Knowledge transfer driven data center computing power energy efficiency modeling method and device

Abstract

The application discloses a kind of knowledge transfer driven data center computing power energy efficiency modeling method and device, method includes: obtaining the energy efficiency data of source computing power equipment and target computing power equipment with mark;The energy efficiency data of source computing power equipment and target computing power equipment is respectively preprocessed, to obtain respective key energy efficiency characteristic space;Based on unbalanced optimal transmission, the energy efficiency model of cross computing power equipment is constructed;The energy efficiency model is trained, and the energy efficiency characteristic space difference learning between heterogeneous computing power equipment is realized;The energy efficiency model trained is used for the energy efficiency modeling of the target computing power equipment to be evaluated, and the energy efficiency performance is evaluated.The application improves the problem that computing power equipment energy efficiency data reuse is difficult, modeling cost is high and few sample modeling precision is poor due to the field drift of heterogeneous computing power equipment energy efficiency characteristic space in the scene that data acquisition is limited.

Description

Technical Field

[0001] This invention belongs to the technical field of data centers, specifically relating to a knowledge transfer-driven data center computing power and energy efficiency modeling method and apparatus. Background Technology

[0002] In recent years, with the continuous expansion of global cloud data center scale, their energy consumption has also increased dramatically. It is estimated that in 2022, the global power consumption of cloud data centers will reach 240-340 TWh, accounting for approximately 1%-1.3% of global final power consumption. Furthermore, the deployment of complex artificial intelligence applications, especially large language models, by hyperscale public cloud providers has further exacerbated the energy consumption problem of data centers. As a critical infrastructure of data centers, massive heterogeneous computing power equipment, i.e., various types of servers, has an average resource utilization rate of 12%-18%, but its energy consumption accounts for 42% of the overall energy consumption of data centers. Therefore, how to achieve efficient utilization of computing power resources has become the key to energy conservation in data centers. Currently, most energy efficiency optimization research focuses on improving the resource utilization efficiency of computing power equipment, i.e., optimizing computing power equipment parameters and dynamically scheduling computing power resources. The common thread in these studies is finding the correlation between computing power equipment resource characteristics and energy consumption, which is the significance of establishing computing power equipment energy efficiency models.

[0003] With the rapid development of Machine Learning (ML), its powerful nonlinear fitting capabilities have been used to model the energy efficiency of computing devices. For example, "Lin W, Wu G, Wang X, et al. An artificial neural network approach to power consumption model construction for servers in cloud data centers[J]. IEEE Transactions on Sustainable Computing, 2020, 5(3):329-340." uses energy efficiency data from servers running different types of workloads to train an artificial neural network model for energy efficiency modeling. Similarly, "Wu W, Lin W, He L, et al. A power consumption model for cloud servers based on elman neural network[J]. IEEE Transactions on Cloud Computing, 2021, 9(4):1268-1277." proposes a server time series energy efficiency model trained based on Elman neural network, realizing time series prediction of computing device energy efficiency. Although many current ML-based methods for modeling the energy efficiency of computing devices have shown superior performance, research on energy efficiency modeling of computing devices for actual data centers still faces the following two challenges:

[0004] Energy efficiency modeling of computing equipment in data-constrained scenarios: On the one hand, energy efficiency models of computing equipment built based on traditional machine learning methods often require the collection of a large amount of labeled data for model training to obtain an effective energy efficiency model. However, due to the short operating time of newly deployed computing equipment in data centers, the amount of energy efficiency data that can be obtained is insufficient to meet the training requirements of the model. At the same time, due to the commercial privacy of data centers, a large amount of energy efficiency data of computing equipment, as a feature that directly reflects the operating status of the data center, is often prohibited from being disclosed to the public. This also leads to the challenge of high data acquisition difficulty and high acquisition cost for energy efficiency modeling of computing equipment in real data centers. Therefore, how to achieve energy efficiency modeling of computing equipment in real data centers under data-constrained scenarios is of great significance.

