Intelligent computing power distribution method based on big data processing

Abstract

The invention discloses an intelligent computing power distribution method based on big data processing, and particularly relates to the field of computing power distribution, and the method comprises the steps: S1, determining a computing power distribution region, which is used for determining data in a target computing server as the computing power distribution region, and dividing the computing power distribution region into sub-regions according to a 5s / time equal-time division mode, sequentially numbering the sub-regions in the computing power distribution region as 1, 2,..., n; s2, data acquisition: acquiring computing power resource data and task data in a target computing power distribution process, transmitting the computing power resource data to a computing power resource data processing step, and transmitting the task data to a task data processing step; according to the method, decision making is carried out by adopting a global perspective, a distribution mode is considered in multiple aspects of own computing power resources and task requirements, and a matching relationship between tasks and resources is deeply and comprehensively considered, so that the condition of local optimization and overall non-optimization is effectively avoided, and the overall performance of the system is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of computing power allocation, and more specifically, to a method for intelligently allocating computing power based on big data processing. Background Art

[0002] With the rapid development of big data technology, the scale of data has exploded, and the demand for computing resources is increasing. Computing systems play an important role in improving computing efficiency, supporting complex computing tasks, and enhancing computer performance. Therefore, an intelligent computing power allocation method is needed to optimize resource utilization.

[0003] Traditional computing power allocation methods include computing power allocation methods, resource inventory, allocation strategy formulation, and allocation execution. The computing power allocation method clearly defines the computing power resources required for computing tasks, understands the resource requirements of the tasks, and provides a basis for subsequent allocation. Resource inventory comprehensively counts the existing computing power resources in the system and clearly defines the total amount of resources available for allocation, so that they can be reasonably allocated according to task requirements. Allocation strategy formulation allocates computing power according to preset rules. Allocation execution allocates the corresponding computing power resources to each computing task according to the formulated allocation strategy, ensuring that the task can obtain the required computing power and begin executing the computing operation.

[0004] However, it still has some shortcomings in actual use, such as difficulty in coping with dynamic changes. Traditional computing power allocation methods lack real-time monitoring and dynamic adjustment mechanisms when facing dynamic changes in computing tasks, and often cannot make effective allocation adjustments in time, resulting in poor adaptability of the system when facing changes; data is incomplete. Traditional computing power allocation mostly makes decisions based on local information, and rarely considers the matching relationship between tasks and resources from the perspective of the entire system, which can easily lead to local optimality and overall non-optimal situations, affecting the overall performance improvement of the system; data is inaccurate. Traditional computing power allocation is usually based on fixed rules or simple indicators, and cannot judge the correlation between tasks or different types of tasks, which affects the accuracy of data in computing power allocation decisions.

[0005] Therefore, it is urgent to provide a method for intelligent allocation of computing power based on big data processing to solve the problems that the existing computing power allocation method is difficult to cope with dynamic changes, incomplete data, and inaccurate data. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for intelligent allocation of computing power based on big data processing, which solves the problems raised in the above-mentioned background technology through the following scheme.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently allocating computing power based on big data processing, comprising:

[0008] S1. Determine the computing power allocation area, which is used to determine the data in the target computing server as the computing power allocation area, and divide it into sub-areas according to the time division method of 5s / second, and number each sub-area in the computing power allocation area in sequence as 1, 2, ..., i, ..., n;

[0009] S2. Data collection, used to obtain computing power resource data and task data during the target computing power allocation process, and transmit the computing power resource data to the computing power resource data processing step, and transmit the task data to the task data processing step;

[0010] S3, computing power resource data processing, is used to analyze and process the computing power resource data obtained in S2 to obtain the hardware resource performance impact coefficient, computing power resource usage impact coefficient, and network data impact coefficient, and transmit them to the comprehensive data analysis step;

[0011] S4, task data processing, is used to analyze and process the task data obtained in S2 to obtain the task complexity influence coefficient and the task resource interaction influence coefficient, and transmit them to the comprehensive data analysis step;

[0012] S5. Comprehensive data analysis: import the hardware resource performance impact coefficient, computing power resource usage impact coefficient, network data impact coefficient, task complexity impact coefficient, and task resource interaction impact coefficient obtained in S3 and S4 into the computing power allocation rationality index mathematical model to obtain the computing power allocation rationality index value, and transmit it to the computing power allocation method judgment step;

