A computing power network computing resource optimization scheduling method and follow-up control device

Abstract

The application discloses a kind of computing power network computing resource optimization scheduling method and follow-up control device, applied to computing power network technical field, the topological structure of target hybrid computing power computing resource network is acquired;According to the topological structure, and the algorithm of preset constraint index, determine constraint index measure value;According to the order of constraint index measure value size, the order of node is determined, and ordering result is obtained;According to the ordering result, computing power scheduling control is carried out;Primary index includes computing power network information flow circulation timeliness, computing power node control quality accuracy, computing resource effective computing power adaptability, computing power bearing service scene efficiency, software running process optimization efficiency, secondary index includes the order queue computing power overflow optimization degree corresponding to electronic computing power node, computing power data queue time sequence optimization degree, quantum server effective computing power capability corresponding to quantum computing power node.Through the above index, computing power resource can be controlled comprehensively, and computing power resource global optimization scheduling is realized.

Description

Technical Field

[0001] This invention relates to the field of computing network technology, and in particular to a computing network computing resource optimization scheduling method, a computing network computing resource optimization scheduling device, a hybrid computing network computing resource optimization scheduling equipment, and a computer-readable storage medium. Background Technology

[0002] Global optimization scheduling and control of computing resources in computing networks based on quantum and electronic computers is a global optimization decision-making problem based on effective computing power. Currently, there is no mature technology for evaluating the global optimization of effective computing power under the "hybrid control" condition of quantum and electronic computers. Existing computing networks have not yet achieved a hybrid configuration of quantum and electronic computers, and there is no algorithm for global optimization decision-making of computing resource control in computing networks based on quantum and electronic computers. Therefore, how to provide a global optimization method for computing resource control in hybrid computing networks is an urgent problem that needs to be solved by those skilled in the art. Summary of the Invention

[0003] The purpose of this invention is to provide a method for optimizing and scheduling computing resources in a computing power network, which can perform global optimization and control of computing resources in a hybrid computing power network. Another purpose of this invention is to provide a device for optimizing and scheduling computing resources in a computing power network, a device for optimizing and scheduling computing resources in a hybrid computing power network, and a computer-readable storage medium, which can perform global optimization and control of computing resources in a hybrid computing power network.

[0004] To address the aforementioned technical problems, this invention provides a method for optimizing the scheduling of computing resources in a computing power network, comprising:

[0005] Obtain the topology of the target hybrid computing resource network;

[0006] Based on the topology and the algorithm for the preset constraint index, the constraint index measure value is determined.

[0007] The order of the nodes is determined according to the order of the constraint index measure values, and the sorting result is obtained;

[0008] Based on the sorting results, the target hybrid computing resource network is subjected to global optimization scheduling control of computing resources.

[0009] The constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators:

[0010] The primary indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation. The secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes.

[0011] Optionally, when the constraint index includes the timeliness of information flow in the computing power network, determining the constraint index measure value based on the topology and the algorithm for the preset constraint index includes:

[0012] Based on the topology of the subnet corresponding to the node and the connectivity of the node, the time-sensitive microstate is determined;

[0013] Based on the aforementioned time-dependent microstate, determine the total time-dependent connectivity of the subnet corresponding to the node;

[0014] Based on the total timeliness correlation, determine the timeliness microstate realization probability value of the subnet corresponding to the node;

[0015] Based on the time-dependent microstate realization probability value, determine the time-dependent entropy of the subnet corresponding to the node;

[0016] Based on the maximum connectivity in the total time-sensitive connectivity and the time-sensitive entropy, determine the maximum time-sensitive entropy of the subnet corresponding to the preferred node;

[0017] Based on the time-efficiency entropy and the maximum time-efficiency entropy, determine the time-efficiency metric value of the subnet corresponding to the node;

[0018] Based on the timeliness metric value, a timeliness vector is formed for each subnet corresponding to the node, which serves as the constraint index metric vector for the node;

[0019] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0020] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0021] Optionally, when the constraint metric includes the accuracy of the control quality of computing nodes, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes:

[0022] Based on the topology of the subnet corresponding to the node and the corresponding maximum span, determine the total number of quality microstates of the node;

[0023] Based on the total number of quality microstates, determine the probability value of the realization of the quality microstate of the subnet corresponding to the node;

[0024] Based on the probability value of realization of the quality microstate, determine the quality entropy and maximum quality entropy of the subnet corresponding to the node;

[0025] Based on the quality entropy and the maximum quality entropy, determine the quality metric value of the subnet corresponding to the node;

[0026] The quality vector of each subnet corresponding to each node is formed based on the quality metric value, and serves as the constraint index metric vector of the node.

[0027] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0028] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0029] Optionally, when the constraint metric includes the adaptability of effective computing power to computing resources, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes:

[0030] Based on the topology of the subnet corresponding to the node, determine the computing power temporal entropy of the electronic node in the node;

[0031] Based on the topology of the subnet corresponding to the node, determine the computing power overflow entropy of the electronic node in the node;

[0032] Based on the topology of the subnet corresponding to the node, determine the quantum volume of the quantum node in the node;

[0033] The computing power data queue timing optimization degree is determined based on the computing power time sequence entropy;

[0034] The optimization degree of sequential queue computing power overflow is determined based on the computing power overflow entropy.

[0035] The effective computing power of the quantum server is determined based on the quantum volume.

[0036] The effective computing power adaptability measure value of the electronic node is determined based on the timing optimization degree of the computing power data queue, the computing power overflow optimization degree of the sequential queue, and the preset weights.

[0037] Based on the effective computing power capability of the quantum server in the quantum computing node, determine the effective computing power adaptability measure value of the quantum node's computing resources;

[0038] The computing resource effective computing power adaptability vector formed based on the computing resource effective computing power adaptability measure value is used as the constraint index measure vector of the node.

[0039] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0040] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0041] Optionally, when the constraint metric includes the computing power data queue timing optimization degree, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes:

[0042] Based on the topology of the subnet corresponding to the node, determine the computing power temporal entropy of the node;

[0043] The computing power data queue timing optimization degree is determined based on the computing power time sequence entropy;

[0044] The time-series optimization degree of the computing power data queue is used as the constraint index measure value.

[0045] Optionally, when the constraint metric includes the optimization degree of sequential queue computing power overflow, determining the constraint metric measure value based on the topology and the algorithm of the preset constraint metric includes:

[0046] Based on the topology of the subnet corresponding to the node, determine the computing power overflow entropy of the node;

[0047] The optimization degree of sequential queue computing power overflow is determined based on the computing power overflow entropy.

[0048] The optimization degree of the sequential queue computing power overflow is used as the measurement value of the node constraint index.

[0049] Optionally, when the constraint metric includes the effective computing power of the quantum server, determining the constraint metric measure value based on the quantum computing node and the algorithm for the preset constraint metric includes:

[0050] The quantum volume of the quantum computing power node is determined based on the quantum computing power node corresponding to the node.

[0051] The effective computing power of the quantum server is determined based on the quantum volume.

[0052] The effective computing power of the quantum server is used as a constraint index measure of the quantum computing node.

[0053] Optionally, when the constraint metric includes the performance of computing power in service scenarios, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes:

[0054] Based on the baseline throughput value measured on the load of each scenario, the actual training throughput of the business scenario in the node, and the weight of the load of the business scenario, the relative entropy of the computing power carrying the business scenario is determined.

[0055] Based on the relative entropy of the computing power carrying service scenario, determine the performance measurement value of the computing power carrying service scenario;

[0056] Based on the computing power carrying service scenario performance measurement value, a computing power carrying service scenario performance vector of the subnet corresponding to the node is formed, which serves as the constraint index measurement vector of the node.

[0057] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0058] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0059] Optionally, when the constraint index includes software runtime optimization performance, determining the constraint index measure value based on the topology and the algorithm for the preset constraint index includes:

[0060] Based on the actual performance standard value of the given test set on the software system, the actual performance of the software under test in the node, and the weight of each software under test, determine the relative entropy for software operation tuning.

