Calculation power distribution method under fusion of energy network and calculation power network

Abstract

The invention relates to the technical field of computing, in particular to a computing power distribution method under fusion of an energy network and a computing power network. According to the technical scheme, the method comprises the steps of obtaining a current calculation instruction, and determining calculation power required by operation according to the current calculation instruction; and obtaining the residual computing power of each computing power node. Judging whether the residual computing power of each computing power node is greater than the computing power required by the current computing instruction, if so, judging that the computing power node is a first-level distributable computing power node, obtaining a power consumption form value, and calculating to obtain a budget computing power according to the computing power required by the current computing instruction and the computing power occupied by each first-level distributable computing power node; and obtaining a budget electricity consumption form value according to the budget calculation power and the corresponding electricity consumption form curve. And screening out a budget electricity consumption form value as a computing power node. According to the method, the computing power network can be analyzed from the perspective of the energy network, so that the energy consumption can be reduced, and the operation speed and the execution efficiency of the computing instruction can be increased.

Description

Technical Field

[0001] The present invention relates to the field of computing technology, and in particular to a computing power allocation method under the fusion of an energy network and a computing power network. Background Art

[0002] The computing network is a new type of Internet infrastructure that organically connects and coordinates computing resources around the world to form a powerful super network, thereby providing high-performance computing, storage, analysis and data processing services. The core technologies of the computing network include high-speed interconnection, resource sharing, load balancing, and optimized scheduling, aiming to meet the growing needs of users for diversified and high-performance computing.

[0003] The advantages of the computing network are mainly reflected in the following aspects: Improve computing efficiency: The computing network integrates idle computing resources from all over the world. Users can use these computing resources on demand anywhere and at any time, which greatly improves computing efficiency. Reduce costs: Users do not need to purchase expensive hardware equipment and invest a lot of money in infrastructure construction. They only need to pay a pay-as-you-go service fee to obtain powerful computing power.

[0004] Environmental protection and sustainable development: The computing network can promote sustainable development by optimizing energy use, reducing carbon dioxide emissions, supporting the development of clean energy. Flexibility and scalability: The computing network can quickly deploy and expand computing resources according to user needs and adapt to different business scenarios. Security: The computing network adopts strict network security measures to effectively protect users' data and computing resources from threats.

[0005] Energy network, also known as smart grid, is an advanced power transmission, distribution and management system that integrates various advanced technologies and equipment to achieve efficient, clean, safe and reliable use of energy. Compared with traditional power grids, energy network has a higher level of intelligence and stronger adaptability, and can effectively solve many problems in the process of energy utilization.

[0006] At present, the main features of energy networks include: High intelligence: The equipment and systems in the energy network have advanced sensors, communication equipment and computing capabilities, which can monitor and control the flow of energy in real time and achieve efficient energy management. Renewable energy: The energy network can fully accept various types of renewable energy, such as solar energy, wind energy, hydropower, etc., which helps to reduce carbon emissions and mitigate global climate change. Distributed energy: The energy network emphasizes the rational use of distributed energy resources, including distributed power generation equipment, energy storage equipment, etc., to achieve local production and consumption of energy and improve energy utilization. Efficient power transmission: The energy network adopts advanced power transmission technology and equipment, such as direct current transmission and flexible alternating current transmission, which can reduce energy loss during power transmission and improve energy transmission efficiency. Flexible scheduling: The energy network can adjust the output and input of energy in real time according to the user's electricity demand and weather conditions, achieve supply and demand balance, and reduce energy waste. Safe energy transmission: The equipment and systems in the energy network have high safety performance and can quickly cut off the power supply when a fault occurs to ensure personal and property safety. Close integration with information systems: Energy networks are combined with information systems to achieve remote monitoring, remote control and intelligent operation of energy equipment, providing users with convenient, safe and reliable energy services.