[0005] Energy efficiency data reuse across computing devices: On the other hand, energy efficiency models for computing devices built using traditional machine learning (ML) methods only exhibit superior performance if the assumption of independent and identically distributed training and testing data is met. However, in real-world data centers, to meet the computing resource needs of different users, a large number of heterogeneous computing devices are often deployed, such as general-purpose computing devices with CPU chips as the computing core and intelligent computing devices with AI chips such as GPUs, FPGAs, TPUs, and NPUs as the computing core. Different computing devices exhibit varying energy efficiency due to differences in hardware or operating load, leading to a drift in the distribution of energy efficiency characteristics across different devices. This makes it difficult for energy efficiency models built using traditional ML to demonstrate good generalization performance. Therefore, obtaining a high-precision energy efficiency model requires collecting a large amount of energy efficiency data from a specified target server for model training, resulting in high data acquisition costs. In fact, although different computing devices have different operating loads or hardware, the data collected from different servers contains shared energy efficiency characteristics, making it difficult for traditional ML methods to achieve effective knowledge mining in non-independent and identically distributed scenarios. Therefore, how to further explore the valuable knowledge contained in the collected energy efficiency data and realize the reuse of energy efficiency data across computing devices, thereby improving the energy efficiency modeling performance when the target computing device has only a small amount of labeled data, is of great significance for energy efficiency modeling of data center computing devices. Summary of the Invention

[0006] The main objective of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a knowledge transfer-driven data center computing power energy efficiency modeling method and apparatus. By learning the differences in energy efficiency characteristic distribution between the source computing power device and the target computing power device, the valuable knowledge contained in the large amount of labeled energy efficiency historical data of the source computing power device can be used to improve the energy efficiency modeling accuracy of the target computing power device with only a small amount of labeled energy efficiency data.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a knowledge transfer-driven data center computing power efficiency modeling method, comprising the following steps:

[0009] Acquire energy efficiency data of source computing power devices and target computing power devices; the source computing power devices include a large amount of labeled energy efficiency data, and the target computing power devices include a small amount of labeled energy efficiency data;

[0010] The energy efficiency data of the source computing power device and the target computing power device are preprocessed to obtain their respective key energy efficiency feature spaces; the key energy efficiency feature space refers to the features most relevant to the energy efficiency of the computing power device.

[0011] An energy efficiency model across computing devices is constructed based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the edge distribution and conditional distribution of key energy efficiency features of the source and target computing devices are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing device is used to improve the energy efficiency modeling accuracy of the target computing device, which has only a small amount of labeled energy efficiency data.

[0012] The energy efficiency model is trained to learn the spatial differences in key energy efficiency characteristics among heterogeneous computing devices, thereby achieving effective energy efficiency knowledge transfer among heterogeneous computing devices.

[0013] The trained energy efficiency model is used to model the energy efficiency of the target computing device to be evaluated, and to assess its energy efficiency performance.

[0014] As a preferred technical solution, the acquisition of energy efficiency data of the source computing power device and the target computing power device specifically includes:

[0015] The computing power device with a large amount of labeled historical energy efficiency data will be designated as the source computing power device X. s The energy efficiency dataset of the source computing power device is represented as D. s ={x s ;y s The number of labeled samples is N. s ;

[0016] The computing power device with a small amount of labeled energy efficiency data is set as the target computing power device X. T The energy efficiency dataset corresponding to the target computing power device is represented as D. T ={x T ;y T The sample size is N. t Among them, energy efficiency data with labels is represented as The number of energy efficiency data is n t ; where n t <<N t <<N s ;

[0017] Let X S The joint probability distribution of the energy efficiency characteristic space is represented by P. S X T The joint probability distribution of the energy efficiency characteristic space is represented by P. T .

[0018] As a preferred technical solution, the energy efficiency data of the source computing power device and the target computing power device are preprocessed to obtain their respective key energy efficiency feature spaces, specifically as follows:

[0019] Principal component analysis is used to reduce the dimensionality of the key energy efficiency feature spaces of the source computing power device and the target computing power device, respectively, so as to obtain the key energy efficiency feature spaces of the source computing power device and the target computing power device.

[0020] To eliminate the impact of differences in the units and numerical ranges of energy efficiency values ​​among labeled energy efficiency data from different computing power devices on the training of the energy efficiency model, the energy efficiency values ​​of the source and target computing power devices are normalized using Min-Max, as follows:

[0021]

[0022] Where y represents the energy efficiency value of the computing device, max(y) and min(y) represent the maximum and minimum energy efficiency values ​​in each energy efficiency dataset, respectively, and y′ represents the normalized energy efficiency value, which ranges from [0,1].