[0061] Based on the relative entropy of the software operation process, determine the optimization performance measure value of the software operation process;

[0062] Based on the software operation process optimization performance measurement value, a software operation process optimization performance vector for the subnet corresponding to the node is formed, which serves as the constraint index measurement vector for the node.

[0063] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0064] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0065] Optionally, when the constraint indicators include at least two of the following: the timeliness of information flow in the computing power network, the accuracy of the quality control of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, the step of determining the constraint indicator measure value based on the topology and the algorithm of the preset constraint indicators includes:

[0066] A comprehensive evaluation matrix is ​​formed based on the constraint index measure vector;

[0067] Based on the comprehensive evaluation matrix and the weights corresponding to each constraint index, the preference vector is determined as the global optimal solution vector;

[0068] The global optimal solution vector is used as the final computing power scheduling control constraint index measurement vector;

[0069] The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes:

[0070] The order of the nodes is determined based on the computing power scheduling control constraint index metric vector, and the sorting result is obtained.

[0071] The present invention also provides a computing power network computing resource optimization and scheduling device, comprising:

[0072] The node acquisition module is used to acquire the topology of the target hybrid computing resource network;

[0073] The constraint index module is used to determine the constraint index measure value based on the topology and the algorithm of the preset constraint index.

[0074] The computing power ranking module is used to determine the order of the nodes according to the order of the constraint index measurement values, and obtain the ranking result;

[0075] The scheduling and control module is used to perform global optimization scheduling and control of computing resources for the target hybrid computing resource network based on the sorting results.

[0076] The constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators:

[0077] The primary indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation. The secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes.

[0078] The present invention also provides a computing resource control and optimization device for a hybrid computing network, the device comprising:

[0079] Memory: Used to store computer programs;

[0080] Processor: Used to implement the computational resource control optimization method for the hybrid computing network as described in any of the preceding claims when executing the computer program.

[0081] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the computing resource control and optimization method for the hybrid computing network as described in any of the preceding claims.

[0082] The present invention provides a method for optimizing and scheduling computing resources in a computing power network, comprising: acquiring the topology of a target hybrid computing power network; determining constraint index measures based on the topology and an algorithm for preset constraint indicators; determining the order of nodes based on the order of constraint index measures to obtain a sorting result; and performing global optimization scheduling control of computing resources in the target hybrid computing power network based on the sorting result. The constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators: primary indicators include the timeliness of information flow in the computing power network, the accuracy of computing node control quality, the adaptability of computing resources to effective computing power, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation; secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes.

[0083] Based on the computing power network, the computing power resources of the hybrid computing power network can be comprehensively controlled by considering the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, thereby achieving global optimization scheduling and control of computing power resources.

[0084] The present invention also provides a computing power network computing resource optimization and scheduling device, a hybrid computing power network computing resource optimization and scheduling device, and a computer-readable storage medium, which have the same beneficial effects as described above, and will not be described in detail here. Attached Figure Description

[0085] To more clearly illustrate the technical solutions of the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0086] Figure 1 This is a schematic diagram of a computing power network structure;

[0087] Figure 2 A flowchart illustrating a method for optimizing and scheduling computing resources in a computing network, as provided in an embodiment of the present invention;

[0088] Figure 3 A flowchart illustrating the first specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0089] Figure 4 A flowchart illustrating the second specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0090] Figure 5 A flowchart illustrating the third specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0091] Figure 6 A flowchart illustrating the fourth specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0092] Figure 7 A flowchart illustrating the fifth specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0093] Figure 8 A flowchart illustrating the sixth specific method for optimizing and scheduling computing resources in a computing network provided in this embodiment of the invention;

[0094] Figure 9 This is a structural block diagram of a computing power network computing resource optimization and scheduling device provided in an embodiment of the present invention;

[0095] Figure 10 This is a structural block diagram of a computing power network computing resource optimization and scheduling device provided in an embodiment of the present invention. Detailed Implementation

[0096] The core of this invention is to provide a global optimization scheduling method for computing resources in a computing power network. Currently, there is no mature technology for effective global optimization evaluation of computing power under the "hybrid control" conditions of quantum and electronic computers. Existing computing power networks have not yet achieved a hybrid configuration of quantum and electronic computers, and there is still no algorithm for global optimization decision-making for computing resource control in hybrid computing power networks based on quantum and electronic computers.

[0097] The present invention provides a global optimization scheduling method for computing resources in a computing power network, comprising: acquiring the topology of a target hybrid computing power network; determining constraint index measure values ​​based on the topology and a preset constraint index algorithm; determining the order of nodes based on the order of constraint index measure values ​​to obtain a sorting result; and performing global optimization scheduling control of computing resources in the target hybrid computing power network based on the sorting result. The constraint indexes include any of the following primary and secondary indicators, or any combination of primary indicators: primary indicators include the timeliness of information flow in the computing power network, the accuracy of computing node control quality, the adaptability of computing resources to effective computing power, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation; secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes.

[0098] Based on the computing power network, the computing power resources of the hybrid computing power network can be comprehensively allocated and controlled by considering the timeliness of information flow in the computing power network, the accuracy of control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, so as to achieve global optimization scheduling of computing power resources.

[0099] To enable those skilled in the art to better understand the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0100] Please refer to Figure 1 as well as Figure 2 , Figure 1 This is a schematic diagram of a computing power network structure; Figure 2 This is a flowchart of a computing power network computing resource optimization scheduling method provided in an embodiment of the present invention.

[0101] See Figure 1 as well as Figure 2 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0102] S101: Obtain the topology of the target hybrid computing resource network.

[0103] In this step, the entire target hybrid computing resource network is scanned to obtain the tags of each computing node within the network. These tags characterize whether a node is a quantum computer node or an electronic computer node. Furthermore, the tags can also represent the location information of each computing node, such as its specific location within a subnet, the subnet's cut, etc. It should be noted that the subnets can be customized according to actual requirements. For example, they can be defined based on geographical information or by dividing the target hybrid computing resource network according to computing power and service functions. The division criteria can be set according to the specific circumstances.

[0104] S102: Determine the constraint index measure value based on the topology and the algorithm of the preset constraint index.

[0105] In this embodiment of the invention, the constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators: the primary indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation process; the secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes.

[0106] The calculation process for each of the above-mentioned constraint indicators will be described in detail in the following embodiments of the invention, and will not be repeated here. It should be noted that when multiple parameters are selected as performance indicators, global optimization can be performed based on the comprehensive evaluation value of multiple primary performance indicators, the specific details of which will be described in detail in the following real-world examples of the invention.

[0107] S103: Determine the order of nodes based on the order of constraint index measure values ​​to obtain the sorting result.

[0108] After the above steps, each node has a corresponding constraint index measure value. The nodes can be sorted according to the size of the constraint index measure value to obtain the sorting result. The sorting result is usually sorted from largest to smallest, and the nodes at the top are usually considered to be the optimal scheduling and allocation nodes.

[0109] S104: Perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0110] In this step, based on the above sorting results, the required nodes are selected from front to back for the global optimization and scheduling control of the target hybrid computing resource network. The sorting results represent the nodes selected for optimization based on computing resources; however, a high-ranking node may not necessarily be available to execute tasks. For example, if a high-ranking node also has a high load, nodes that meet the computing power service requirements will be selected according to the above sorting results. Specifically, the global optimization and scheduling control of the target hybrid computing resource network can perform business operations such as computing power scheduling, computing power control, and computing power allocation.

[0111] It should be noted that in this step, the nodes determined by the above sorting results are controlled to achieve global optimization scheduling control of computing resources for the entire target hybrid computing resource network.

[0112] The present invention provides a global optimization scheduling method for computing resources in a computing power network. Based on the computing power network, the method comprehensively controls the computing power resources of the hybrid computing power network by considering the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, thereby achieving global optimization scheduling of computing power resources.

[0113] The specific details of the global optimization scheduling method for computing resources in a computing power network provided by this invention will be described in detail in the following embodiments.

[0114] Please refer to Figure 3 , Figure 3 The flowchart illustrates the first specific method for optimizing and scheduling computing resources in a computing network, as provided in this embodiment of the invention.

[0115] See Figure 3 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0116] S201: Obtain the topology of the target hybrid computing resource network.

