GPU power consumption control method and system, computer equipment and storage medium

Abstract

The invention provides a GPU power consumption control method and system, computer equipment and a storage medium. The method comprises the steps that historical load information of a GPU in a first target time period is acquired; training a machine learning model according to the historical load information of the GPU in the first target time period, and taking the trained machine learning model as a load prediction model; predicting load information of the GPU in a second time period in the future according to the load prediction model; adjusting the working voltage and the working frequency of the GPU in the second time period in the future according to the load information of the GPU in the second time period in the future; and taking the load information corresponding to the current working voltage and working frequency of the GPU as historical load information, and obtaining the historical load information of the GPU in the first target time period again. According to the method, the balance of the performance and the energy consumption of the GPU is realized, new data can be regularly collected, and the prediction model is continuously updated and optimized so as to adapt to load changes in different application scenes.

Description

Technical Field

[0001] The present invention belongs to the technical field of electrical digital data processing, and particularly relates to a GPU power consumption control method, system, computer device, and storage medium. Background Art

[0002] With the rapid development of high-computation-load applications such as deep learning, real-time rendering, and scientific computing, the performance requirements of GPUs have been continuously increasing, resulting in a sharp increase in power consumption. Traditional static power management methods are difficult to achieve the best balance between performance and energy consumption, especially in scenarios where the multi-task load changes dynamically.

[0003] The existing technology mainly adjusts the voltage and frequency of the GPU through a preset fixed policy (DVFS). However, this method cannot adapt to complex computational load changes in real time, and is prone to resource waste or performance degradation. Therefore, there is an urgent need for a management solution that can dynamically adjust the GPU power consumption according to the real-time workload. Summary of the Invention

[0004] The present invention provides a GPU power consumption control method, system, computer device, and storage medium to dynamically adjust the GPU power consumption according to the real-time workload.

[0005] In a first aspect, the present invention provides a GPU power consumption control method, including:

[0006] S1. Obtain the historical load information of the GPU within a first target time period; wherein the historical load information includes the historical core utilization rate;

[0007] S2. Train a machine learning model according to the historical load information of the GPU within the first target time period, and use the trained machine learning model as a load prediction model;

[0008] S3. Predict the load information of the GPU within a future second time period according to the load prediction model;

[0009] S4. Adjust the working voltage and working frequency of the GPU within the future second time period according to the load information of the GPU within the future second time period;

[0010] S5. Use the load information corresponding to the current working voltage and working frequency of the GPU as historical load information, and return to execute the operation of S1.

[0011] Optionally, the training of the machine learning model according to the historical load information of the GPU within the first target time period and using the trained machine learning model as a load prediction model includes:

[0012] Construct a first loss function L1:

[0013]

[0014] Among them, M is the total number of historical load information; is the value of the m-th load information predicted by the machine learning model; n m is the actual value of the m-th load information;

[0015] And / or, construct the second loss function L2:

[0016]

[0017] Among them, M' is the total number of categories of load information; h m' is the true label of the m'-th category of load information; is the probability that the machine learning model predicts that the load information belongs to the m'-th category of load information;

[0018] Determine the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2;

[0019] Train the machine learning model according to the total loss of the machine learning model, and use the trained machine learning model as the load prediction model.

[0020] Optionally, the determining the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2 includes:

[0021] Calculate the total loss L of the machine learning model according to the following formula f :

[0022] L f = αL1 + βL2;

[0023] Among them, α is the preset weight of the first loss function; β is the preset weight of the second loss function.

[0024] Optionally, the adjusting the working voltage and working frequency of the GPU in the future second time period according to the load information of the GPU in the future second time period includes:

[0025] Obtain the predicted value of the core utilization rate by the load prediction model at the future target moment;

[0026] When the predicted value of the core utilization rate is less than the first target ratio, use the first preset voltage and the first preset frequency as the working voltage and working frequency of the GPU;

[0027] When the predicted value of the core utilization rate is greater than or equal to the first target ratio and less than or equal to the second target ratio, use the second preset voltage and the second preset frequency as the working voltage and working frequency of the GPU;

[0028] When the predicted value of the core utilization rate is greater than the second target ratio, the third preset voltage and the second preset frequency are used as the operating voltage and operating frequency of the GPU.

