counters 201 can count the utilization rate of the corresponding functional module 401, such as the ratio of the working time and idle time of the functional module 401 within a preset time interval.

[0081] The graphics processor 40 also includes a temperature sensor 403 for acquiring ambient temperature.

[0082] The graphics processor 40 also includes an integrated voltage sensing module 404 for acquiring the power supply voltage of the graphics processor 40.

[0083] The graphics processor 40 also includes at least one counter 406 for counting. A target value is preset, and when the counter 406 reaches the target value, the utilization counter 201, the temperature sensor 403, and the integrated voltage sensing module 404 transmit the collected operating parameters to the shadow register 405.

[0084] The graphics processor 40 also includes a power prediction model 202. The power prediction model 202 retrieves multiple operating parameters from the shadow register 405 and uses these parameters to predict the total power consumption of the graphics processor 40 and / or the power consumption of one or more functional modules 401.

[0085] The graphics processor 40 also includes a processing module (not shown) for adjusting the frequency of the graphics processor according to the power consumption of the graphics processor.

[0086] For more detailed implementations of this embodiment, please refer to the foregoing embodiments, which will not be repeated here.

[0087] This application's embodiments obtain the utilization rate of each functional module to reflect the impact of key functional modules in the graphics processor on power consumption. This utilization rate, combined with other operating parameters, contributes to the prediction of graphics processor power consumption. In other words, by using more comprehensive operating parameters, the accuracy of graphics processor power consumption prediction is improved. Furthermore, the graphics processor's frequency is adjusted based on its power consumption to match the frequency with its power consumption, thereby ensuring the graphics processor's performance.

[0088] Furthermore, the power consumption prediction model 202 calculates the power consumption of the graphics processor using the following formula:

[0089] ,

[0090] in, This indicates the power consumption of the graphics processor. This indicates the total utilization rate of the graphics processor. Indicates the load capacitance. This indicates the power supply voltage. Indicates clock frequency. Indicates the short-circuit time. Indicates short-circuit current. This indicates leakage current.

[0091] Furthermore, the graphics processor 40 also includes a locking module for locking the plurality of operating parameters in response to the count reaching a target value;

[0092] The restart module is used to store the multiple operating parameters in the memory and recount them.

[0093] Regarding the functional modules / units included in the various devices and products described in the above embodiments, they can be software modules / units, hardware modules / units, or a combination of both. For example, for various devices and products applied to or integrated into a chip, all of their functional modules / units can be implemented using hardware methods such as circuits, or at least some modules / units can be implemented using software programs that run on a processor integrated within the chip, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits; for various devices and products applied to or integrated into a chip module, all of their functional modules / units can be implemented using hardware methods such as circuits, and different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components of the chip module, or at least some modules / units can be implemented using hardware methods such as circuits. The implementation can be achieved using software programs that run on a processor integrated within the chip module. The remaining (if any) modules / units can be implemented using hardware methods such as circuits. For various devices or products applied to or integrated into terminal devices, each of their functional modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components within the terminal device. Alternatively, at least some modules / units can be implemented using software programs that run on a processor integrated within the terminal device. The remaining (if any) modules / units can be implemented using hardware methods such as circuits.

[0094] This application also discloses a storage medium, which is a computer-readable storage medium storing a computer program thereon. When the computer program is executed, it can perform the steps of the method shown in the foregoing embodiments. The storage medium may include read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc. The storage medium may also include non-volatile memory or non-transitory memory, etc.

[0095] This application also discloses a terminal device, which includes the aforementioned graphics processor (or graphics processor); or, the terminal device includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the steps of the aforementioned graphics processor performance control method when running the computer program.

[0096] In the embodiments of this application, "multiple" refers to two or more.

[0097] The descriptions of "first," "second," etc., appearing in the embodiments of this application are for illustrative purposes and to distinguish the objects being described. They have no order and do not indicate any special limitation on the number of devices in the embodiments of this application, nor do they constitute any limitation on the embodiments of this application.

[0098] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means.

[0099] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0100] In the several embodiments provided in this application, it should be understood that the disclosed methods, apparatuses, and systems can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for example, the division of units is merely a logical functional division, and other division methods may exist in actual implementation; for example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can be physically included separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0103] The integrated unit implemented as a software functional unit described above can be stored in a computer-readable storage medium. This software functional unit, stored in a storage medium, includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute some steps of the methods described in the various embodiments of this application.

[0104] While this application discloses the above information, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of this application; therefore, the scope of protection of this application shall be determined by the scope defined in the claims.

Claims

1. A performance control method for a graphics processor, characterized in that, include: Obtain multiple operating parameters of the graphics processor and the utilization rate of each functional module within the graphics processor; The power consumption of the graphics processor is predicted using the multiple operating parameters and the utilization rate of each functional module. The frequency of the graphics processor is adjusted according to the power consumption of the graphics processor.

2. The performance control method for a graphics processor according to claim 1, characterized in that, The utilization rate of each functional module is the ratio of the working time and idle time of each functional module within a preset time interval.

3. The performance control method for a graphics processor according to claim 1, characterized in that, Determining the power consumption of the graphics processor using the multiple operating parameters and the utilization rate of each functional module includes: The weights corresponding to the utilization rates of each functional module are predicted using the multiple operating parameters, and the total utilization rate of the graphics processor is determined based on the predicted weights and the utilization rates of each functional module. The power consumption of the graphics processor is determined based on the total utilization of the graphics processor and the plurality of operating parameters.

4. The performance control method for a graphics processor according to claim 3, characterized in that, The various operating parameters include power supply voltage, temperature, clock frequency, load, electrical characteristic parameters, and process data.

5. The performance control method for a graphics processor according to claim 4, characterized in that, The process data includes leakage current, and the acquisition of multiple operating parameters of the graphics processor includes: The temperature was collected; The leakage current is determined based on the temperature and a first mapping relationship, wherein the first mapping relationship includes each temperature and its corresponding leakage current.

6. The performance control method for a graphics processor according to claim 4, characterized in that, The process data includes leakage current, and the electrical characteristic parameters include short-circuit current and short-circuit time. The power consumption of the graphics processor is calculated using the following formula: ， in, This indicates the power consumption of the graphics processor. This represents the total utilization rate. Indicates the load capacitance. This indicates the power supply voltage. Indicates clock frequency. Indicates the short-circuit time. Indicates short-circuit current. This indicates leakage current.

7. The performance control method for a graphics processor according to claim 1, characterized in that, The process of obtaining multiple operating parameters of the graphics processor also includes: In response to the count reaching the target value, the plurality of operating parameters are locked; The multiple operating parameters are stored in the memory and counted again.

8. The performance control method for a graphics processor according to claim 1, characterized in that, The method of predicting the power consumption of the graphics processor using the multiple operating parameters and the utilization rate of each functional module includes: The multiple operating parameters and the utilization rate of each functional module are input into the power consumption prediction model, which is used to output the power consumption of the graphics processor.

9. A graphics processor, characterized in that, include: At least one functional module; A utilization counter is used to generate the utilization rate of each functional module. A power consumption prediction model is used to obtain multiple operating parameters of the graphics processor and predict the power consumption of the graphics processor using these operating parameters and the utilization rate of each functional module. A processing module is used to adjust the frequency of the graphics processor according to the power consumption of the graphics processor.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program, when run by the processor, performs the steps of the performance control method for the graphics processor as described in any one of claims 1 to 8.

11. A terminal device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, The processor performs the steps of the performance control method for the graphics processor according to any one of claims 1 to 8, or includes the graphics processor according to claim 9.

Images