[0117] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0118] S202: Determine the time-sensitive microstate based on the topology of the subnet corresponding to the node and the connectivity of the node.

[0119] In this embodiment of the invention, the timeliness of information flow through the computing power network is used as a constraint indicator. The timeliness of information flow through the computing power network quantitatively characterizes the speed at which information flows through the computing power network, imposing constraints from the perspective of time efficiency. That is, when the constraint indicator includes the timeliness of information flow through the computing power network, it is first determined based on node a obtained in S201 above. i (i = 1, 2, ..., n), determine its corresponding subnet topology P i (i = 1, 2, ..., n), this subnet specifically contains s(s = s1, s2, ..., sn). x ) cuts, for any node That is, nodes It is a subnet P i A certain cut s (s = s1, s2, ..., s2) x In this step, the nodes in the topology will first be determined based on the above topology. Subnet topology P i Connectivity:

[0120]

[0121] and computing nodes Subnet topology P i The number of connections k (k = 1, 2, ..., N);

[0122] Then, in this step, we will calculate the number of nodes when the connection number is k (k = 1, 2, ... N). Subnet topology P i Connection degree corresponding to the s-th cut Time-sensitive microstate

[0123]

[0124] S203: Determine the total time-dependent connectivity of the subnet corresponding to the node based on the time-dependent micro-state.

[0125] In this step, the nodes will be calculated. The total timeliness of the subnet is denoted as

[0126]

[0127] S204: Based on the total timeliness connectivity, determine the timeliness microstate realization probability value of the subnet corresponding to the node.

[0128] This step will calculate the nodes. Subnet P i The time-dependent microstate realization probability is denoted as

[0129]

[0130] S205: Determine the time entropy of the subnet corresponding to the node based on the time-dependent microstate realization probability value.

[0131] This step will calculate the nodes. Subnet P i The time-dependent entropy is denoted as

[0132] S206: Determine the maximum time-dependent entropy of the subnet corresponding to the preferred node based on the maximum time-dependent connection degree in the total time-dependent connection degree and the time-dependent entropy.

[0133] The first step in this process is to determine the nodes. Subnet P i The maximum degree of connection is denoted as A. imax :

[0134]

[0135] Then, based on the maximum connectivity and the time-dependent entropy, the maximum time-dependent entropy will be determined and denoted as .

[0136]

[0137] S207: Determine the timeliness measure value of the subnet corresponding to the node based on the timeliness entropy and the maximum timeliness entropy.

[0138] compute nodes Subnet P i The timeliness measure value is denoted as

[0139]

[0140] S208: Based on the timeliness measure value, form the timeliness vector of the subnet corresponding to each node, which serves as the constraint index measure vector of the node.

[0141] Determine subnet P i The corresponding node is a i ∈P i (i = 1, 2, ..., n), calculate the subnet P corresponding to its optimal path. i The time-dependent vector of (i = 1, 2, ..., n) is denoted as R. T :

[0142]

[0143] In this step, the timeliness vector will be denoted as the constraint index measure vector.

[0144] S209: Determine the order of nodes based on the constraint index measure vector to obtain the sorting result.

[0145] In this step, when it is determined that only the timeliness of information flow in the computing power network is used as the constraint index, that is, when only the timeliness vector calculated in S208 is used as the constraint index measure vector, the values ​​will be sorted from largest to smallest according to their optimal values ​​in this step. The definition is as follows:

[0146]

[0147]

[0148] Following this pattern, by arranging the values ​​from largest to smallest, we obtain the time-efficient local optimum solution vector for a single objective, denoted as X. T* :

[0149] X T* ={a 11 a 12 , ..., a 1n}

[0150] In this step, the optimal time-efficiency solution vector for the single objective will be used as the sorting result.

[0151] S210: Perform computing power scheduling control on the target hybrid computing power computing resource network based on the sorting results.

[0152] This step is basically the same as S104 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0153] The specific details of the computing resource control and optimization method for a computing power network provided by this invention will be described in detail in the following embodiments.

[0154] Please refer to Figure 4 , Figure 4 The flowchart illustrates a second specific method for optimizing and scheduling computing resources in a computing network, as provided in this embodiment of the invention.

[0155] See Figure 4 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0156] S301: Obtain the topology of the target hybrid computing resource network.

[0157] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0158] S302: Determine the total number of quality microstates of a node based on the topology of the subnet corresponding to the node and the corresponding maximum span.

[0159] In this embodiment of the invention, the accuracy of computing node control quality is used as a constraint indicator. The accuracy of computing node control quality quantitatively characterizes the accuracy of information flow through the computing network, thus imposing constraints from the perspective of accuracy quality. That is, when the constraint indicator includes the accuracy of computing node control quality, it is first determined based on the node a obtained in S301 above. i (i = 1, 2, ..., n), determine its corresponding subnet topology P i (i = 1, 2, ..., n), this subnet specifically contains s(s = s1, s2, ..., sn). x ) cuts, for any node That is, nodes It is a subnet P i A certain cut s (s = s1, s2, ..., s2) x In this step, the nodes in the topology will first be determined based on the above topology. Cut s (s = s1, s2, ..., s2) x The span of ) is recorded

[0160] Next, the maximum span will be determined, i.e., the number of nodes will be calculated. span is Cut s (s = s1, s2, ..., s2) x Maximum span:

[0161]

[0162] Finally, the total number of mass microstates of the nodes will be determined, i.e., the number of compute nodes will be calculated. span is Total number of mass microstates:

[0163]

[0164] S303: Determine the realization probability value of the quality microstate of the subnet corresponding to the node based on the total number of quality microstates.

[0165] This step will calculate the nodes. The corresponding subnet P i The probability value of realization of the mass microstate is denoted as .

[0166]

[0167] S304: Determine the quality entropy and maximum quality entropy of the subnet corresponding to the node based on the probability value of the realization of the quality microstate.

[0168] This step will calculate the nodes. The corresponding subnet P i mass entropy

[0169]

[0170] After calculating the mass entropy mentioned above, the maximum mass entropy will be determined, i.e., the computation node. The corresponding subnet P i The maximum mass entropy, denoted as

[0171]

[0172] S305: Determine the quality metric value of the subnet corresponding to the node based on the quality entropy and the maximum quality entropy.

[0173] This step will calculate the nodes. The corresponding subnet P i The control quality value of the computing power node is denoted as

[0174]

[0175] S306: Based on the quality metric values, form the quality vector of the subnet corresponding to each node, which serves as the constraint index metric vector of the node.

[0176] In this step, based on the subnet topology P i Middle node a i ∈P i (i = 1, 2, ..., n), calculate the quality vector corresponding to the optimal path and denot it as R. M :

[0177]

[0178] In this step, the aforementioned quality vector can be used as the constraint index measurement vector.

[0179] S307: Determine the order of nodes based on the constraint index measure vector to obtain the sorting result.

[0180] In this step, when it is determined that only the accuracy of the quality control of the computing power node is used as the constraint index, that is, only the quality vector calculated in S306 is used as the constraint index measure vector, this step will sort the values ​​from largest to smallest according to the optimal values, and find the local optimal quality solution vector for the single objective, defined as follows:

[0181]

[0182]

[0183] Following this pattern, by arranging the values ​​from largest to smallest, we obtain the single-objective optimal quality solution vector, denoted as X. M* :

[0184] X M* ={a 11 a 12 , ..., a 1n}

[0185] In this step, the optimal quality solution vector for the single objective will be used as the sorting result.

[0186] S308: Perform computing power scheduling control on the target hybrid computing power computing resource network based on the sorting results.

[0187] This step is basically the same as S104 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0188] The specific details of the computing resource control and optimization method for a computing power network provided by this invention will be described in detail in the following embodiments.

[0189] Please refer to Figure 5 , Figure 5The flowchart illustrates a third specific method for optimizing and scheduling computing resources in a computing network, as provided in this embodiment of the invention.

[0190] See Figure 5 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0191] S401: Obtain the topology of the target hybrid computing resource network.

[0192] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0193] S402: Determine the computing power temporal entropy when the node is an electronic computing power node based on the topology of the subnet corresponding to the node.