[0029] In a second aspect, the present invention provides a GPU power consumption control system, including:

[0030] An acquisition module, configured to acquire the historical load information of the GPU within a first target time period; wherein the historical load information includes the historical core utilization rate;

[0031] A training module, configured to train a machine learning model according to the historical load information of the GPU within the first target time period, and use the trained machine learning model as a load prediction model;

[0032] A prediction module, configured to predict the load information of the GPU within a future second time period according to the load prediction model;

[0033] An adjustment module, configured to adjust the operating voltage and operating frequency of the GPU within the future second time period according to the load information of the GPU within the future second time period;

[0034] A determination module, configured to use the load information corresponding to the current operating voltage and operating frequency of the GPU as the historical load information, and return to execute the operation of the acquisition module.

[0035] Optionally, the training module includes:

[0036] A first construction unit, configured to construct a first loss function L1:

[0037]

[0038] where M is the total number of historical load information; is the value of the m-th load information predicted by the machine learning model; n m is the actual value of the m-th load information;

[0039] and / or, a second construction unit, configured to construct a second loss function L2:

[0040]

[0041] where M' is the total number of categories of load information; h m' is the true label of the m'-th category of load information; is the probability that the machine learning model predicts that the load information belongs to the m'-th category of load information;

[0042] A first determination unit, configured to determine the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2;

[0043] A training unit for training a machine learning model according to the total loss of the machine learning model and using the trained machine learning model as a load prediction model.

[0044] Optionally, the first determination unit includes:

[0045] A computing device for computing the total loss L of the machine learning model according to the following formula f :

[0046] L f = αL1 + βL2;

[0047] Where α is the preset weight of the first loss function; β is the preset weight of the second loss function.

[0048] Optionally, the adjustment module includes:

[0049] An acquisition unit for acquiring the predicted value of the core utilization rate by the load prediction model at a future target time;

[0050] A second determination unit for using the first preset voltage and the first preset frequency as the working voltage and working frequency of the GPU when the predicted value of the core utilization rate is less than the first target ratio;

[0051] A third determination unit for using the second preset voltage and the second preset frequency as the working voltage and working frequency of the GPU when the predicted value of the core utilization rate is greater than or equal to the first target ratio and less than or equal to the second target ratio;

[0052] A fourth determination unit for using the third preset voltage and the second preset frequency as the working voltage and working frequency of the GPU when the predicted value of the core utilization rate is greater than the second target ratio.

[0053] In a third aspect, the present invention provides a computer device including a processor and a memory; wherein, when the processor executes the computer program stored in the memory, the steps of the GPU power consumption control method described in the first aspect are implemented.

[0054] In a fourth aspect, the present invention provides a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, the steps of the GPU power consumption control method described in the first aspect are implemented.

[0055] The present invention provides a GPU power consumption control method, system, computer device and storage medium. In the method, a prediction model is trained based on historical data to accurately predict load changes, and combined with a dynamic power consumption adjustment strategy, the balance between the performance and energy consumption of the GPU is achieved; the self-learning mechanism can regularly collect new data and continuously update and optimize the prediction model to adapt to load changes in different application scenarios, providing strong support for the application of the GPU in multiple fields. Brief Description of the Drawings

[0056] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0057] Figure 1 It is a schematic flowchart of a GPU power consumption control method provided by an embodiment of the present invention;

[0058] Figure 2 It is a schematic structural diagram of a GPU power consumption control system provided by an embodiment of the present invention. Detailed Embodiments

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0060] Embodiment 1

[0061] As Figure 1 shown, this embodiment provides a GPU power consumption control method, including:

[0062] S1. Obtain the historical load information of the GPU within the first target time period; the historical load information includes the historical core utilization rate.

[0063] Historical data is the basis for training the prediction model, mainly generated through the data recorded by the GPU runtime log and monitoring module, including but not limited to the following:

[0064] Task queue length: Records the number and processing time of tasks to be processed to reflect the current workload situation, and can be obtained through the registers of the GPU hardware scheduler. For example, the GPU hardware API (such as CUDA or ROCm) provides an interface to allow querying the task queue status.

[0065] Core utilization rate: The utilization rate of the GPU core, used to measure the intensity of computing resource usage.

[0066] Video memory utilization rate: The allocation and usage of video memory, reflecting the video memory pressure.

[0067] Temperature curve: Includes the time series changes of the GPU core temperature and video memory temperature, used to evaluate the impact of thermal load on performance.

[0068] Current and voltage: record the actual status of the power supply system and analyze the power consumption trend.

[0069] Fan speed: As auxiliary information, it reflects the operating status of the cooling system.

[0070] Frame rate (FPS): Records frame rate changes in real-time rendering tasks to measure the real-time requirements of the tasks.

[0071] Task category label: records the specific type of task (such as deep learning reasoning, image rendering, scientific computing, etc.) to facilitate classification and processing.