[0007] The existing computing power network allocates computing power according to demand and cannot be deeply integrated, which often results in excessive energy consumption and low computing power efficiency. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a computing power allocation method under the fusion of energy network and computing power network, which is used to analyze the energy and computing power of each computing power node and intelligently allocate the computing power. The computing power allocation method under the fusion of energy network and computing power network may include the following steps:

[0009] S1. Get the current computing instruction and determine the computing power C required for its operation based on the current computing instruction. In the computing execution program, there will be many computing instructions to be transmitted, and the specific computing program is also executed according to the transmitted computing execution, so the computing instruction is the basis of program execution.

[0010] S2. Obtain the remaining computing power C' of each computing power node;

[0011] S3, determine whether the remaining computing power C' of each computing power node is greater than the computing power C required by the current computing instruction. If yes, execute S4, if not, execute S5. By comparing the values, you can select computing power nodes that meet the computing power requirements. For the consideration of computing power nodes, satisfying computing power is the primary factor. Through the judgment, you can make a preliminary screening to ensure that the allocated computing power nodes meet the computing power requirements;

[0012] S4. Determine that this computing power node is a first-level allocable computing power node, and obtain the power consumption form value E of each first-level allocable computing power node k , where k is the number of each first-level allocable computing power node. k It reflects the working output capacity of a computing power node. It is necessary to consider not only the output of capacity but also the energy consumption, so as to make a comprehensive evaluation. When the computing power node executes computing instructions, the computing power node consumes a certain amount of electricity, but there will be some loss when the consumed electricity is used for computing power execution. This ratio value is not the same for all computing power nodes, but varies with the performance of each computing power node itself, the running time, the running environment, the number of computing instructions that have been run, etc.

[0013] S5. Determine that this computing power node is a first-level discarded computing power node and do not process it. Not processing it means performing the first round of screening and exclusion. The purpose of this determination is that if this computing power node cannot meet the computing power requirements, we will not consider this computing power allocation;

[0014] S6: Based on the computing power C required by the current computing instruction and the computing power G occupied by each first-level allocable computing power node k Calculate the budget computing power G k ', then according to the budget computing power G k ' and the corresponding power consumption curve to obtain the budgeted power consumption value E k '. The budget computing power G k ' is the computing power after the computing power is allocated to the first-level allocable computing power node numbered k;

[0015] S7. Filter out the budgeted electricity consumption value E k 'The first-level allocable computing power node corresponding to the maximum value is then used as the allocated execution computing power node for the current computing instruction.

[0016] Preferably, the method for obtaining the computing power C required for the execution of the current computing instruction may specifically include: decomposing the computing instruction to obtain one or more subtasks, and then obtaining the computing power C through subtask calculations.

[0017] Preferred: Hashrate Where i and j are the numbers of the subtasks in the current computing instruction, and i≠j, I is the total number of subtasks in the current computing instruction, i, j = 1, 2, ..., I; c i is the computing power required for the calculation of the subtask numbered i, c ij ' is the computational interference factor of the subtask numbered j on the subtask numbered i; L ij is the length of the interference interval between the subtask numbered j and the subtask numbered i in the calculation instruction.i The units can be measured in FLOPS, hash rate or TPS, etc., and their values ​​can be obtained through their specific content, of course, they can also be obtained through experiments, specifically through a pre-prepared subtask-computational amount information table. The subtasks and their computational amounts are one-to-one corresponding, and the computational amount corresponding to each subtask can be obtained through advance experiments, and then constructed into an information table.

[0018] In the embodiment of the present invention, the calculation interference factor c ij ' is the mutual interference between the two subtasks during operation. Its value is generally between -0.1 and 0.1, and theoretically it is 0. Its specific value can also be obtained from the subtask compatibility interference table factor information table. c ij ' can be positive or negative. ij When the value of ' is positive, it means that the computational effort of the two subtasks will drag down each other when they are executed. ij When the value of ' is negative, it means that the two subtasks will promote each other. Both the subtask-computational information table and the subtask compatibility interference table factor information table can be obtained through experiments, so before doing actual work, a large amount of basic data needs to be collected and constructed. The computing power calculated by this method is accurate, not only considering its own computing needs but also considering the mutual interference between subtasks. This method can obtain computing data through a basic database, which is convenient and fast to calculate, with little interference. It can be applied to the analysis and calculation of various computing instructions and has a wide range of applications.