[0194] In this step, the node a obtained in S301 above will be used first. i (i = 1, 2, ..., n), determine its corresponding subnet topology P i (i = 1, 2, ..., n), this subnet specifically contains s(s = s1, s2, ..., sn). x ) cuts, for any node That is, nodes It is a subnet P i A certain cut s (s = s1, s2, ..., s2) x ) nodes in ).

[0195] Before proceeding to this step, it is necessary to first determine whether the node is an electronic node or a quantum node; a quantum node is a computing power node composed of quantum computers, while an electronic node is an electronic computing power node composed of electronic computers.

[0196] For electronic computing nodes, this step will first record and store the nodes. Data object is Specify a maximum allowed time threshold Current time value Data objects timescale value Used to calculate the time-series entropy value of relevant data computing power.

[0197] The first step in this process is to perform normalization:

[0198]

[0199] Then calculate the nodes. The following measures of temporal uncertainty in the computing power data queue are used as the temporal entropy of the node's computing power:

[0200]

[0201] S403: Determine the timing optimization degree of the computing power data queue based on the computing power timing entropy.

[0202] This step will calculate the timing optimization degree of the computing power data queue. And store:

[0203]

[0204] S404: Use the timing optimization degree of the computing power data queue as a constraint indicator measure.

[0205] Before this step, it is necessary to determine whether the timing optimization degree of the computing power data queue should be used as a constraint indicator to control the nodes. The timing optimization degree of the computing power data queue corresponding to the electronic computing power node quantitatively characterizes the degree of temporal inconsistency between related data, imposing constraints from the perspective of computing power. If the determination result is yes, the timing optimization degree of the computing power data queue calculated in S403 will be used as the constraint indicator measure value, and then the node order will be determined according to the order of the constraint indicator measure values; otherwise, the subsequent process will continue.

[0206] S405: Determine the computing power overflow entropy when the node is an electronic computing power node based on the topology of the subnet corresponding to the node.

[0207] This step will first record and store the nodes. Data object is Specify a maximum computing power threshold Available computing power demand value Computing power available standard value Used to calculate the entropy of sequential queue computing power overflow.

[0208] The first step in this process is to perform normalization:

[0209]

[0210] Then calculate the nodes. The uncertainty measure of sequential queue computing power overflow is the computing power overflow entropy of a node:

[0211]

[0212] S406: Determine the optimization degree of sequential queue computing power overflow based on computing power overflow entropy.

[0213] This step will calculate the optimization degree of sequential queue computing power overflow. And store:

[0214]

[0215] S407: Use the optimization degree of sequential queue computing power overflow as a constraint index measure value.

[0216] Before this step, it is necessary to determine whether the sequential queue computing power overflow optimization degree should be used as a constraint indicator to control the nodes. The sequential queue computing power overflow optimization degree corresponding to the electronic computing power node quantitatively characterizes the degree of sequential queue computing power overflow, providing constraints from the perspective of computing power. If the result here is yes, the sequential queue computing power overflow optimization degree calculated in S406 will be used as the constraint indicator measure value, and then the node order will be determined according to the order of the constraint indicator measure values; otherwise, the subsequent process will continue.

[0217] It should be noted that S402 to S404 and S405 to S407 can be executed in parallel or sequentially, and their specific order is not specifically limited here.

[0218] S408: Determine the effective computing power adaptability measure of electronic nodes based on the computing power data queue timing optimization degree, sequential queue computing power overflow optimization degree, and preset weights.

[0219] Before this step, the weight coefficient values ​​need to be recorded and stored. So that the optimal solution vector can be calculated and stored later.

[0220] During the execution of this step, the effective computing power adaptability of computing resources is determined as a constraint indicator. That is, when the constraint indicator includes the effective computing power adaptability of computing resources, this step will calculate... The effective computing power adaptability measure of a node's computing resources is denoted as:

[0221]

[0222] in, The weights of the time-series optimization degree of the computing power data queue and the computing power overflow optimization degree of the sequential queue are respectively assigned.

[0223] S409: Determine the quantum volume of the quantum node in the node based on the topology of the subnet corresponding to the node.

[0224] For a quantum node, this step first records and stores the initial state |ψ> and the final state |φ> after the operation, denoted as ρ=|ψ><ψ|, σ=|φ><φ|; and records and stores the effective error rate (fidelity) F of the operation. n Quantum volume QV; recording and storing the effective computing power E of a quantum server a .

[0225] The first step in this process is to calculate the effective error rate of the quantum system:

[0226]

[0227] Then, the quantum volume QV measure of a quantum computing node for an arbitrary n-qubit system is calculated:

[0228]

[0229] Where m∈[2,n].

[0230] S410: Determine the effective computing power of the quantum server based on the quantum volume.

[0231] This step will calculate the effective computing power E of the quantum server. a And store:

[0232]

[0233] S411: Determine the effective computing power adaptability measure of quantum nodes based on the effective computing power capability of the quantum server.

[0234] In this step, the effective computing power of the quantum server will be used as a measure of the adaptability of the quantum node's computing resources to the effective computing power, so as to perform subsequent operations.

[0235] It should be noted that before this step, it is first necessary to determine whether the node is an electronic node or a quantum node. If it is a quantum node, the effective computing power of the quantum server needs to be used as a constraint indicator to control the node. The effective computing power of the quantum server corresponding to the quantum computing node is a quantitative characterization of the effective computing power of the quantum computing node, imposing constraints from the perspective of computing power. If the determination result here is yes, the effective computing power of the quantum server calculated in S410 will be used as the constraint indicator measure value, and then the order of the nodes will be determined according to the order of the size of the constraint indicator measure value; otherwise, the subsequent process will continue.

[0236] S412: The computing resource effective computing power adaptability vector formed based on the computing resource effective computing power adaptability measure value is used as the constraint index measure vector of the node.

[0237] In this step, it will be based on subnet P i Middle node a i ∈P i (i = 1, 2, ..., n), the solution vector of the measure of the consistency of computing power adaptation of computing resources, is used as the effective computing power adaptation vector of computing resources, denoted as X. (3) :

[0238]

[0239] In this step, the aforementioned effective computing power adaptability vector of computing resources can be used as a constraint index measurement vector. Effective computing power adaptability refers to the degree to which the actual computing power obtained from the computing power network itself effectively represents the business capabilities of computing power nodes, thus imposing constraints from the perspective of adapting computing power.

[0240] S413: Determine the order of nodes based on the constraint index measure vector to obtain the sorting result.

[0241] In this step, when the effective computing power adaptability of computing resources is determined as the constraint index, the computing node a is computed in this step. i The optimal solution for consistent adaptation of computing resources and computing power for (i = 1, 2, ..., n):

[0242] definition:

[0243]

[0244] Following this pattern, by sorting the optimal values ​​from largest to smallest, we obtain the single-objective local optimal resource computing power adaptation consistency solution vector, denoted as X. (3)* :

[0245] X (3)* =(a 31 a 32 , ..., a 3n )

[0246] In this step, the consistency solution vector of the local optimal resource computing power adaptation for a single objective will be used as the sorting result.

[0247] S414: Perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0248] The above-described S414 is similar to the corresponding content in the above-described embodiments of the invention, and will not be described again here.

[0249] The specific details of the computing resource control and optimization method for a computing power network provided by this invention will be described in detail in the following embodiments.

[0250] Please refer to Figure 6 , Figure 6 The flowchart illustrates the fourth specific method for optimizing and scheduling computing resources in a computing network, as provided in this embodiment of the invention.

[0251] See Figure 6 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0252] S501: Obtain the topology of the target hybrid computing resource network.

[0253] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0254] S502: Based on the baseline throughput value measured on each scenario load, the actual training throughput of the business scenario in the node, and the weight of the business scenario load, determine the relative entropy of the computing power carrying the business scenario.