[0072] These data are collected through hardware sensors and driver layers and periodically saved in a historical database as training data.

[0073] S2: Train a machine learning model based on historical load information of the GPU in the first target time period, and use the trained machine learning model as a load prediction model.

[0074] In this step, illustratively, a first loss function L1 is constructed:

[0075]

[0076] Where M is the total number of historical load information; The value of the mth load information predicted by the machine learning model; n m is the actual value of the mth load information.

[0077] And / or, construct the second loss function L2:

[0078]

[0079] Where M' is the total number of load information categories; h m' is the true label of the m'th type of load information; The probability that the load information predicted by the machine learning model belongs to the m'th category of load information.

[0080] The total loss of the machine learning model is determined according to the first loss function L1 and / or the second loss function L2.

[0081] The machine learning model is trained according to the total loss of the machine learning model, and the trained machine learning model is used as a load prediction model.

[0082] Among them, the total loss L of the machine learning model is calculated according to the following formula f :

[0083] L f =αL1+βL2.

[0084] Among them, α is the preset weight of the first loss function; β is the preset weight of the second loss function.

[0085] Adjust the loss function weight according to the scenario to continuously optimize the load prediction model and better adapt to different scenarios.

[0086] S3. Predict the load information of the GPU in the second future time period according to the load prediction model.

[0087] S4. Adjust the working voltage and working frequency of the GPU in the second future time period according to the load information of the GPU in the second future time period.

[0088] Exemplarily, this step includes:

[0089] Obtain the predicted value of the core utilization rate by the load prediction model at the future target time.

[0090] When the predicted value of the core utilization rate is less than the first target ratio (e.g., 30%), use the first preset voltage and the first preset frequency as the working voltage and working frequency of the GPU (i.e., the operating frequency of the GPU core clock).

[0091] When the predicted value of the core utilization rate is greater than or equal to the first target ratio and less than or equal to the second target ratio (e.g., 70%), use the second preset voltage and the second preset frequency as the working voltage and working frequency of the GPU.

[0092] When the predicted value of the core utilization rate is greater than the second target ratio, use the third preset voltage and the second preset frequency as the working voltage and working frequency of the GPU.

[0093] Adjust the voltage and frequency of the GPU according to different load conditions. During the adjustment process, ensure that the voltage change takes precedence over the frequency change to prevent insufficient power supply caused by too high frequency. In low-load scenarios (predicted core utilization rate is less than 30%), reduce the voltage and frequency to avoid resource waste and effectively save energy consumption; in high-load scenarios (predicted core utilization rate is greater than 70%), quickly increase the voltage and frequency. While meeting the real-time requirements of tasks, accurately match the computing resources with the load, significantly improve the energy utilization efficiency, and greatly improve the energy efficiency ratio compared with the traditional fixed strategy.

[0094] S5. Use the load information corresponding to the current working voltage and working frequency of the GPU as historical load information, and return to perform the operation of S1.

[0095] The self-learning mechanism regularly collects the latest operating data during operation, expands the training data set, and uses incremental learning methods to update the parameters of the load prediction model online. At the same time, the loss function weight is dynamically adjusted according to the application scenario, such as increasing the error penalty for real-time tasks. This allows the load prediction model to be continuously optimized, better adapt to the dynamic changes of GPU load in different application scenarios, and maintain good control performance for a long time.

[0096] In summary, this embodiment provides a GPU power consumption control method, which trains a prediction model based on historical data to accurately predict load changes, and combines a dynamic power consumption adjustment strategy to achieve a balance between GPU performance and energy consumption; the self-learning mechanism can regularly collect new data and continuously update and optimize the prediction model to adapt to load changes in different application scenarios, providing strong support for the application of GPU in multiple fields.

[0097] Based on the same inventive concept as Example 1, this embodiment provides a GPU power consumption control system. Since the principle of solving the problem by this system is similar to the aforementioned GPU power consumption control method, the implementation of this system can refer to the implementation of the GPU power consumption control method.

[0098] like Figure 2 As shown, the GPU power consumption control system includes:

[0099] The acquisition module 10 is used to acquire the historical load information of the GPU within the first target time period; wherein the historical load information includes the historical core usage rate.

[0100] The training module 20 is used to train a machine learning model according to the historical load information of the GPU in the first target time period, and use the trained machine learning model as a load prediction model.

[0101] The prediction module 30 is used to predict the load information of the GPU in a future second time period according to the load prediction model.

[0102] The adjustment module 40 is used to adjust the operating voltage and the operating frequency of the GPU in the future second time period according to the load information of the GPU in the future second time period.