[0019] Preferably: the interference interval length L ij The specific method of obtaining the subtask can be as follows: obtain the subtask between subtasks numbered j and i through the current calculation instruction and name it as the interval subtask, and then obtain the calculation amount c of the interval subtask by searching the obtained subtask-calculation amount information table n , where n is the sorting number of the interval subtask, and then according to the calculation amount c of the interval subtask n Calculate the interference interval length L ij .

[0020] Preferred: Interference interval length Where n is the interval subtask ranking number between subtasks numbered j and i, N is the total number of subtasks between subtasks numbered j and i, n = 1, 2, ..., N; c 标The standard calculation amount can generally be calculated using the average computing power of the subtasks. This method can be used to calculate in accordance with the actual situation of interference. The closer the two subtasks are, the greater the interference. When the two subtasks are farther apart, the interference will be greatly reduced. Therefore, setting the interference interval length for evaluation is more accurate and completely consistent with the actual interference.

[0021] Preferably: The specific method for obtaining the remaining computing power can be Among them C 有效 is the effective computing power of the corresponding computing power node, and its value can be obtained by detecting the current computing power node, m is the number of each computing instruction running in the computing power node, and M is the total number of computing instructions running in the computing power node, m = 1, 2, ..., M; C m is the computing power required for the computation instruction numbered m, C m ' is the computing power loss of the computing instruction numbered m. This method can effectively calculate the operating status of each computing power node, the calculation is accurate and takes into account the performance and operating performance of each node itself, and the calculation is convenient and accurate.

[0022] Preferred: The effective computing capacity C of the computing power node 有效 The maximum output computing power of the current computing power node, which needs to take into account the specific performance of the node, the operating environment, etc. The theoretical computing power of general equipment is known and can be obtained through equipment parameters. 有效 =ηC, where C is the theoretical computing power of the computing power node, which can be obtained through its device parameters and this value will not change. η is the effective coefficient, and its value is generally 0.5-1, which depends on the actual situation.

[0023] Preferred: Effectiveness coefficient Where s is the loss coefficient number, η s is the loss rate of number s, S is the total number of loss coefficients. s=n=1、2、…、S.

[0024] Preferably: the loss coefficient may include but is not limited to life loss rate and usage condition loss rate.

[0025] Preferably: the power consumption value Among them G k It is the computing power occupied by the first-level allocable computing power node numbered k, which can be obtained through the above data calculation. The details will not be elaborated here. k It is the current electrical power of the first-level allocable computing power node numbered k.

[0026] By introducing the power consumption form value, the present invention can effectively evaluate the input and output status of the computing power node. This status value can not only evaluate the application efficiency of electric energy, but also effectively protect the computing power node to avoid its overload operation.

[0027] Optimally, this computing power is the sum of the computing power C required by the current computing instruction of the computing power node numbered k and the remaining computing power C' of each computing power node. That is, the budget computing power G k '=C+G k .

[0028] Preferably, the power consumption curve can be pre-set according to the performance of each computing node. Since the performance of each computing node is different, the power consumption curves therein may also be different, so it is necessary to pre-prepare the power consumption curve according to the actual situation and update it in real time.

[0029] The specific method for constructing the power consumption form curve may include: constructing a plane coordinate system, wherein the horizontal coordinate of the plane coordinate system is the occupied computing power, and the vertical coordinate is the power consumption form value. Then, each computing power node is detected, and the calculation instructions of each computing power value are executed on each computing power node, and then the corresponding power consumption form value is calculated, and each power consumption form value and its corresponding computing power value are used to construct the power consumption form curve of each computing power node. Of course, it can also be constructed through historical operation data, which will not be described in detail here. Then, the horizontal coordinate is the budgeted computing power G k ', find the power consumption curve corresponding to the computing node, and the corresponding ordinate value is the budget power consumption value E corresponding to the computing node k '.