[0255] In this embodiment of the invention, the computing power carrying service scenario performance is used as a constraint indicator. The computing power carrying service scenario performance quantitatively characterizes the effective computing power obtained by the tested system and the test system under a given service scenario (including training scenario) of the computing power network, and imposes constraints from the perspective of obtaining effective computing power from the service scenario. That is, when the constraint indicator includes the computing power carrying service scenario performance, the first step is to obtain the computing power at the node according to the above S501. For a given set of business scenarios S i The corresponding scenario load is s(∈S) i Using a specific reference computing system, the throughput was measured on s. As a baseline, the actual training throughput of the tested system was Th. is The weight of each scenario's job load s is τ. s

[0256] The definition of relative entropy above is:

[0257] Let Ω = {0, 1, ...,}, x i y i These are two probability measures on Ω;

[0258] Let x i ≥0, y i ≥0, i=1, 2,...,n; and Then it is said:

[0259] Let X be the relative entropy of X with respect to Y, where X = {x1, x2, ..., x...} n} T Y = {y1, y2, ..., y} n} T .

[0260] In this step, normalization will be performed first:

[0261]

[0262]

[0263] Then, based on the normalized data, the relative entropy of the computing power supporting the business scenario is calculated.

[0264]

[0265] S503: Determine the performance measurement value of the computing power carrying service scenario based on the relative entropy of the computing power carrying service scenario.

[0266] This step will calculate the performance metric value for computing power supporting business scenarios.

[0267]

[0268] S504: Based on the performance measurement value of computing power carrying service scenarios, form the computing power carrying service scenario performance vector of the subnet corresponding to the node, which serves as the constraint indicator measurement vector of the node.

[0269] In this step, it will be based on subnet P i Middle node a i ∈P i (i = 1, 2, ..., n), calculate the measure vector of computing power carrying capacity service scenario effectiveness, and use it as the computing power carrying capacity service scenario effectiveness vector, denoted as X. (4) :

[0270]

[0271] In this step, the performance vector of the computing power supporting the business scenario will be denoted as the constraint index measurement vector.

[0272] S505: Determine the order of nodes based on the constraint index measure vector to obtain the sorting result.

[0273] In this step, when it is determined that only the performance of computing power in the business scenario is used as a constraint indicator, node a will be calculated. i The optimal solution for calculating the single-objective computing power support for a business scenario for (i = 1, 2, ..., n) is defined as follows:

[0274]

[0275]

[0276] Following this pattern, by arranging the values ​​from largest to smallest, we obtain the solution vector for the single-objective local optimal computing power support service scenario, denoted as X. (4)* :

[0277] X (4)* =(a 41 a 42 , ..., a 4n )

[0278] In this step, the solution vector of the single-objective local optimal computing power to support the service scenario will be used as the sorting result.

[0279] S506: Perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0280] This step is basically the same as S104 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0281] The specific details of the global optimization scheduling method for computing resources in a computing power network provided by this invention will be described in detail in the following embodiments.

[0282] Please refer to Figure 7 , Figure 7 The flowchart illustrates the fifth specific method for optimizing and scheduling computing network resources provided in this embodiment of the invention.

[0283] See Figure 7 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0284] S601: Obtain the topology of the target hybrid computing resource network.

[0285] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0286] S602: Based on the actual performance standard value of the given test set on the software system, the actual performance of the test set in the node of the tested software, and the weight of each tested software, determine the relative entropy for software operation tuning.

[0287] In this embodiment of the invention, software runtime optimization performance is used as a constraint indicator. Software runtime optimization performance quantitatively characterizes the optimization process performed at the software layer (operating system, computing engine, and framework) during the actual operation of the target software in the computing power network, thus imposing constraints from the perspective of software optimization performance. That is, when the constraint indicator includes software runtime optimization performance, the performance at the node is first obtained according to the above-described S601. Actual performance of a given test set on a standard software system The corresponding target software is s(∈S) i The actual performance of the test set on the target software system is P. is The weight of each tested software s is σ. s .

[0288] In this step, normalization will be performed first:

[0289]

[0290] Then, based on the normalized data, the relative entropy of the software runtime optimization is calculated, i.e., the relative entropy of the software runtime optimization:

[0291]

[0292] S603: Determine the performance measurement value of software operation process optimization based on the relative entropy of software operation process optimization.

[0293] In this step, the software runtime optimization performance measurement value will be calculated according to the following formula.

[0294]

[0295] S604: Based on the software operation process optimization performance measurement value, form the software operation process optimization performance vector of the subnet corresponding to the node, and use it as the constraint index measurement vector of the node.

[0296] In this step, it will be based on subnet P i Middle node a i ∈P i (i = 1, 2, ..., n), the measure solution vector of the software runtime optimization performance is used as the software runtime optimization performance vector, denoted as X. (5) :

[0297]

[0298] In this step, the optimization performance vector of the software operation process will be denoted as the constraint index measure vector.

[0299] S605: Determine the order of nodes based on the constraint index measure vector to obtain the sorting result.

[0300] In this step, when it is determined that only the performance of software runtime optimization is used as the constraint indicator, node a will be calculated. i The optimal solution for optimizing the performance of a single-objective computation software for (i = 1, 2, ..., n) during runtime is defined as follows:

[0301]

[0302]

[0303] Following this pattern, by arranging the values ​​from largest to smallest, we obtain the solution vector for optimizing the local optimum software operation process for a single objective, denoted as X. (5)* :

[0304] X (5)* =(a 51 a 52 , ..., a5n )

[0305] In this step, the solution vector of the local optimal software operation process optimization efficiency will be used as the sorting result.

[0306] S606: Perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0307] This step is basically the same as S104 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0308] The specific details of the global optimization scheduling algorithm for computing resources in a computing network provided by this invention will be described in detail in the following embodiments.

[0309] Please refer to Figure 8 , Figure 8 The flowchart illustrates the sixth specific method for optimizing and scheduling computing network resources provided in this embodiment of the invention.

[0310] See Figure 8 In this embodiment of the invention, the method for optimizing and scheduling computing resources in a computing power network includes:

[0311] S701: Obtain the topology of the target hybrid computing resource network.

[0312] This step is basically the same as S101 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0313] S702: Determine the constraint index measure value based on the topology and the algorithm of the preset constraint index.

[0314] The specific details of this step have been described in detail in the above embodiments of the invention, and will not be repeated here. It should be noted that, in the embodiments of the present invention, global optimization will be performed. At least two of the primary indicators will be selected from the various vectors described above, and typically five constraint indicator measurement vectors will be selected as the evaluation matrix of the constraint indicators. That is, the embodiments of the present invention will be executed when the constraint indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation.

[0315] S703: Form a comprehensive evaluation matrix based on the constraint index measurement vector.

[0316] This step will call the preferred node a in the computing power network of the storage module. i The evaluation values ​​x of multiple constraint objectives corresponding to (i = 1, 2, ..., n) ij(i = 1, 2, ..., n, j = 1, 2, ..., 5), forming a matrix X = (x ij ) n×5 Generally, based on the constrained objective o j (j = 1, 2, ..., m), based on node a i (i = 1, 2, ..., n) Policy measure x ij The mathematical method determined by (i = 1, 2, ..., n, j = 1, 2, ..., m) calls the comprehensive evaluation matrix of computing resources in the data layer computing power network: X = (x ij ) n×m :

[0317] Next, in this step, x will be... ij Regarding the normalization of i, the normalized matrix B = (b ij ) n×m ,in, This serves as the evaluation matrix for subsequent steps.

[0318] S704: Based on the comprehensive evaluation matrix and the weights corresponding to each constraint index, determine the preference vector as the global optimal solution vector.

[0319] Before proceeding to this step, it is necessary to first calculate the constraint target o. j The weights w of (j = 1, 2, ..., m) j (j = 1, 2, ..., m), the specific weight can be obtained by the structural entropy weight method. For details on the structural entropy weight method, please refer to the existing technology, which will not be elaborated here.

[0320] In this step, the specific calculation of the computing power network node {a} will be performed. i Strategy set A (i = 1, 2, ..., n) {i，j} The global optimal solution is obtained by first calculating the preference vector. in:

[0321]

[0322] S705: Use the global optimal solution vector as the final computing power scheduling control constraint index measurement vector.

[0323] This step will record and store node {a i The global optimal solution value of computing network computing resources in strategy set A (i = 1, 2, ..., n) The solution vector is formed and used as the final constraint index measure vector.