[0103] The determination module 50 is used to use the load information corresponding to the current operating voltage and operating frequency of the GPU as the historical load information, and return to execute the operation of the acquisition module.

[0104] Exemplarily, the training module includes:

[0105] The first construction unit is used to construct the first loss function L1:

[0106]

[0107] Where M is the total number of historical load information; is the value of the m-th load information predicted by the machine learning model; n m is the actual value of the m-th load information;

[0108] And / or, a second construction unit for constructing a second loss function L2:

[0109]

[0110] Where M' is the total number of load information categories; h m' is the true label of the m'-th category of load information; is the probability that the machine learning model predicts that the load information belongs to the m'-th category of load information;

[0111] A first determination unit for determining the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2;

[0112] A training unit for training the machine learning model according to the total loss of the machine learning model and using the trained machine learning model as a load prediction model.

[0113] Exemplarily, the first determination unit includes:

[0114] A calculation device for calculating the total loss L of the machine learning model according to the following formula f :

[0115] L f = αL1 + βL2;

[0116] Where α is the preset weight of the first loss function; β is the preset weight of the second loss function.

[0117] Exemplarily, the adjustment module includes:

[0118] An acquisition unit for acquiring the predicted value of the core utilization rate by the load prediction model at a future target time;

[0119] A second determination unit for using the first preset voltage and the first preset frequency as the working voltage and working frequency of the GPU when the predicted value of the core utilization rate is less than the first target ratio;

[0120] A third determination unit for using the second preset voltage and the second preset frequency as the working voltage and working frequency of the GPU when the predicted value of the core utilization rate is greater than or equal to the first target ratio and less than or equal to the second target ratio;

[0121] A fourth determination unit, configured to use a third preset voltage and a second preset frequency as the operating voltage and operating frequency of the GPU when a predicted value of the core utilization rate is greater than a second target ratio.

[0122] For the more specific working processes of the above-mentioned various modules, reference may be made to the corresponding content disclosed in Embodiment 1, and details will not be elaborated here.

[0123] Embodiment 3

[0124] This embodiment provides a computer device, including a processor and a memory; wherein, when the processor executes a computer program stored in the memory, the steps of the GPU power consumption control method described in Embodiment 1 are implemented.

[0125] For the more specific process of the above method, reference may be made to the corresponding content disclosed in Embodiment 1, and details will not be elaborated here.

[0126] Embodiment 4

[0127] This embodiment provides a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, the steps of the GPU power consumption control method described in Embodiment 1 are implemented.

[0128] For the more specific process of the above method, reference may be made to the corresponding content disclosed in Embodiment 1, and details will not be elaborated here.

[0129] Embodiment 5

[0130] This embodiment provides a computer program product, including computer-executable instructions or a computer program. When the computer-executable instructions or the computer program are executed by a processor, the steps of the GPU power consumption control method described in Embodiment 1 are implemented.

[0131] For the more specific process of the above method, reference may be made to the corresponding content disclosed in Embodiment 1, and details will not be elaborated here.

[0132] In this specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the systems, devices, storage media, and computer program products disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0133] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions in the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of the present invention.

[0134] In some embodiments, the computer-executable instructions can be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including being deployed as an independent program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0135] As an example, the computer-executable instructions may or may not correspond to files in a file system, and can be stored as part of a file that stores other programs or data. For example, they can be stored in one or more scripts in a Hyper Text Markup Language (HTML) document, stored in a single file dedicated to the program being discussed, or stored in multiple cooperating files (for example, files that store one or more modules, subroutines, or code portions).

[0136] As an example, the computer-executable instructions can be deployed to be executed on one electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed at multiple locations and interconnected through a communication network.

[0137] The present invention has been described in detail above in conjunction with specific embodiments and exemplary examples, but these descriptions should not be construed as limiting the present invention. Those skilled in the art understand that without departing from the spirit and scope of the present invention, various equivalent substitutions, modifications, or improvements can be made to the technical solutions of the present invention and their implementation manners, and all of these fall within the scope of the present invention. The protection scope of the present invention is subject to the appended claims.

Claims

1. A GPU power consumption control method, characterized in that: include: S1, obtaining historical load information of the GPU within a first target time period; The historical load information includes historical core usage; S2, training a machine learning model according to historical load information of the GPU in the first target time period, and using the trained machine learning model as a load prediction model; S3, predicting the GPU load information in a second future time period according to the load prediction model; S4, adjusting the operating voltage and operating frequency of the GPU in the second time period in the future according to the load information of the GPU in the second time period in the future; S5, taking the load information corresponding to the current operating voltage and operating frequency of the GPU as the historical load information, and returning to execute the operation of S1.