[0023] The objective function of the energy efficiency model, as a preferred technical solution, is as follows:

[0024]

[0025] Here, f(·) represents the energy efficiency modeling function, which belongs to the regenerative Hilbert space. This represents the joint probability distribution of energy efficiency data for unlabeled target computing devices. It represents the projection space of two joint probability distributions. This represents the i-th energy efficiency characteristic data of the source computing power device. This indicates the corresponding energy efficiency value; This represents the j-th energy efficiency characteristic data of the target computing power device. This represents the corresponding energy efficiency value; 1 represents the identity matrix; Γ i,j ∈Γ represents the unbalanced optimal transmission plan, C(·) represents a joint cost function, which includes the energy efficiency sample distance between the source and target computing devices; Ω(·) represents the regularization term, D KL (·) represents the KL divergence, denoted as D. KL (z) = zlog(z) - z; δ≥0 indicates that it is a parameter used to balance the complexity of the cost function C(·) and the energy efficiency modeling function f; λ1 and λ2 represent the penalty hyperparameters, respectively; the target computing power equipment training data contains n t Labeled energy efficiency data were used to fit the energy efficiency modeling function f(·) in a semi-supervised setting. As a constraint condition for training the energy efficiency model.

[0026] The preferred technical solution includes general-purpose computing devices with CPU chips as the computing core and intelligent computing devices with AI chips as the computing core.

[0027] As a preferred technical solution, training the energy efficiency model to achieve spatial differences in energy efficiency characteristics among heterogeneous computing devices specifically involves:

[0028] The large amount of labeled energy efficiency data from the source computing power device and the small amount of labeled energy efficiency data from the target computing power device are used as training data.

[0029] By fixing f(·), we first solve for Γ, and then iteratively update the optimization objective function based on the Majorization-Minimization algorithm to obtain the transmission plan Γ.

[0030] With the transmission plan Γ fixed, the objective function is further expressed in the following form:

[0031]

[0032] in, The predicted energy efficiency value, n t <j≤N t ;

[0033] Finally, the above optimization objective is fitted using the training data until the model converges or reaches the maximum number of iterations, thus obtaining the optimal energy efficiency modeling function f(·).

[0034] As a preferred technical solution, unlabeled energy efficiency data of the target computing power device is input into a trained energy efficiency model to achieve energy efficiency evaluation of the target computing power device, and the performance of the energy efficiency prediction model is quantitatively analyzed using mean square error and mean absolute error.

[0035] Secondly, the present invention provides a knowledge transfer-driven data center computing power energy efficiency modeling system, which is applied to the knowledge transfer-driven data center computing power energy efficiency modeling method, including a data acquisition module, a preprocessing module, an energy efficiency model construction module, an energy efficiency model training module, and an energy efficiency evaluation module.

[0036] The data acquisition module is used to acquire energy efficiency data of the source computing power device and the target computing power device; the source computing power device includes a large amount of labeled energy efficiency data, and the target computing power device includes a small amount of labeled energy efficiency data;

[0037] The preprocessing module is used to preprocess the energy efficiency data of the source computing power device and the target computing power device respectively to obtain their respective key energy efficiency feature spaces; the key energy efficiency feature space refers to the features most relevant to the energy efficiency of the computing power device.

[0038] The energy efficiency model construction module is used to construct an energy efficiency model across computing power devices based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the energy efficiency edge distribution and conditional distribution of the source computing power device and the target computing power device are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing power device is used to improve the energy efficiency modeling accuracy of the target computing power device, which has only a small amount of labeled energy efficiency data.

[0039] The energy efficiency model training module is used to train the energy efficiency model, learn the key energy efficiency feature spatial differences between heterogeneous computing devices, and thus realize effective energy efficiency knowledge transfer between heterogeneous computing devices.

[0040] The energy efficiency assessment module is used to apply the trained energy efficiency model to the energy efficiency modeling of the target computing power device to be assessed, and to evaluate the energy efficiency performance.

[0041] Thirdly, the present invention provides an electronic device, the electronic device comprising:

[0042] At least one processor; and,

[0043] A memory communicatively connected to the at least one processor; wherein,

[0044] The memory stores computer program instructions that can be executed by the at least one processor, which are then executed by the at least one processor to enable the at least one processor to execute the knowledge transfer-driven data center computing power and energy efficiency modeling method.

[0045] Fourthly, the present invention provides a computer-readable storage medium storing a program, which, when executed by a processor, implements the knowledge transfer-driven data center computing power and energy efficiency modeling method.