[0324] S706: Determine the order of nodes based on the computing power scheduling control constraint index metric vector to obtain the sorting result.

[0325] In this step, it will be based on The sizes are reordered to obtain a new sorted vector. Optimal solution set The corresponding sorting is the node strategy set corresponding to the globally optimal configuration of computing resources in the computing power network. That is, the optimal solution vector corresponding to the globally optimal configuration node. The corresponding node set is the globally optimal configuration node set for computing resources in the computing power network.

[0326] In this step, the globally optimal configuration set of nodes for computing resources in the computing power network will be used as the sorting result.

[0327] S707: Perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0328] This step will optimize the global configuration of the computing resources of the computing network by setting a set of nodes. Pick It is the node corresponding to the global optimal solution of computing resources in the computing power network, and performs business operations such as computing power scheduling, computing power control and computing power allocation on the target hybrid computing power computing resource network.

[0329] The rest of this step is basically the same as S104 in the above-described embodiment of the invention. For details, please refer to the above-described embodiment of the invention. It will not be repeated here.

[0330] The present invention provides a global optimization scheduling method for computing resources in a computing power network. Based on a judgment matrix constructed from multiple constraint index measure vectors, a globally optimal solution can be obtained. Based on the globally optimal solution, business operations such as computing power scheduling, computing power control, and computing power allocation are performed on the target hybrid computing power network, thereby achieving globally optimal scheduling and allocation of computing power resources.

[0331] The following describes a computing power network computing resource optimization scheduling follow-up control device provided by an embodiment of the present invention. The computing power network computing resource optimization scheduling follow-up control device described below can be referred to in correspondence with the computing power network computing resource global optimization scheduling method described above.

[0332] In this embodiment of the invention, the follow-up control device includes a hardware-based software control module, a memory for carrying the software control module, a processor, and a computer-readable storage medium storing computer programs for implementing the above embodiments of the invention.

[0333] Please refer to Figure 9 , Figure 9 This is a structural block diagram of a computing power network computing resource optimization and scheduling device provided in an embodiment of the present invention.

[0334] See Figure 9 In this embodiment of the invention, the computing power network computing resource optimization and scheduling device may include:

[0335] The node acquisition module 100 is used to acquire the topology of the target hybrid computing power resource network;

[0336] The constraint index module 200 is used to determine the constraint index measure value based on the topology and the algorithm of the preset constraint index.

[0337] The computing power ranking module 300 is used to determine the order of the nodes according to the order of the size of the constraint index measurement values, and obtain the ranking result;

[0338] The scheduling control module 400 is used to perform global optimization scheduling control of computing resources for the target hybrid computing resource network based on the sorting results.

[0339] Preferably, in this embodiment of the invention, when the constraint index includes the timeliness of information flow in the computing power network, the constraint index module 200 includes:

[0340] The time-dependent microstate unit is used to determine the time-dependent microstate based on the topology of the subnet corresponding to the node and the connectivity of the node.

[0341] The total timeliness connectivity unit is used to determine the total timeliness connectivity of the subnet corresponding to the node based on the timeliness microstate.

[0342] The time-dependent microstate realization probability value unit is used to determine the time-dependent microstate realization probability value of the subnet corresponding to the node based on the total time-dependent connectivity.

[0343] The time-entropy unit is used to determine the time-entropy of the subnet corresponding to the node based on the time-entropy microstate realization probability value;

[0344] The maximum time-entropy unit is used to determine the maximum time-entropy of the subnet corresponding to the preferred node based on the maximum connection degree in the total time-entropy degree and the time-entropy.

[0345] The timeliness measurement value unit is used to determine the timeliness measurement value of the subnet corresponding to the node based on the timeliness entropy and the maximum timeliness entropy;

[0346] The timeliness vector unit is used to form the timeliness vector of each subnet corresponding to the node based on the timeliness metric value, which serves as the constraint index metric vector of the node.

[0347] Also includes:

[0348] The node prediction module is used to determine whether the computing power node is an electronic computing power node or a quantum computing power node based on the type of computing power node in the hybrid computing power network.

[0349] The computing power sorting module 300 is used for:

[0350] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0351] Preferably, in this embodiment of the invention, when the constraint index includes the accuracy of the computing power node control quality, the constraint index module 200 includes:

[0352] The total number of quality microstates unit is used to determine the total number of quality microstates of a node based on the topology of the subnet corresponding to the node and the corresponding maximum span.

[0353] The quality microstate realization probability value unit is used to determine the quality microstate realization probability value of the subnet corresponding to the node based on the total number of quality microstates.

[0354] The mass entropy and maximum mass entropy unit is used to determine the mass entropy and maximum mass entropy of the subnet corresponding to the node based on the mass microstate realization probability value.

[0355] A quality value unit is used to determine the quality value of the subnet corresponding to the node based on the quality entropy and the maximum quality entropy.

[0356] A quality vector unit is used to form a quality vector based on the quality value of the subnet corresponding to the node, and to serve as the constraint index measure vector of the node.

[0357] The computing power sorting module 300 is used for:

[0358] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0359] Preferably, in this embodiment of the invention, when the constraint index includes the adaptability of effective computing power to computing resources, the constraint index module 200 includes:

[0360] The first computing power temporal entropy unit is used to determine the computing power temporal entropy of the electronic nodes in the node according to the topology of the subnet corresponding to the node.

[0361] The first computing power overflow entropy unit is used to determine the computing power overflow entropy of the electronic nodes in the node according to the topology of the subnet corresponding to the node.

[0362] A quantum volume unit is used to determine the quantum volume of a quantum node in the node based on the topology of the subnet corresponding to the node.

[0363] The first computing power timing optimization unit is used to determine the computing power data queue timing optimization degree based on the computing power timing entropy.

[0364] The first computing power overflow optimization unit is used to determine the sequential queue computing power overflow optimization degree based on the computing power overflow entropy.

[0365] The first quantum server effective computing power unit is used to determine the effective computing power of the quantum server based on the quantum volume.

[0366] The electronic node computing resource effective computing power adaptability measurement value unit is used to determine the electronic node's computing resource effective computing power adaptability measurement value based on the computing power data queue timing optimization degree, the sequential queue computing power overflow optimization degree, and a preset weight.

[0367] The quantum node computing resource effective computing power adaptability measurement value unit is used to determine the quantum node's computing resource effective computing power adaptability measurement value based on the effective computing power capability of the quantum server;

[0368] The computing resource effective computing power adaptability vector unit is used to form a computing resource effective computing power adaptability vector based on the computing resource effective computing power adaptability measure value, and to use it as the constraint index measure vector of the node.

[0369] The computing power sorting module 300 is used for:

[0370] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0371] Preferably, in this embodiment of the invention, when the constraint index includes the computing power data queue timing optimization degree, the constraint index module 200 includes:

[0372] The second computing power temporal entropy unit is used to determine the computing power temporal entropy of the node based on the topology of the subnet corresponding to the node.

[0373] The second computing power timing optimization unit is used to determine the computing power data queue timing optimization degree based on the computing power timing entropy.

[0374] The third computing power timing optimization unit is used to use the computing power data queue timing optimization degree as the constraint index measure value of the node.

[0375] Preferably, in this embodiment of the invention, when the constraint index includes the sequential queue computing power overflow optimization degree, the constraint index module 200 includes:

[0376] The second computing power overflow entropy unit is used to determine the computing power overflow entropy of the node based on the topology of the subnet corresponding to the node.

[0377] The second computing power overflow optimization unit is used to determine the sequential queue computing power overflow optimization degree based on the computing power overflow entropy.

[0378] The third computing power overflow optimization unit is used to use the sequential queue computing power overflow optimization degree as the constraint index measure value of the node.

[0379] Preferably, in this embodiment of the invention, when the constraint index includes the effective computing power capability of the quantum server, the constraint index module 200 includes:

[0380] The first quantum system effective error rate unit is used to determine the quantum system effective error rate of the quantum node based on the quantum computing power node corresponding to the quantum node;

[0381] The second quantum volume unit is used to determine the quantum volume based on the effective error rate of the quantum system;

[0382] The third quantum server effective computing power unit is used to determine the effective computing power of the quantum server based on the quantum volume.