[0030] Technical effects and advantages of the present invention: The present invention allocates the computing power of the computing power network based on the energy network. When the computing power nodes perform computing power execution, energy consumption will be generated. The more serious these energy losses are, the lower the efficiency of computing power execution. The present invention allocates computing power through the fusion calculation of the energy network and the computing power network, and can analyze the computing power network from the perspective of the energy network, which can not only reduce energy consumption, but also increase the running speed and execution efficiency of computing instructions, ensure the highest efficiency operation of computing power nodes, avoid excessive loss of computing power nodes and damage to equipment, and reduce its service life. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a structural block diagram of the computing power allocation method under the fusion of energy network and computing power network proposed in the present invention.

[0032] Figure 2 This is a structural block diagram of the computing power allocation method under the fusion of energy network and computing power network proposed in the present invention. DETAILED DESCRIPTION

[0033] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0034] Example 1

[0035] refer to Figure 1 In this embodiment, a computing power allocation method under the fusion of energy network and computing power network is proposed, which is used to analyze the energy and computing power of each computing power node and intelligently allocate the computing power. The computing power allocation method under the fusion of energy network and computing power network may include the following steps:

[0036] S1. Get the current computing instruction, and determine the computing power C required for its operation according to the current computing instruction. In the computing execution program, there will be many computing instructions for transmission, and the specific computing program is also executed according to the transmitted computing execution, so the computing instruction is the basis for program execution. The computing instruction can be an operation signal, which can be an instruction sent by the switchboard or an operation instruction processed by the user or each computing node signal processing countermeasure, and the execution object of the instruction is each computing node.

[0037] The computing power node can be any computing or operating device at the computing power end in the current energy computing power network, which can be a server, and is not limited here.

[0038] The method for obtaining the computing power C required for the current computing instruction to run may specifically include: each computing instruction is composed of one or more subtasks to be calculated. First, when a computing instruction is received, the computing instruction needs to be decomposed to obtain one or more subtasks. The decomposition standard is to obtain each subtask as a basic subtask. The basic subtask is a subtask for us to pre-construct the content of the information table. The details will not be repeated here.

[0039] Then we can calculate the computing power based on each subtask Where i and j are the numbers of the subtasks in the current computing instruction, and i≠j, I is the total number of subtasks in the current computing instruction, i, j = 1, 2, ..., I; c i is the computing power required for the calculation of the subtask numbered i, c ij ' is the computational interference factor of the subtask numbered j on the subtask numbered i. ijTo calculate the interference interval length of the subtask numbered j in the instruction to the subtask numbered i, its value can be obtained by looking up a table or calculating. i The units can be measured in FLOPS, hash rate or TPS, etc. Its value can be obtained through its specific content, of course, it can also be obtained through experiments, and specifically through a pre-prepared subtask-computational information table. The subtask and its computational amount are one-to-one corresponding. We can obtain the computational amount corresponding to each subtask through advance experiments, and then construct an information table. Its specific value is not the subject of protection of this application, so it will not be repeated here. The computational interference factor c ij ' is the mutual interference between the two subtasks during operation. Its value is generally between -0.1 and 0.1, and theoretically is 0. Its specific value can also be obtained from the subtask compatibility interference table factor information table;

[0040] Furthermore, in this embodiment, c ij ' can be positive or negative. ij When the value of ' is positive, it means that the computational effort of the two subtasks will drag down each other when they are executed. ij When the value of ' is negative, it means that the two subtasks will promote each other. Both the subtask-calculation amount information table and the subtask compatibility interference table factor information table can be obtained through experiments. Therefore, before carrying out actual work, a large amount of basic data needs to be collected and constructed. The specific content is not the content that needs to be protected in this application and will not be elaborated here.

[0041] The computing power calculated by the present invention through this method is accurate, not only considering the computing needs of the present invention but also considering the mutual interference between subtasks. This method can obtain computing data through a basic database, the calculation is convenient and fast, the interference is small, and it can be applied to the analysis and calculation of various computing instructions, and has a wide range of applications.