2. The GPU power consumption control method according to claim 1, characterized in that: The step of training a machine learning model according to historical load information of the GPU in the first target time period, and using the trained machine learning model as a load prediction model, includes: Construct the first loss function L1: Where M is the total number of historical load information; The value of the mth load information predicted by the machine learning model; n m is the actual value of the mth load information; And / or, construct the second loss function L2: Where M' is the total number of load information categories; h m' is the true label of the m'th type of load information; The probability that the load information predicted by the machine learning model belongs to the m'th type of load information; Determine the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2; The machine learning model is trained according to the total loss of the machine learning model, and the trained machine learning model is used as a load prediction model.

3. The GPU power consumption control method according to claim 2, characterized in that: Determining the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2 includes: The total loss L of the machine learning model is calculated according to the following formula f : L f =αL1+βL2; Among them, α is the preset weight of the first loss function; β is the preset weight of the second loss function.

4. The GPU power consumption control method according to claim 1, characterized in that: The step of adjusting the operating voltage and the operating frequency of the GPU in the second time period in the future according to the load information of the GPU in the second time period in the future includes: Obtain the predicted value of the core utilization rate at the future target time by the load prediction model; When the predicted value of the core usage rate is less than the first target ratio, the first preset voltage and the first preset frequency are used as the operating voltage and the operating frequency of the GPU; When the predicted value of the core usage rate is greater than or equal to the first target proportion and less than or equal to the second target proportion, the second preset voltage and the second preset frequency are used as the operating voltage and the operating frequency of the GPU; When the predicted value of the core usage rate is greater than the second target ratio, the third preset voltage and the second preset frequency are used as the operating voltage and the operating frequency of the GPU.

5. A GPU power consumption control system, characterized in that: include: An acquisition module, used to acquire historical load information of the GPU within a first target time period; The historical load information includes historical core usage; A training module, used to train a machine learning model according to historical load information of the GPU in a first target time period, and use the trained machine learning model as a load prediction model; A prediction module, used for predicting the load information of the GPU in a second future time period according to the load prediction model; An adjustment module, used for adjusting the operating voltage and operating frequency of the GPU in the future second time period according to the load information of the GPU in the future second time period; The determination module is used to use the load information corresponding to the current working voltage and working frequency of the GPU as the historical load information, and return to execute the operation of the acquisition module.

6. The GPU power consumption control system according to claim 5, characterized in that: The training module includes: The first construction unit is used to construct the first loss function L1: Where M is the total number of historical load information; The value of the mth load information predicted by the machine learning model; n m is the actual value of the mth load information; And / or, a second construction unit, for constructing a second loss function L2: Where M' is the total number of load information categories; h m' is the true label of the m'th type of load information; The probability that the load information predicted by the machine learning model belongs to the m'th type of load information; A first determining unit, used to determine the total loss of the machine learning model according to the first loss function L1 and / or the second loss function L2; The training unit is used to train the machine learning model according to the total loss of the machine learning model, and use the trained machine learning model as a load prediction model.

7. The GPU power consumption control system according to claim 6, characterized in that: The first determining unit includes: A computing device for calculating the total loss L of the machine learning model according to the following formula f : L f =αL1+βL2; Among them, α is the preset weight of the first loss function; β is the preset weight of the second loss function.

8. The GPU power consumption control system according to claim 5, characterized in that: The adjustment module comprises: An acquisition unit, used to acquire a predicted value of the core usage rate at a future target moment by a load prediction model; A second determining unit is configured to use a first preset voltage and a first preset frequency as an operating voltage and an operating frequency of the GPU when the predicted value of the core usage rate is less than a first target ratio; A third determining unit is configured to use the second preset voltage and the second preset frequency as the operating voltage and the operating frequency of the GPU when the predicted value of the core usage rate is greater than or equal to the first target proportion and less than or equal to the second target proportion; The fourth determination unit is used to use the third preset voltage and the second preset frequency as the operating voltage and the operating frequency of the GPU when the predicted value of the core usage rate is greater than the second target ratio.

9. A computer device, characterized in that: The method comprises a processor and a memory; wherein the processor implements the steps of the GPU power consumption control method described in any one of claims 1 to 4 when executing a computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: Used to store computer programs; when the computer programs are executed by a processor, the steps of the GPU power consumption control method described in any one of claims 1-4 are implemented.

Images