[0383] The fourth unit, the quantum server effective computing power unit, is used to use the quantum server effective computing power of the quantum node as the constraint index measure value.

[0384] Preferably, in this embodiment of the invention, when the constraint index includes the performance of computing power carrying service scenarios, the constraint index module 200 includes:

[0385] The relative entropy unit for computing power carrying business scenarios is used to determine the relative entropy of computing power carrying business scenarios based on the baseline value of throughput measured by the load of each scenario, the actual training throughput of the business scenarios in the node, and the weight of the load of the business scenarios.

[0386] The computing power carrying service scenario performance measurement value unit is used to determine the computing power carrying service scenario performance measurement value based on the relative entropy of the computing power carrying service scenario.

[0387] The computing power carrying service scenario performance vector unit is used to form the computing power carrying service scenario performance vector of the subnet corresponding to the node based on the computing power carrying service scenario performance measurement value, and to serve as the constraint index measurement vector of the node.

[0388] The computing power sorting module 300 is used for:

[0389] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0390] Preferably, in this embodiment of the invention, when the constraint index includes software runtime optimization performance, the constraint index module 200 includes:

[0391] The software runtime optimization relative entropy unit is used to determine the software runtime optimization relative entropy based on the actual performance standard value of the given test set on the software system, the actual performance of the software under test in the node, and the weight of each software under test.

[0392] The software operation process optimization performance measurement value unit is used to determine the software operation process optimization performance measurement value based on the relative entropy of the software operation process optimization.

[0393] The software operation process optimization performance vector unit is used to form the software operation process optimization performance vector of the subnet corresponding to the node based on the software operation process optimization performance measurement value, and to serve as the constraint index measurement vector of the node.

[0394] The computing power sorting module 300 is used for:

[0395] The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

[0396] Preferably, in this embodiment of the invention, when the constraint indicators include at least two of the following: the timeliness of information flow in the computing power network, the accuracy of the quality control of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, the constraint indicator module 200 includes:

[0397] The evaluation matrix unit is used to form a comprehensive evaluation matrix based on the constraint index measure vector;

[0398] The preference vector unit is used to determine the preference vector as the global optimal solution vector based on the comprehensive evaluation matrix and the weights corresponding to each constraint index.

[0399] The final constraint index unit is used to take the global optimal solution vector as the final computing power global optimization scheduling control constraint index measurement vector.

[0400] The computing power sorting module 300 is used for:

[0401] The order of the nodes is determined based on the computing power scheduling control constraint index metric vector, and the sorting result is obtained.

[0402] The computing power network computing resource optimization and scheduling device of this embodiment is used to implement the aforementioned computing power network computing resource global optimization and scheduling method. Therefore, the specific implementation of the computing power network computing resource optimization and scheduling device can be found in the embodiment section of the computing power network computing resource global optimization and scheduling method above. For example, the node acquisition module 100, constraint index module 200, computing power ranking module 300, and scheduling control module 400 are respectively used to implement steps S101 to S104 in the aforementioned computing power network computing resource global optimization and scheduling method. Therefore, its specific implementation can be referred to the description of the corresponding embodiments, and will not be repeated here.

[0403] The following describes a computing power network computing resource optimization and scheduling device provided by an embodiment of the present invention. The computing power network computing resource optimization and scheduling device described below can be referred to in correspondence with the computing power network computing resource global optimization and scheduling method and computing power network computing resource optimization and scheduling device described above.

[0404] Please refer to Figure 10 , Figure 10 This is a structural block diagram of a computing power network computing resource global optimization scheduling device provided in an embodiment of the present invention.

[0405] Reference Figure 10 The computing resource global optimization scheduling and control device of the computing power network may include a processor 11 and a memory 12.

[0406] The memory 12 is used to store computer programs; the processor 11 is used to execute the computer programs to implement the global optimization scheduling method for computing resources of the computing power network described in the above embodiments of the invention.

[0407] In this embodiment, the processor 11 in the computing resource global optimization scheduling control device of the computing power network is used to install the computing resource global optimization scheduling follow-up control device of the hybrid computing power network described in the above-mentioned embodiments. Simultaneously, the processor 11, combined with the memory 12, can implement the computing resource global optimization scheduling method of the computing power network described in any of the above-mentioned embodiments. Therefore, the specific implementation of the computing resource global optimization scheduling control device of the computing power network can be found in the embodiments section of the computing resource global optimization scheduling method of the computing power network described above. The specific implementation can be referred to the descriptions of the corresponding embodiments, and will not be repeated here.

[0408] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a global optimization scheduling and control method for computing resources in a computing network as described in any of the above embodiments. Further details can be found in the prior art and will not be elaborated upon here.

[0409] The present invention also provides a global optimization scheduling and control system for computing resources in a hybrid computing network, which typically includes a user layer, an application server layer, and a data layer.

[0410] User layer S1: This layer simultaneously satisfies the data selection for computing resource allocation and scheduling in the computing power network, as well as the data interaction for the optimal computing power node strategy. In other words, within the computing power network topology, this module layer communicates with various functional modules and allocates scheduling data regarding the selection process for candidate optimal computing power nodes.

[0411] Application Server Layer S2: This module serves as an intermediary layer between the data layer and the user layer. It enables unified database management, knowledge base management, user access to the database, and ensures database security and operational efficiency. The user interface connects to the data layer through this intermediary layer.

[0412] The application server layer provides database management functions, including user and permission management, assigning different permissions to different users.

[0413] The knowledge base management function at the application server layer manages and maintains databases and statistical databases containing characteristic parameters of quantum and electronic computers, including adding, deleting, editing, and querying information.

[0414] Data Layer S3: Used to store or retrieve data, including optimized data that meets the computing power optimization node strategy, and stored or retrieved data of the computing power nodes to be optimized in various functional modules, some data that communicates with the database, pre-set data, intermediate data that needs to be stored, and data allocated and scheduled. This module acts as a data storage repository.

[0415] The function of storing or retrieving data includes, for example, the computing power optimization node a. i (i = 1, 2, ..., n) correspond to computing nodes in the subnet. The data is based on the constraint target. j (j = 1, 2, ..., m) obtains the relevant initial state data, intermediate operation data, analysis process data, sub-item operation data, comprehensive operation data, and allocation and scheduling data corresponding to the judgment measurement data; stores or calls the network system computing power optimization node. The number of n, s x Optimal computing power node a i (i = 1, 2, ..., n) Strategy set A {i，j} The global optimal solution has n components {x} i1 x i2 , ..., x in}(i=1,2,...,n), constraint target o j(j = 1, 2, ..., m) dimension m (corresponding to m constraints) and data for allocation and scheduling, etc.

[0416] The computing resource global optimization scheduling and control system based on the aforementioned computing power network can specifically implement the computing resource global optimization scheduling and control method for a computing power network provided in any of the above embodiments. Its specific details have been described in detail in the above embodiments and will not be repeated here.

[0417] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0418] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0419] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0420] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0421] The above provides a detailed description of the computing resource control optimization method and servo control device for computing power networks provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this invention.