[0042] Furthermore, in the embodiment of the present disclosure, the interference interval length L ij The method of obtaining can be obtained by table lookup, of course, it can also be obtained by calculation. The table lookup method is not described here. The interference interval length L ij The specific method of obtaining by calculation can be as follows: obtain the subtask between subtasks numbered j and i through the current calculation instruction and name it as the interval subtask, and then obtain the calculation amount c of the interval subtask by searching the obtained subtask-calculation amount information table. n , where n is the sorting number of the interval subtask, and then the interference interval length is calculated Where n is the interval subtask ranking number between subtasks numbered j and i, N is the total number of subtasks between subtasks numbered j and i, n = 1, 2, ..., N; c 标 It is the standard computing amount, which can generally be calculated using the average computing power of subtasks.

[0043] Furthermore, when |ij|=1, n=N=0, then L ij The specific value is 1, and the specific value is not calculated here. This method can be used to calculate in accordance with the actual situation of interference. The closer the two subtasks are, the greater the interference. When the two subtasks are farther apart, the interference will be greatly reduced. Therefore, setting the interference interval length for evaluation is more accurate and completely consistent with the actual interference.

[0044] Further, in the disclosed embodiment, S2, obtain the remaining computing power C' of each computing power node. In actual work, each computing power node has a certain computing power, which can be obtained according to the performance of the computing power node. Different computing power devices, different models, different usage time, and different usage environments may have different computing power. However, for the same model of computing power equipment, its theoretical computing power can generally be understood as the computing power at the factory, and its value is a fixed value. However, in the case of some loss factors, its computing power will decrease.

[0045] Among them, the loss factors include but are not limited to its running time and usage environment. Of course, there are other situations that cause its computing power to change, which will not be elaborated here.

[0046] In the disclosed embodiment, the remaining computing power may be obtained through experience or testing experiments. To obtain it through experience, the staff is required to have long-term operation and management experience of the system, which will not be elaborated here.

[0047] It is obtained through testing experiments. It is necessary to frequently test each computing power node to update each data. Of course, it can also be obtained through calculation. The specific method for obtaining the remaining computing power can be the remaining computing power Among them C 有效 is the effective computing power of the corresponding computing power node, and its value can be obtained by detecting the current computing power node, m is the number of each computing instruction running in the computing power node, and M is the total number of computing instructions running in the computing power node, m = 1, 2, ..., M; C m is the computing power required for the computation instruction numbered m, C m ' is the computing power loss of the computing instruction numbered m, C m and C m 'All can be obtained through experiments or tests, and the details will not be elaborated here.

[0048] Therefore, the present invention can effectively calculate the operating status of each computing power node through this method, the calculation is accurate and takes into account the performance and operating performance of each node itself, and the calculation is convenient and accurate. Of course, it can also be obtained by direct detection, which will not be described in detail here. The effective computing power capacity C of the computing power node 有效 The maximum output computing power of the current computing power node, which needs to take into account the specific performance of the node, the operating environment, etc. The theoretical computing power of general equipment is known and can be obtained through equipment parameters. 有效 =ηC, where C is the theoretical computing power of the computing power node, which can be obtained through its device parameters and this value will not change. η is the effective coefficient, and its value is generally 0.5-1, which depends on the actual situation.

[0049] For example, if the effective coefficient η is 0.90 and the theoretical computing power C of the computing power node is 1T / s, then the effective computing power C can be calculated. 有效 =ηC=0.9×1T / s=0.9T / s. Of course, this is a simple example and is not universal. The specific details will not be repeated here. The effective coefficient η can be obtained by testing. Of course, this requires frequent testing. The effective coefficient can also be obtained by calculation. Where s is the loss coefficient number, η s is the loss rate of number s, and S is the total number of loss coefficients.

[0050] Furthermore, the loss coefficient may include but is not limited to the life loss rate, the use condition loss rate, etc. The effective coefficient will gradually decrease with the use time and will also decrease under certain circumstances. For example, the total number of loss coefficients is 2, and the life loss rate η1 of a server is 0.05, and the use condition loss rate η2 is 0.08. Then the effective coefficient Of course, this is just a simple example and may not be universal, so I will not go into details here.