[0013] S6: Determine the computing power allocation method, compare the computing power allocation rationality index value obtained in S5 with the preset computing power allocation rationality index values under multiple standards, calculate the minimum value of the difference between the computing power allocation rationality index value and the preset computing power allocation rationality index values under multiple standards, and when the difference value is less than the preset difference value, transmit the judgment result and the data in the target computing server to the human-computer interaction step; when the difference value is greater than the preset difference value, transmit the data in the target computing server to the computing power allocation feedback step;

[0014] S7, computing power allocation feedback, used to increase the computing power resources in the target computing server to the next standard according to S6, and re-transmit it to the data collection step;

[0015] S8, human-computer interaction, is used to transmit the obtained judgment results and the data in the target computing server to the user information terminal, providing reference data for the user to make adjustment measures.

[0016] Preferably, the computing resource data includes parameters affecting hardware resource performance, parameters affecting computing resource usage, and parameters affecting network data; and the task data includes parameters affecting task complexity and parameters affecting task resource interaction.

[0017] Preferably, the parameters affecting the performance of hardware resources include the CPU core frequency, denoted as f; the GPU memory bandwidth, denoted as b; the memory read and write speed, denoted as s; the storage device IOPS, denoted as S; the parameters affecting the usage of computing resources include the CPU usage rate, denoted as u; the GPU load, denoted as l; the remaining memory capacity, denoted as r; the network bandwidth occupancy rate, denoted as o; the network data influencing parameters include the network device packet loss rate, denoted as Lr; the number of network paths, denoted as Nr; the number of network node connection points, denoted as Nc; the task complexity influencing parameters include the number of algorithm nesting layers, denoted as Cl, the number of data dependencies, denoted as Cd, the number of code logic branches, denoted as Tb, and the number of external function calls, denoted as Tf; the task resource interaction influencing parameters include the number of device read and write times, denoted as Tr; the number of network protocol interactions, denoted as Tn; the data transmission volume, denoted as Tc, and the resource mutex waiting time, denoted as T.

[0018] Preferably, the computing power resource data processing step includes a hardware resource performance impact coefficient calculation step, a computing power resource usage impact coefficient calculation step, and a network data impact coefficient calculation step.

[0019] Preferably, the hardware resource performance impact coefficient calculation step is used to import the hardware resource performance impact parameters into the hardware resource performance impact coefficient mathematical model to obtain the hardware resource performance impact coefficient value; the computing power resource usage impact coefficient calculation step is used to import the computing power resource usage impact parameters into the computing power resource usage impact coefficient mathematical model to obtain the computing power resource usage impact coefficient value; the network data impact coefficient calculation step is used to import the network data impact parameters into the network data impact coefficient mathematical model to obtain the network data impact coefficient value.

[0020] Preferably, the mathematical model of the hardware resource performance impact coefficient is specifically:

[0021]

[0022] The mathematical model of the impact coefficient of computing power resource usage is as follows:

[0023]

[0024] The mathematical model of network data influence coefficient is as follows:

[0025]

[0026] where f iIndicates the CPU core frequency in the i-th time period, f max Indicates the maximum CPU core frequency, f min Indicates the minimum CPU core frequency, b i represents the GPU memory bandwidth in the i-th time period, b max Indicates the maximum GPU memory bandwidth, b min Indicates the minimum GPU memory bandwidth; s i Indicates the memory read and write speed in the i-th time period, s max Indicates the maximum memory read and write speed, s min Indicates the minimum memory read and write speed, S i represents the storage device IOPS in the i-th time period, S max Indicates the maximum IOPS of the storage device, u i Indicates the CPU usage in the i-th time period, l i represents the GPU load in the i-th time period; o i represents the network bandwidth utilization rate in the i-th time period, r i Indicates the remaining memory capacity in the i-th time period, r max Indicates the maximum memory capacity, Lr i Indicates the amount of packet loss of the network device in the i-th time period, Lr max Indicates the maximum packet loss of the network device; NC i Indicates the number of network node connection points in the i-th time period, NC i-1 Indicates the number of network node connection points in the i-1th time period, Nr i represents the number of network paths in the i-th time period, Indicates the average number of network paths.

[0027] Preferably, the computing power resource data processing step includes a task complexity influence coefficient calculation step and a task resource interaction influence coefficient calculation step.

[0028] Preferably, the task complexity influence coefficient calculation step is used to import the task complexity influence parameters into the task complexity influence coefficient mathematical model to obtain the task complexity influence coefficient value; the task resource interaction influence coefficient calculation step is used to import the task resource interaction influence parameters into the task resource interaction influence coefficient mathematical model to obtain the task resource interaction influence coefficient value.