Claims

1. A method for optimizing the scheduling of computing resources in a computing power network, characterized in that, include: Obtain the topology of the target hybrid computing resource network; Based on the topology and the algorithm for the preset constraint index, the constraint index measure value is determined. The order of nodes is determined based on the order of the constraint index measure values, and the sorting result is obtained. Based on the sorting results, the target hybrid computing resource network is subjected to global optimization scheduling control of computing resources. The constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators: The primary indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation. The secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes. When the constraint metric includes the adaptability of effective computing power to computing resources, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes: Based on the topology of the subnet corresponding to the node, determine the computing power temporal entropy of the electronic node in the node; Based on the topology of the subnet corresponding to the node, determine the computing power overflow entropy of the electronic node in the node; Based on the topology of the subnet corresponding to the node, determine the quantum volume of the quantum node in the node; The computing power data queue timing optimization degree is determined based on the computing power time sequence entropy; The optimization degree of sequential queue computing power overflow is determined based on the computing power overflow entropy. The effective computing power of the quantum server is determined based on the quantum volume. The effective computing power adaptability measure value of the electronic node is determined based on the timing optimization degree of the computing power data queue, the computing power overflow optimization degree of the sequential queue, and the preset weights. Based on the effective computing power capability of the quantum server, determine the effective computing power adaptability measure value of the quantum node's computing resources; The computing resource effective computing power adaptability vector formed based on the computing resource effective computing power adaptability measure value is used as the constraint index measure vector of the node. The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

2. The method according to claim 1, characterized in that, When the constraint index includes the timeliness of information flow in the computing power network, determining the constraint index measure value based on the topology and the algorithm for the preset constraint index includes: Based on the topology of the subnet corresponding to the node and the connectivity of the node, the time-sensitive microstate is determined; Based on the aforementioned time-dependent microstate, determine the total time-dependent connectivity of the subnet corresponding to the node; Based on the total timeliness correlation, determine the timeliness microstate realization probability value of the subnet corresponding to the node; Based on the time-dependent microstate realization probability value, determine the time-dependent entropy of the subnet corresponding to the node; Based on the maximum connectivity in the total time-sensitive connectivity and the time-sensitive entropy, determine the maximum time-sensitive entropy of the subnet corresponding to the node; Based on the time-efficiency entropy and the maximum time-efficiency entropy, determine the time-efficiency metric value of the subnet corresponding to the node; Based on the timeliness metric value, a timeliness vector is formed for each subnet corresponding to the node, which serves as the constraint index metric vector for the node; The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

3. The method according to claim 1, characterized in that, When the constraint metric includes the accuracy of the control quality of computing nodes, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes: Based on the topology of the subnet corresponding to the node and the corresponding maximum span, determine the total number of quality microstates of the node; Based on the total number of quality microstates, determine the probability value of the realization of the quality microstate of the subnet corresponding to the node; Based on the probability value of realization of the quality microstate, determine the quality entropy and maximum quality entropy of the subnet corresponding to the node; Based on the quality entropy and the maximum quality entropy, determine the quality metric value of the subnet corresponding to the node; The quality vector of each subnet corresponding to each node is formed based on the quality metric value, and serves as the constraint index metric vector of the node. The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

4. The method according to claim 1, characterized in that, When the constraint metric includes the timing optimization degree of the computing power data queue, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes: Based on the topology of the subnet corresponding to the node, determine the computing power temporal entropy of the node; The computing power data queue timing optimization degree is determined based on the computing power time sequence entropy; The time-series optimization degree of the computing power data queue is used as the constraint index measure value.

5. The method according to claim 1, characterized in that, When the constraint metric includes the optimization degree of sequential queue computing power overflow, determining the constraint metric measure value based on the topology and the algorithm of the preset constraint metric includes: Based on the topology of the subnet corresponding to the node, determine the computing power overflow entropy of the node; The optimization degree of sequential queue computing power overflow is determined based on the computing power overflow entropy. The optimization degree of the sequential queue computing power overflow is used as the measurement value of the node constraint index.

6. The method according to claim 1, characterized in that, When the constraint metric includes the effective computing power of the quantum server, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes: The quantum volume of the quantum computing node is determined based on the quantum computing node. The effective computing power of the quantum server is determined based on the quantum volume. The effective computing power of the quantum server is used as the metric value of the node constraint index.

7. The method according to claim 1, characterized in that, When the constraint metric includes the performance of computing power support service scenarios, determining the constraint metric measure value based on the topology and the algorithm for the preset constraint metric includes: Based on the baseline throughput value measured for each scenario load, the actual training throughput of the business scenario in the node, and the weight of the business scenario load, the relative entropy of the computing power carrying the business scenario is determined. Based on the relative entropy of the computing power carrying service scenario, determine the performance measurement value of the computing power carrying service scenario; Based on the computing power carrying service scenario performance measurement value, a computing power carrying service scenario performance vector of the subnet corresponding to the node is formed, which serves as the constraint index measurement vector of the node. The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

8. The method according to claim 1, characterized in that, When the constraint index includes software runtime optimization performance, determining the constraint index measure value based on the topology and the algorithm for the preset constraint index includes: Based on the actual performance standard values ​​of the given test set on the software system, the actual performance values ​​of the software under test in the node, and the weight of each software under test, determine the relative entropy for software operation tuning. Based on the relative entropy of the software operation process, determine the optimization performance measure value of the software operation process; Based on the software operation process optimization performance measurement value, a software operation process optimization performance vector for the subnet corresponding to the node is formed, which serves as the constraint index measurement vector for the node. The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

9. The method according to any one of claims 1 to 8, characterized in that, When the constraint indicators include at least two of the following: the timeliness of information flow in the computing power network, the accuracy of the quality control of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power in carrying business scenarios, and the optimization efficiency of software operation, the step of determining the constraint indicator measure value based on the topology and the algorithm of the preset constraint indicators includes: A comprehensive evaluation matrix is ​​formed based on the constraint index measure vector; Based on the comprehensive evaluation matrix and the weights corresponding to each constraint index, the preference vector is determined as the global optimal solution vector; The global optimal solution vector is used as the final computing power scheduling control constraint index measurement vector; The process of determining the order of the nodes based on the order of the constraint index measure values, and obtaining the sorting result, includes: The order of the nodes is determined based on the computing power scheduling control constraint index metric vector, and the sorting result is obtained.

10. A computing power network computing resource optimization and scheduling device, characterized in that, include: The node acquisition module is used to acquire the topology of the target hybrid computing resource network; The constraint index module is used to determine the constraint index measure value based on the topology and the algorithm of the preset constraint index. The computing power ranking module is used to determine the order of nodes based on the order of the constraint index measurement values, and obtain the ranking result. The scheduling and control module is used to perform global optimization scheduling and control of computing resources for the target hybrid computing resource network based on the sorting results. The constraint indicators include any of the following primary and secondary indicators, or any combination of primary indicators: The primary indicators include the timeliness of information flow in the computing power network, the accuracy of the control quality of computing power nodes, the effective computing power adaptability of computing resources, the efficiency of computing power to support business scenarios, and the optimization efficiency of software operation. The secondary indicators include the optimization degree of sequential queue computing power overflow corresponding to electronic computing power nodes, the optimization degree of computing power data queue timing, and the effective computing power capability of quantum servers corresponding to quantum computing power nodes. When the constraint index includes the effective computing power adaptability of computing resources, the constraint index module includes: The first computing power temporal entropy unit is used to determine the computing power temporal entropy of the electronic nodes in the node according to the topology of the subnet corresponding to the node. The first computing power overflow entropy unit is used to determine the computing power overflow entropy of the electronic nodes in the node according to the topology of the subnet corresponding to the node. A quantum volume unit is used to determine the quantum volume of a quantum node in the node based on the topology of the subnet corresponding to the node. The first computing power timing optimization unit is used to determine the computing power data queue timing optimization degree based on the computing power timing entropy. The first computing power overflow optimization unit is used to determine the sequential queue computing power overflow optimization degree based on the computing power overflow entropy. The first quantum server effective computing power unit is used to determine the effective computing power of the quantum server based on the quantum volume. The electronic node computing resource effective computing power adaptability measurement value unit is used to determine the electronic node's computing resource effective computing power adaptability measurement value based on the computing power data queue timing optimization degree, the sequential queue computing power overflow optimization degree, and a preset weight. The quantum node computing resource effective computing power adaptability measurement value unit is used to determine the quantum node's computing resource effective computing power adaptability measurement value based on the effective computing power capability of the quantum server; The computing resource effective computing power adaptability vector unit is used to form a computing resource effective computing power adaptability vector based on the computing resource effective computing power adaptability measure value, and to use it as the constraint index measure vector of the node. The computing power ranking module is used for: The order of the nodes is determined based on the constraint index measure vector, and the sorting result is obtained.

11. A hybrid computing power network computing resource optimization and scheduling device, characterized in that, The device includes: Memory: Used to store computer programs; Processor: Used to implement the steps of the computing power network computing resource optimization scheduling method as described in any one of claims 1 to 9 when executing the computer program.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the computing power network computing resource optimization scheduling method as described in any one of claims 1 to 9.

Images