[0051] S3. Determine whether the remaining computing power C' of each computing power node is greater than the computing power C required by the current computing instruction. If yes, execute S4; if not, execute S5. By comparing the values, the computing power nodes that meet the computing power requirements can be screened. For the consideration of computing power nodes, satisfying the computing power is the primary factor. Preliminary screening can be performed through the judgment to ensure that the allocated computing power nodes meet the computing power requirements.

[0052] S4. Determine that this computing power node is a first-level allocable computing power node, and obtain the power consumption form value E of each first-level allocable computing power node k , where k is the number of each first-level allocable computing power node. kIt reflects the working output capacity of a computing power node. It not only takes into account the output of capacity, but also the energy consumption, so as to make a comprehensive evaluation. When the computing power node executes the calculation instruction, the computing power node consumes a certain amount of electricity, but the consumed electricity will also have some loss while executing the calculation power. This ratio value is not the same for all computing power nodes, but varies with the performance of each computing power node itself, the running time, the running environment, the number of computing instructions that have been run, etc. Among them G k is the computing power occupied by the first-level allocable computing power node numbered k, which can be obtained through the above data calculation. The details will not be elaborated here. k It is the current electric power of the first-level allocable computing power node numbered k. Its value can be obtained by detecting it through an electric meter, and the details are not repeated here. By introducing the power consumption form value, the input and output status of the computing power node can be effectively evaluated. This status value can not only evaluate the application efficiency of electric energy, but also effectively protect the computing power node and avoid its overload operation.

[0053] S5. This computing power node is determined to be a first-level abandoned computing power node, and no processing is performed on it. Not processing it means the first round of screening and exclusion. The purpose of this determination is that if this computing power node cannot meet the computing power requirements, we will not consider the computing power allocation this time. The details are not repeated here. If all computing power nodes are first-level abandoned computing power nodes, there will be no computing power nodes that meet the computing power requirements. This situation basically does not exist, but when it occurs, an alarm can be processed, and the staff can perform manual intervention after receiving the alarm. The details are not repeated here.

[0054] S6: Based on the computing power C required by the current computing instruction and the computing power G occupied by each first-level allocable computing power node k Calculate the budget computing power G k ', then according to the budget computing power G k ' and the corresponding power consumption curve to obtain the budgeted power consumption value E k '. The budget computing power G k ' is the computing power of the first-level allocable computing power node numbered k after computing power allocation. This computing power is the sum of the computing power C required for the current computing instruction of the computing power node numbered k and the remaining computing power C' of each computing power node. That is, the budget computing power G k '=C+G kOf course, there may be other algorithms, which will not be described in detail here. The power consumption curve can be pre-set according to the performance of each computing power node. Since the performance of each computing power node is different, the power consumption curves that exist may not be the same, so it is necessary to pre-prepare the power consumption curve according to the actual situation and update it in real time.

[0055] For further information, please refer to Figure 2 , the specific construction method of the power consumption form curve may include: constructing a plane coordinate system, the horizontal coordinate of the plane coordinate system is the occupied computing power, and the vertical coordinate is the power consumption form value. Then, each computing power node is detected, and the calculation instructions of each computing power value are run on each computing power node, and then the corresponding power consumption form value is calculated, and each power consumption form value and its corresponding computing power value are used to construct the power consumption form curve of each computing power node. Of course, it can also be constructed through historical operation data, which will not be described in detail here. Then, the horizontal coordinate is the budget computing power G k ', find the power consumption curve corresponding to the computing node, and the corresponding ordinate value is the budget power consumption value E corresponding to the computing node k '.

[0056] S7. Filter out the budgeted electricity consumption value E k 'The first-level allocable computing power node corresponding to the maximum value is then used as the allocated execution computing power node for the current computing instruction. This method can complete computing power allocation through energy status, evaluate the computing power network and energy network of each computing power node, and analyze the computing power network from the perspective of the energy network. This can not only reduce energy consumption, but also increase the running speed and execution efficiency of computing instructions, ensure the highest efficiency of computing power nodes, avoid excessive loss of computing power nodes and damage to equipment, and reduce their service life.