[0029] Preferably, the mathematical model of the task complexity influence coefficient is specifically:

[0030]

[0031] The mathematical model of the task-resource interaction coefficient is as follows:

[0032]

[0033] Among them, Cl i Indicates the number of algorithm nesting layers in the i-th time period, Cl max Indicates the maximum number of algorithm nesting layers, Cd i represents the number of data dependencies in the i-th time period, Cd max Indicates the maximum number of data dependencies, Tb i Indicates the number of code logic branches in the i-th time period, Tb max Indicates the maximum number of code logic branches, Tf i Indicates the number of external function calls in the i-th time period, Tf max Indicates the maximum number of external function calls, Tr i Indicates the number of device reads and writes in the i-th time period, Tr max Indicates the maximum number of device read and write times, Tr min Indicates the minimum number of device read and write times, Tn i represents the number of network protocol interactions in the i-th time period, Tn max Indicates the maximum number of network protocol interactions, Tn min Indicates the minimum number of network protocol interactions, Tc i Indicates the amount of data transmitted in the i-th time period, Tc max Indicates the maximum data transmission rate, Tc min Indicates the minimum data transmission volume, T i Indicates the resource mutex waiting time in the i-th time period, T max Indicates the maximum resource mutex lock waiting time.

[0034] Preferably, the mathematical model of the computing power allocation rationality index is specifically:

[0035]

[0036] Technical effects and advantages of the present invention:

[0037] 1. When faced with dynamic changes in computing tasks, the present invention has a real-time monitoring and dynamic adjustment mechanism, which can quickly make effective allocation adjustments, thereby improving its adaptability and ability to cope with dynamic changes;

[0038] 2. This invention adopts a global perspective to make decisions, considers the allocation method of its own computing resources and task requirements from multiple aspects, and comprehensively considers the matching relationship between tasks and resources, thereby effectively avoiding the situation where local optimality leads to overall suboptimality, and significantly improves the overall performance of the system;

[0039] 3. The present invention performs computational analysis based on the essential requirements of the tasks, and can accurately determine the relevance between tasks and task requirements, thereby greatly improving the accuracy of data in computing power allocation decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] As attached Figure 1 The method shown is an intelligent computing power allocation method based on big data processing, including S1, computing area determination, S2, data collection, S3, computing power resource data processing, S4, task data processing, S5, comprehensive data analysis, S6, computing power allocation method judgment, S7, computing power allocation feedback, and S8, human-computer interaction.

[0043] S1. Determine the computing power allocation area, which is used to determine the data in the target computing server as the computing power allocation area, and divide it into sub-areas according to the time division method of 5s / time, and number each sub-area in the computing power allocation area in sequence as 1, 2, ..., i, ..., n.

[0044] S2. Data collection is used to obtain computing power resource data and task data during the target computing power allocation process, and transmit the computing power resource data to the computing power resource data processing step, and transmit the task data to the task data processing step.

[0045] In this embodiment, it should be specifically explained that the computing power resource data includes hardware resource performance influencing parameters, computing power resource usage influencing parameters and network data influencing parameters; task data includes task complexity influencing parameters and task resource interaction influencing parameters.

[0046] In this embodiment, it is specifically necessary to explain that in the present embodiment, it is specifically necessary to explain that the parameters affecting the performance of hardware resources include the CPU core frequency, denoted as f; the GPU memory bandwidth, denoted as b; the memory read and write speed, denoted as s; the storage device IOPS, denoted as S; the parameters affecting the usage of computing resources include the CPU utilization rate, denoted as u; the GPU load, denoted as l; the remaining memory capacity, denoted as r; the network bandwidth occupancy rate, denoted as o; the network data influencing parameters include the network device packet loss rate, denoted as Lr; the number of network paths, denoted as Nr; the number of network node connection points, denoted as Nc; the task complexity influencing parameters include the number of algorithm nesting layers, denoted as Cl, the number of data dependencies, denoted as Cd, the number of code logic branches, denoted as Tb, and the number of external function calls, denoted as Tf; the task resource interaction influencing parameters include the number of device read and write times, denoted as Tr; the number of network protocol interactions, denoted as Tn; the data transmission volume, denoted as Tc, and the resource mutex waiting time, denoted as T.