[0046] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0047] This invention, starting from the practical aspects of energy efficiency modeling for data center computing equipment, proposes a knowledge transfer-driven energy efficiency modeling method for data center computing equipment that lacks labeled energy efficiency data. By learning the spatial distribution differences of key energy efficiency characteristics between the source and target computing equipment, it leverages the valuable knowledge contained in the large amount of labeled energy efficiency historical data of the source computing equipment to improve the energy efficiency modeling accuracy of the target computing equipment, which has only a small amount of labeled energy efficiency data. This invention overcomes the shortcomings of existing energy efficiency modeling methods, improves the generalization performance of the energy efficiency model for target computing equipment, reduces the data acquisition cost for energy efficiency modeling in real data centers, and enhances the usability of the energy efficiency modeling method in actual data centers. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a flowchart of the knowledge transfer-driven data center computing power and energy efficiency modeling method according to an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram of the structure of the knowledge transfer-driven data center computing power and energy efficiency modeling system according to an embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0052] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0053] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0054] Please see Figure 1 This embodiment proposes a knowledge transfer-driven data center computing power energy efficiency modeling method, which includes the following steps:

[0055] Step 1: Set the computing power device with a large amount of labeled historical energy efficiency data as the source computing power device X. S The corresponding energy efficiency dataset is represented as D. S ={x S ;y S The number of labeled samples is N. s At the same time, the computing power device with a very small amount of labeled data is set as the target computing power device X. TThe corresponding energy efficiency dataset is represented as D. T ={x T ;y T The sample size is N. t The labeled data is represented as The sample size is n t And assume X S The joint probability distribution of the energy efficiency characteristic space is represented by P. s X T The joint probability distribution of the energy efficiency characteristic space is represented by P. T It is important to note that n t <<N t <<N s .

[0056] Step 2: Preprocess the energy efficiency data of the source computing power device and the target computing power device respectively to obtain their respective key energy efficiency characteristics.

[0057] Furthermore, to obtain the key energy efficiency characteristics of both the source and target computing devices while preserving their key energy efficiency characteristic spaces, Principal Component Analysis (PCA) is first used to reduce the dimensionality of the energy efficiency characteristic spaces of both devices. Then, the differences in energy efficiency characteristic spaces between different types of computing devices deployed in data centers, including but not limited to general-purpose computing devices with CPU chips as the computing core and intelligent computing devices with AI chips such as GPUs, FPGAs, TPUs, and NPUs as the computing core, are analyzed to explore the feasibility of energy efficiency knowledge transfer between different types of computing devices.

[0058] Meanwhile, to eliminate the impact of differences in the dimensions and numerical ranges of energy efficiency values ​​among labeled energy efficiency data from different computing devices on model training, the energy efficiency values ​​of the source and target computing devices were normalized using Min-Max, as follows:

[0059]

[0060] Where y represents the energy efficiency value of the computing device, max(y) and min(y) represent the maximum and minimum energy efficiency values ​​in each dataset, respectively, and y′ represents the normalized label, with a value range of [0,1].

[0061] Step 3: Construct a semi-supervised domain adaptive regression model based on imbalanced optimal transport, whose objective function can be expressed as:

[0062]

[0063] Here, f(·) represents the energy efficiency modeling function, which belongs to the regenerative Hilbert space. Γ i,j ∈Γ represents an unbalanced optimal transmission plan. This represents the joint probability distribution of the target computing devices in a semi-supervised setup. It represents the projection space of two joint probability distributions. This represents the i-th energy efficiency characteristic data of the source computing power device. This indicates the corresponding energy efficiency value. This represents the j-th energy efficiency characteristic data of the target computing power device. This represents the corresponding energy efficiency value. Ω(·) represents the regularization term, D KL (·) represents the KL divergence, denoted as D. KL (z) = zlog(z) - z. δ≥0 represents the parameter used to balance the complexity of the cost function C(·) and the energy efficiency modeling function f(·). Furthermore, the target server training data contains n t Labeled energy efficiency data were used to fit the energy efficiency modeling function f(·) in a semi-supervised setting. As a constraint on the model, C(·) represents a joint cost function, which includes the energy efficiency sample distance between the source and target computing devices, and a loss function L(·) that measures the difference between the corresponding energy efficiency values, expressed as:

[0064]

[0065] Where q(·) represents the energy efficiency data of the computing power source device. Energy efficiency data of target computing equipment The difference is a function. σ is a hyperparameter used to balance the difference in energy efficiency characteristic space with the energy efficiency value loss L(·).