[0057] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. The computing power allocation method under the fusion of energy network and computing power network is characterized by: The computing power allocation method under the fusion of the energy network and the computing power network includes: S1. Obtain the current computing instruction and determine the computing power C required for its operation according to the current computing instruction; S2. Obtain the remaining computing power C' of each computing power node; S3, determine whether the remaining computing power C' of each computing power node is greater than the computing power C required by the current computing instruction, if yes, execute S4, if no, execute S5; S4. Determine that this computing power node is a first-level allocable computing power node, and obtain the power consumption form value E of each first-level allocable computing power node k , where k is the number of each first-level allocatable computing power node; S5. Determine that the computing power node is a first-level discarded computing power node; S6: Based on the computing power C required by the current computing instruction and the computing power G occupied by each first-level allocable computing power node k Calculate the budget computing power G k ', then according to the budget computing power G k ' and the corresponding power consumption curve to obtain the budgeted power consumption value E k ', the budget computing power G k ' is the computing power after the computing power is allocated to the first-level allocable computing power node numbered k; S7. Filter out the budgeted electricity consumption value E k 'The first-level allocable computing power node corresponding to the maximum value is then used as the allocated execution computing power node for the current computing instruction.

2. The computing power allocation method under the fusion of energy network and computing power network according to claim 1 is characterized in that: The method for obtaining the computing power C required for the execution of the current computing instruction includes: decomposing the computing instruction to obtain one or more subtasks, and then obtaining the computing power C through the subtask calculation.

3. The computing power allocation method under the fusion of energy network and computing power network according to claim 1 is characterized in that: Remaining computing power Among them C 有效 is the effective computing power of the corresponding computing power node; m is the number of each computing instruction being run in the computing power node, and M is the total number of computing instructions being run in the computing power node; m = 1, 2, ..., M; C m is the computing power required for the computation instruction numbered m, C m ' is the computing power loss of the computing instruction numbered m.

4. The computing power allocation method under the fusion of energy network and computing power network according to claim 3 is characterized in that: Effective computing power C 有效 =ηC, where C is the theoretical computing power of the computing power node and η is the effective coefficient.

5. The computing power allocation method under the fusion of energy network and computing power network according to claim 4 is characterized in that: Effective coefficient Where s is the loss coefficient number, η s is the loss rate of number s, S is the total number of loss coefficients; s=n=1, 2,…, S.

6. The computing power allocation method under the fusion of energy network and computing power network according to claim 5 is characterized in that: The loss coefficient includes life loss rate and usage condition loss rate.

7. The computing power allocation method under the fusion of energy network and computing power network according to claim 1 is characterized in that: The power consumption value Among them G k is the computing power already occupied by the first-level allocable computing power node numbered k, P k It is the current electrical power of the first-level allocable computing power node numbered k.

8. The computing power allocation method under the fusion of energy network and computing power network according to claim 7 is characterized in that: m is the number of each running computing instruction in the computing power node, and M is the total number of each running computing instruction in the computing power node; m = 1, 2, ..., M; C m is the computing power required for the computation instruction numbered m, C m ' is the computing power loss of the computing instruction numbered m.

9. The computing power allocation method under the fusion of energy network and computing power network according to claim 8 is characterized in that: Budgeted computing power G k '=C+G k .

10. The computing power allocation method under the fusion of energy network and computing power network according to claim 7 is characterized in that: The specific method for constructing the power consumption pattern curve includes: constructing a plane coordinate system, wherein the horizontal coordinate of the plane coordinate system is the occupied computing power, and the vertical coordinate is the power consumption pattern value; then detecting each computing power node, running the calculation instructions of each computing power value on each computing power node, and then calculating the corresponding power consumption pattern value, and constructing the power consumption pattern curve of each computing power node by combining each power consumption pattern value with its corresponding computing power value.

Images