[0047] S3, computing power resource data processing, is used to analyze and process the computing power resource data obtained in S2 to obtain the hardware resource performance impact coefficient, computing power resource usage impact coefficient, and network data impact coefficient, and transmit them to the comprehensive data analysis step.

[0048] In this embodiment, it should be specifically explained that the computing power resource data processing step includes the hardware resource performance impact coefficient calculation step, the computing power resource usage impact coefficient calculation step, and the network data impact coefficient calculation step.

[0049] In this embodiment, it is specifically necessary to explain that the hardware resource performance impact coefficient calculation step is used to import the hardware resource performance impact parameters into the hardware resource performance impact coefficient mathematical model to obtain the hardware resource performance impact coefficient value; the computing power resource usage impact coefficient calculation step is used to import the computing power resource usage impact parameters into the computing power resource usage impact coefficient mathematical model to obtain the computing power resource usage impact coefficient value; the network data impact coefficient calculation step is used to import the network data impact parameters into the network data impact coefficient mathematical model to obtain the network data impact coefficient value.

[0050] In this embodiment, it should be specifically noted that the mathematical model of the hardware resource performance impact coefficient is:

[0051]

[0052] The mathematical model of the impact coefficient of computing power resource usage is as follows:

[0053]

[0054] The mathematical model of network data influence coefficient is as follows:

[0055]

[0056] where f i Indicates the CPU core frequency in the i-th time period, f max Indicates the maximum CPU core frequency, f min Indicates the minimum CPU core frequency, b i represents the GPU memory bandwidth in the i-th time period, b max Indicates the maximum GPU memory bandwidth, b min Indicates the minimum GPU memory bandwidth; s i Indicates the memory read and write speed in the i-th time period, s max Indicates the maximum memory read and write speed, s min Indicates the minimum memory read and write speed, S i represents the storage device IOPS in the i-th time period, S max Indicates the maximum IOPS of the storage device, u i Indicates the CPU usage in the i-th time period, l i represents the GPU load in the i-th time period; o i represents the network bandwidth utilization rate in the i-th time period, r i Indicates the remaining memory capacity in the i-th time period, r max Indicates the maximum memory capacity, Lr i Indicates the amount of packet loss of the network device in the i-th time period, Lr max Indicates the maximum packet loss of the network device; NC i Indicates the number of network node connection points in the i-th time period, NC i-1 Indicates the number of network node connection points in the i-1th time period, Nr i represents the number of network paths in the i-th time period, Indicates the average number of network paths.

[0057] S4, task data processing, is used to analyze and process the task data obtained in S2 to obtain the task complexity influence coefficient and the task resource interaction influence coefficient, and transmit them to the comprehensive data analysis step.

[0058] In this embodiment, it should be specifically explained that the computing power resource data processing step includes a task complexity influence coefficient calculation step and a task resource interaction influence coefficient calculation step.

[0059] In this embodiment, it should be specifically explained that the task complexity influence coefficient calculation step is used to import the task complexity influence parameters into the task complexity influence coefficient mathematical model to obtain the task complexity influence coefficient value; the task resource interaction influence coefficient calculation step is used to import the task resource interaction influence parameters into the task resource interaction influence coefficient mathematical model to obtain the task resource interaction influence coefficient value.

[0060] In this embodiment, it should be specifically noted that the mathematical model of the task complexity influence coefficient is:

[0061]

[0062] The mathematical model of the task-resource interaction coefficient is as follows:

[0063]

[0064] Among them, Cl i Indicates the number of algorithm nesting layers in the i-th time period, Cl max Indicates the maximum number of algorithm nesting layers, Cd i represents the number of data dependencies in the i-th time period, Cd max Indicates the maximum number of data dependencies, Tb i Indicates the number of code logic branches in the i-th time period, Tb max Indicates the maximum number of code logic branches, Tf i Indicates the number of external function calls in the i-th time period, Tf max Indicates the maximum number of external function calls, Tr i Indicates the number of device reads and writes in the i-th time period, Tr max Indicates the maximum number of device read and write times, Tr min Indicates the minimum number of device read and write times, Tn i represents the number of network protocol interactions in the i-th time period, Tn max Indicates the maximum number of network protocol interactions, Tn min Indicates the minimum number of network protocol interactions, Tc i Indicates the amount of data transmitted in the i-th time period, Tc max Indicates the maximum data transmission rate, Tc min Indicates the minimum data transmission volume, T i Indicates the resource mutex waiting time in the i-th time period, T max Indicates the maximum resource mutex lock waiting time.