[0066] In summary, the ultimate goal of the above objective function is to find an optimal unbalanced optimal transmission plan Γ and an energy efficiency modeling function f(·). By optimizing the above objective function, the differences in the marginal and conditional distributions of energy efficiency of the source and target computing devices can be learned simultaneously. This allows the valuable energy efficiency knowledge contained in the large amount of energy efficiency data from the source computing devices to improve the accuracy of energy efficiency modeling for the target computing devices, which have only a small amount of sample data.

[0067] Step 4: Using the large amount of labeled data from the source computing power device and the very small amount of labeled data from the target computing power device as training data, train the semi-supervised domain adaptive regression model constructed in Step 3 to obtain the optimal transmission plan Γ and the energy efficiency modeling function f(·). Given the training dataset, since the objective function defined in Step 3 has smoothness, during model training, one function can be fixed while solving the other. First, f(·) can be fixed, and Γ can be solved first. The optimization objective function in Step 3 is iteratively updated based on the Majorization-Minimization algorithm, expressed as follows:

[0068]

[0069] Here, `diag(·)` represents the diagonal element extraction operation of a matrix. `⊙` represents the matrix multiplication operation. `exp(·)` represents the exponential function form. m This represents an m-dimensional identity matrix. 1 n This represents an n-dimensional identity matrix. Let λ = λ1 = λ2. k represents the number of iterations.

[0070] Once the transmission plan Γ is fixed, the objective function constructed in step 3 can be further expressed in the following form:

[0071]

[0072] in, This represents the predicted energy efficiency value.

[0073] Furthermore, the objective function described above can be rewritten in kernel form as follows:

[0074]

[0075] Here, k(·) represents the kernel function. ω j ∈ω represents the weight. Used as a constraint condition for training the constraint model.

[0076] Furthermore, the objective function described above can be rewritten in Lagrange form as follows:

[0077]

[0078] in, Furthermore, ρ represents a parameter, and k is represented as:

[0079]

[0080] The optimal parameters can be obtained by setting the first derivative of the Lagrange objective function to zero, as follows:

[0081]

[0082] Step 5: Input the unlabeled energy efficiency data of the target computing power device into the trained model to evaluate the energy efficiency of the target computing power device. Furthermore, use MSE and MAE to quantitatively analyze the performance of the energy efficiency prediction model. The calculation formula is as follows:

[0083]

[0084] Where M represents the number of test samples; p i and These represent the actual value and the prediction of the target computing power device, respectively.

[0085] In another embodiment of the present invention, the feasibility of the present invention was verified, mainly including the following two steps:

[0086] Step 1: Experiment setup, details as follows:

[0087] Dataset Selection: This embodiment was conducted on the historical dataset SURFsara from a real data center. Three consecutive days of energy efficiency data from six CPU-based computing devices and three GPU-based computing devices were selected for the experiment. During the implementation, it was assumed that the source computing device had 15,000 labeled energy efficiency data points, while the target computing device had 500, with only 10 labeled energy efficiency data points. Specific device information is shown in Table 1.

[0088] Table 1. Information on Experimental Computing Equipment

[0089] CPU architecture r10n20, r11n15, r12n6, r13n30, r25n30, r27n13 GPU architecture r30n5, r31n1, r31n2

[0090] Comparison Method: In this embodiment, the method proposed in this invention is compared and analyzed with four popular traditional ML energy efficiency modeling methods (i.e., Linear Regression, SVM, AdaBoost and Random Forest) and two energy efficiency modeling methods based on transfer learning (i.e., JDOT and DARE-GRAM).

[0091] Simulation Setup: This embodiment sets up performance analysis for cross-computing device energy efficiency modeling in two scenarios (i.e., cross-homogeneous computing device energy efficiency modeling and cross-heterogeneous computing device energy efficiency modeling). For cross-homogeneous computing device energy efficiency modeling, this embodiment verifies the interaction between CPU architecture computing devices. For cross-heterogeneous computing device energy efficiency modeling, this embodiment considers the interaction between CPU architecture computing devices and GPU architecture computing devices.

[0092] Step 2: Performance comparison, details are as follows:

[0093] This embodiment demonstrates a performance comparison between the proposed method and traditional ML methods for energy efficiency modeling across computing devices. It can be seen that the proposed method exhibits superior performance under different task settings. In contrast, traditional ML methods struggle to address the differences in energy efficiency feature spaces between different computing devices, resulting in poorer performance. The comparison results are shown in Figure 2.

[0094] Table 2. Performance comparison between this embodiment and the traditional ML-based cross-computing power device energy efficiency modeling method.

[0095]

[0096] This embodiment demonstrates a performance comparison between the proposed method and the energy efficiency modeling of cross-homogeneous computing devices based on transfer learning. As can be seen, the proposed method is significantly superior to other comparative methods in the energy efficiency modeling task of cross-homogeneous computing devices. The comparison results are shown in Table 3.

[0097] Table 3. Performance comparison between this embodiment and the energy efficiency modeling method for cross-isomorphic computing devices based on transfer learning.

[0098]

[0099] This embodiment demonstrates a performance comparison between the proposed method and the energy efficiency modeling of heterogeneous computing devices based on transfer learning. As can be seen, the proposed method is significantly superior to other comparative methods in the energy efficiency modeling task of heterogeneous computing devices, further illustrating the effectiveness of the proposed method. The comparison results are shown in Table 4.

[0100] Table 4. Performance comparison between this embodiment and the energy efficiency modeling method for cross-heterogeneous computing devices based on transfer learning.

[0101]

[0102] It should be noted that, for the sake of simplicity, the aforementioned method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously.

[0103] Based on the same idea as the knowledge transfer-driven data center computing power energy efficiency modeling method in the above embodiments, the present invention also provides a knowledge transfer-driven data center computing power energy efficiency modeling system, which can be used to execute the above-described knowledge transfer-driven data center computing power energy efficiency modeling method. For ease of explanation, the structural diagram of the knowledge transfer-driven data center computing power energy efficiency modeling system embodiment only shows the parts related to the embodiments of the present invention. Those skilled in the art will understand that the illustrated structure does not constitute a limitation on the device, and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0104] Please see Figure 2 In another embodiment of this application, a knowledge transfer-driven data center computing power energy efficiency modeling system 100 is provided. The system includes a data acquisition module 101, a preprocessing module 102, an energy efficiency model construction module 103, an energy efficiency model training module 104, and an energy efficiency evaluation module 105.

[0105] The data acquisition module 101 is used to acquire energy efficiency data of the source computing power device and the target computing power device; the source computing power device includes a large amount of labeled energy efficiency data, and the target computing power device includes a small amount of labeled energy efficiency data.

[0106] The preprocessing module 102 is used to preprocess the energy efficiency data of the source computing power device and the target computing power device respectively to obtain their respective key energy efficiency feature spaces.

[0107] The energy efficiency model construction module 103 is used to construct an energy efficiency model across computing power devices based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the energy efficiency edge distribution and conditional distribution of the source computing power device and the target computing power device are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing power device is used to improve the energy efficiency modeling accuracy of the target computing power device with only a small amount of sample data.

[0108] The energy efficiency model training module 104 is used to train the energy efficiency model, learn the differences in energy efficiency feature space between heterogeneous computing devices, and thus realize effective energy efficiency knowledge transfer between heterogeneous computing devices.

[0109] The energy efficiency evaluation module 105 is used to apply the trained energy efficiency model to the energy efficiency modeling of the target computing power device to be evaluated, and to evaluate the energy efficiency performance.

[0110] It should be noted that the knowledge transfer-driven data center computing power energy efficiency modeling system of the present invention corresponds one-to-one with the knowledge transfer-driven data center computing power energy efficiency modeling method of the present invention. The technical features and beneficial effects described in the embodiments of the knowledge transfer-driven data center computing power energy efficiency modeling method are all applicable to the embodiments of knowledge transfer-driven data center computing power energy efficiency modeling. For details, please refer to the description in the embodiments of the method of the present invention, which will not be repeated here.

[0111] Furthermore, in the implementation of the knowledge transfer-driven data center computing power and energy efficiency modeling system in the above embodiments, the logical division of each program module is only an example. In actual applications, the above functions can be assigned to different program modules as needed, for example, for the configuration requirements of the corresponding hardware or for the convenience of software implementation. That is, the internal structure of the knowledge transfer-driven data center computing power and energy efficiency modeling system is divided into different program modules to complete all or part of the functions described above.

[0112] Please see Figure 3 In one embodiment, an electronic device is provided for implementing a knowledge transfer-driven data center computing power energy efficiency modeling method. The electronic device 200 may include a first processor 201, a first memory 202 and a bus, and may also include a computer program stored in the first memory 202 and executable on the first processor 201, such as a knowledge transfer-driven data center computing power energy efficiency modeling program 203.