[0065] S5. Comprehensive data analysis is performed, and the hardware resource performance impact coefficient values, computing power resource usage impact coefficient values, network data impact coefficient values, task complexity impact coefficient values, and task resource interaction impact coefficient values obtained in S3 and S4 are imported into the computing power allocation rationality index mathematical model to obtain the computing power allocation rationality index value, and the index value is transmitted to the computing power allocation method judgment step.

[0066] In this embodiment, it should be specifically noted that the mathematical model of the computing power allocation rationality index is specifically:

[0067]

[0068] S6. Determine the computing power allocation method. Compare the computing power allocation rationality index value obtained in S5 with the preset computing power allocation rationality index values under multiple standards. Calculate the minimum value of the difference between the computing power allocation rationality index value and the preset computing power allocation rationality index values under multiple standards. When the difference value is less than the preset difference value, transmit the judgment result and the data in the target computing server to the human-computer interaction step. When the difference value is greater than the preset difference value, transmit the data in the target computing server to the computing power allocation feedback step.

[0069] In this embodiment, it should be specifically noted that the preset computing power allocation rationality index value under the multiple standards is the average of the computing power allocation rationality index values under different historical computing power allocation modes, and the preset difference value is the average of the difference between the computing power allocation rationality index value and the preset computing power allocation rationality index value.

[0070] S7, computing power allocation feedback, is used to increase the data in the target computing server to the next standard computing power resource according to S6, and re-transmit it to the data collection step.

[0071] S8, human-computer interaction, is used to transmit the obtained judgment results and the data in the target computing server to the user information terminal, providing reference data for the user to make adjustment measures.

[0072] Secondly: The drawings of the embodiments disclosed in the present invention only involve structures related to the embodiments disclosed in the present invention. Other structures may refer to conventional designs. The same embodiment and different embodiments of the present invention may be combined with each other without conflict.

[0073] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for intelligent allocation of computing power based on big data processing, characterized in that: include: S1. Determine the computing power allocation area, which is used to determine the data in the target computing server as the computing power allocation area, and divide it into sub-areas according to the time division method of 5s / second, and number each sub-area in the computing power allocation area in sequence as 1, 2, ..., i, ..., n; S2. Data collection, used to obtain computing power resource data and task data during the target computing power allocation process, and transmit the computing power resource data to the computing power resource data processing step, and transmit the task data to the task data processing step; S3, computing power resource data processing, is used to analyze and process the computing power resource data obtained in S2 to obtain the hardware resource performance impact coefficient, computing power resource usage impact coefficient, and network data impact coefficient, and transmit them to the comprehensive data analysis step; S4, task data processing, is used to analyze and process the task data obtained in S2 to obtain the task complexity influence coefficient and the task resource interaction influence coefficient, and transmit them to the comprehensive data analysis step; S5. Comprehensive data analysis: import the hardware resource performance impact coefficient, computing power resource usage impact coefficient, network data impact coefficient, task complexity impact coefficient, and task resource interaction impact coefficient obtained in S3 and S4 into the computing power allocation rationality index mathematical model to obtain the computing power allocation rationality index value, and transmit it to the computing power allocation method judgment step; S6: Determine the computing power allocation method, compare the computing power allocation rationality index value obtained in S5 with the preset computing power allocation rationality index values under multiple standards, calculate the minimum value of the difference between the computing power allocation rationality index value and the preset computing power allocation rationality index values under multiple standards, and when the difference value is less than the preset difference value, transmit the judgment result and the data in the target computing server to the human-computer interaction step; when the difference value is greater than the preset difference value, transmit the data in the target computing server to the computing power allocation feedback step; S7, computing power allocation feedback, used to increase the computing power resources in the target computing server to the next standard according to S6, and re-transmit it to the data collection step; S8, human-computer interaction, is used to transmit the obtained judgment results and the data in the target computing server to the user information terminal, providing reference data for the user to make adjustment measures.

2. The method for intelligently allocating computing power based on big data processing according to claim 1, characterized in that: The computing resource data includes parameters affecting hardware resource performance, computing resource usage, and network data; the task data includes parameters affecting task complexity and task resource interaction.