[0113] The first memory 202 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the first memory 202 can be an internal storage unit of the electronic device 200, such as the portable hard drive of the electronic device 200. In other embodiments, the first memory 202 can also be an external storage device of the electronic device 200, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 200. Furthermore, the first memory 202 can include both internal storage units and external storage devices of the electronic device 200. The first memory 202 can be used not only to store application software and various types of data installed on the electronic device 200, such as the code of the knowledge transfer-driven data center computing power efficiency modeling program 203, but also to temporarily store data that has been output or will be output.

[0114] In some embodiments, the first processor 201 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The first processor 201 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the first memory 202 and calls data stored in the first memory 202 to perform various functions of the electronic device 200 and process data.

[0115] Figure 3 Only electronic devices with components are shown; those skilled in the art will understand that... Figure 3 The structure shown does not constitute a limitation on the electronic device 200, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0116] The knowledge transfer-driven data center computing power efficiency modeling program 203 stored in the first memory 202 of the electronic device 200 is a combination of multiple instructions. When run in the first processor 201, it can achieve the following:

[0117] Acquire energy efficiency data from source computing devices and target computing devices; the source computing devices include a large amount of labeled energy efficiency data, and the target computing devices include a small amount of labeled energy efficiency data;

[0118] The energy efficiency data of the source computing power device and the target computing power device are preprocessed to obtain their respective key energy efficiency feature spaces;

[0119] An energy efficiency model across computing devices is constructed based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the energy efficiency edge distribution and conditional distribution of the source computing device and the target computing device are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing device is used to improve the energy efficiency modeling accuracy of the target computing device with only a small amount of sample data.

[0120] The energy efficiency model is trained to learn the spatial differences in energy efficiency characteristics among heterogeneous computing devices, thereby achieving effective energy efficiency knowledge transfer among heterogeneous computing devices.

[0121] The trained energy efficiency model is used to model the energy efficiency of the target computing device to be evaluated, and to assess its energy efficiency performance.

[0122] Furthermore, if the modules / units integrated in the electronic device 200 are implemented as software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0123] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0124] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A knowledge transfer-driven data center computing power energy efficiency modeling method, characterized in that, Includes the following steps: Acquire energy efficiency data of source computing power devices and target computing power devices; the source computing power devices include a large amount of labeled energy efficiency data, and the target computing power devices include a small amount of labeled energy efficiency data; The energy efficiency data of the source computing power device and the target computing power device are preprocessed to obtain their respective key energy efficiency feature spaces; The key energy efficiency feature space refers to the features most relevant to the energy efficiency of computing equipment. An energy efficiency model across computing devices is constructed based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the edge distribution and conditional distribution of key energy efficiency features of the source and target computing devices are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing device is used to improve the energy efficiency modeling accuracy of the target computing device, which has only a small amount of labeled energy efficiency data. The energy efficiency model is trained to learn the spatial differences in key energy efficiency characteristics among heterogeneous computing devices, thereby achieving effective energy efficiency knowledge transfer among heterogeneous computing devices. The trained energy efficiency model is used to model the energy efficiency of the target computing power device to be evaluated, and the energy efficiency performance is assessed. The objective function of the energy efficiency model is as follows: in, This represents an energy efficiency modeling function, which belongs to the regenerative Hilbert space. , This represents the joint probability distribution of energy efficiency data for unlabeled target computing devices. It represents the projection space of two joint probability distributions. This represents the i-th energy efficiency characteristic data of the source computing power device. This indicates the corresponding energy efficiency value; This represents the j-th energy efficiency characteristic data of the target computing power device. This indicates the corresponding energy efficiency value; It represents the identity matrix; This represents an unbalanced optimal transmission plan. It is represented as a joint cost function, which includes the energy efficiency sample distance between the source computing power device and the target computing power device; This represents the regularization term. This represents the KL divergence, expressed as... ; This is used to balance the cost function. With energy efficiency modeling function The parameter of complexity; and These represent the penalty hyperparameters; the training data for the target computing power device contains... Labeled energy efficiency data was used to fit an energy efficiency modeling function in a semi-supervised setting. , As a constraint for training the energy efficiency model; The training of the energy efficiency model to learn the spatial differences in energy efficiency characteristics among heterogeneous computing devices specifically involves: The large amount of labeled energy efficiency data from the source computing power device and the small amount of labeled energy efficiency data from the target computing power device are used as training data. Will To fix it, first... The solution is obtained by iteratively updating the objective function based on the Majorization-Minimization algorithm, leading to the transmission plan. ; Transmission Plan The objective function is further expressed in the following form: in, ; Finally, the training data is used to fit the above optimization objective until the model converges or reaches the maximum number of iterations, thus obtaining the optimal energy efficiency modeling function. .

2. The knowledge transfer-driven data center computing power energy efficiency modeling method according to claim 1, characterized in that, The acquisition of energy efficiency data for the source and target computing devices specifically involves: Configure computing power devices with a large amount of labeled historical energy efficiency data as source computing power devices. The energy efficiency dataset of the source computing power device is represented as The number of labeled samples is ; Set computing power devices with limited labeled energy efficiency data as target computing power devices. The energy efficiency dataset corresponding to the target computing power device is represented as follows: The sample size is Among them, energy efficiency data with labels is represented as The number of energy efficiency data is ;in, ; set up The joint probability distribution of the energy efficiency characteristic space is expressed as: , The joint probability distribution of the energy efficiency characteristic space is expressed as: .

3. The knowledge transfer-driven data center computing power energy efficiency modeling method according to claim 1, characterized in that, The energy efficiency data of the source computing power device and the target computing power device are preprocessed to obtain their respective key energy efficiency feature spaces, specifically as follows: Principal component analysis is used to reduce the dimensionality of the key energy efficiency feature spaces of the source computing power device and the target computing power device, respectively, so as to obtain the key energy efficiency feature spaces of the source computing power device and the target computing power device. To eliminate the impact of differences in the units and numerical ranges of energy efficiency values ​​among labeled energy efficiency data from different computing power devices on the training of the energy efficiency model, the energy efficiency values ​​of the source and target computing power devices are normalized using Min-Max, as follows: in, This represents the energy efficiency value of computing equipment. and These represent the maximum and minimum energy efficiency values ​​in each energy efficiency dataset, respectively. This represents the normalized energy efficiency value, which ranges from [0,1].

4. The knowledge transfer-driven data center computing power energy efficiency modeling method according to claim 1, characterized in that, The computing power devices include general-purpose computing power devices with CPU chips as the computing core and intelligent computing power devices with AI chips as the computing core.

5. The knowledge transfer-driven data center computing power energy efficiency modeling method according to claim 1, characterized in that, Unlabeled energy efficiency data of the target computing power device is input into a trained energy efficiency model to evaluate the energy efficiency of the target computing power device, and the performance of the energy efficiency prediction model is quantitatively analyzed using mean square error and mean absolute error.

6. A knowledge transfer-driven data center computing power and energy efficiency modeling system, characterized in that, The knowledge transfer-driven data center computing power energy efficiency modeling method applied to any one of claims 1-5 includes a data acquisition module, a preprocessing module, an energy efficiency model construction module, an energy efficiency model training module, and an energy efficiency evaluation module; The data acquisition module is used to acquire energy efficiency data of the source computing power device and the target computing power device; the source computing power device includes a large amount of labeled energy efficiency data, and the target computing power device includes a small amount of labeled energy efficiency data; The preprocessing module is used to preprocess the energy efficiency data of the source computing power device and the target computing power device respectively to obtain their respective key energy efficiency feature spaces. The key energy efficiency feature space refers to the features most relevant to the energy efficiency of computing equipment. The energy efficiency model construction module is used to construct an energy efficiency model across computing power devices based on unbalanced optimal transmission. The objective function of the energy efficiency model is to obtain the optimal unbalanced optimal transmission plan and the energy efficiency modeling function. By optimizing the objective function, the differences in the energy efficiency edge distribution and conditional distribution of the source computing power device and the target computing power device are learned simultaneously. In turn, the valuable energy efficiency knowledge contained in the energy efficiency data of the source computing power device is used to improve the energy efficiency modeling accuracy of the target computing power device, which has only a small amount of labeled energy efficiency data. The energy efficiency model training module is used to train the energy efficiency model, learn the key energy efficiency feature spatial differences between heterogeneous computing devices, and thus realize effective energy efficiency knowledge transfer between heterogeneous computing devices. The energy efficiency assessment module is used to apply the trained energy efficiency model to the energy efficiency modeling of the target computing power device to be assessed, and to evaluate the energy efficiency performance.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor, which enables the at least one processor to perform the knowledge migration-driven data center computing power and energy efficiency modeling method as described in any one of claims 1-5.

8. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the knowledge transfer-driven data center computing power and energy efficiency modeling method according to any one of claims 1-5.

Images