3. The method for intelligently allocating computing power based on big data processing according to claim 2, characterized in that: The parameters affecting the performance of hardware resources include the CPU core frequency, denoted as f; the GPU memory bandwidth, denoted as b; the memory read and write speed, denoted as s; the storage device IOPS, denoted as S; the parameters affecting the usage of computing resources include the CPU usage rate, denoted as u; the GPU load, denoted as l; the remaining memory capacity, denoted as r; the network bandwidth occupancy rate, denoted as o; the network data influencing parameters include the network device packet loss rate, denoted as Lr; the number of network paths, denoted as Nr; the number of network node connection points, denoted as Nc; the parameters affecting the task complexity include the number of algorithm nesting layers, denoted as Cl, the number of data dependencies, denoted as Cd, the number of code logic branches, denoted as Tb, and the number of external function calls, denoted as Tf; the parameters affecting the interaction of task resources include the number of device read and write times, denoted as Tr; the number of network protocol interactions, denoted as Tn; the data transmission volume, denoted as Tc, and the resource mutex waiting time, denoted as T.

4. The method for intelligently allocating computing power based on big data processing according to claim 1, characterized in that: The computing power resource data processing step includes a hardware resource performance impact coefficient calculation step, a computing power resource usage impact coefficient calculation step, and a network data impact coefficient calculation step.

5. The method for intelligently allocating computing power based on big data processing according to claim 4, characterized in that: The hardware resource performance impact coefficient calculation step is used to import the hardware resource performance impact parameters into the hardware resource performance impact coefficient mathematical model to obtain the hardware resource performance impact coefficient value; the computing power resource usage impact coefficient calculation step is used to import the computing power resource usage impact parameters into the computing power resource usage impact coefficient mathematical model to obtain the computing power resource usage impact coefficient value; the network data impact coefficient calculation step is used to import the network data impact parameters into the network data impact coefficient mathematical model to obtain the network data impact coefficient value.

6. The method for intelligently allocating computing power based on big data processing according to claim 5, characterized in that: The mathematical model of the hardware resource performance impact coefficient is specifically: The mathematical model of the impact coefficient of computing power resource usage is as follows: The mathematical model of network data influence coefficient is as follows: where f i Indicates the CPU core frequency in the i-th time period, f max Indicates the maximum CPU core frequency, f min Indicates the minimum CPU core frequency, b i represents the GPU memory bandwidth in the i-th time period, b max Indicates the maximum GPU memory bandwidth, b min Indicates the minimum GPU memory bandwidth; s i Indicates the memory read and write speed in the i-th time period, s max Indicates the maximum memory read and write speed, s min Indicates the minimum memory read and write speed, S i represents the storage device IOPS in the i-th time period, S max Indicates the maximum IOPS of the storage device, u i Indicates the CPU usage in the i-th time period, l i represents the GPU load in the i-th time period; o i represents the network bandwidth utilization rate in the i-th time period, r i Indicates the remaining memory capacity in the i-th time period, r max Indicates the maximum memory capacity, Lr i Indicates the amount of packet loss of the network device in the i-th time period, Lr max Indicates the maximum packet loss of the network device; NC i Indicates the number of network node connection points in the i-th time period, NC i-1 Indicates the number of network node connection points in the i-1th time period, Nr i represents the number of network paths in the i-th time period, Indicates the average number of network paths.

7. The method for intelligently allocating computing power based on big data processing according to claim 1, characterized in that: The computing power resource data processing step includes a task complexity influence coefficient calculation step and a task resource interaction influence coefficient calculation step.

8. The method for intelligently allocating computing power based on big data processing according to claim 7, characterized in that: The task complexity influence coefficient calculation step is used to import the task complexity influence parameter into the task complexity influence coefficient mathematical model to obtain the task complexity influence coefficient value; the task resource interaction influence coefficient calculation step is used to import the task resource interaction influence parameter into the task resource interaction influence coefficient mathematical model to obtain the task resource interaction influence coefficient value.

9. The method for intelligently allocating computing power based on big data processing according to claim 8, characterized in that: The mathematical model of the task complexity influence coefficient is specifically: The mathematical model of the task-resource interaction coefficient is as follows: Among them, Cl i Indicates the number of algorithm nesting layers in the i-th time period, Cl max Indicates the maximum number of algorithm nesting layers, Cd i represents the number of data dependencies in the i-th time period, Cd max Indicates the maximum number of data dependencies, Tb i Indicates the number of code logic branches in the i-th time period, Tb max Indicates the maximum number of code logic branches, Tf i Indicates the number of external function calls in the i-th time period, Tf max Indicates the maximum number of external function calls, Tr i Indicates the number of device reads and writes in the i-th time period, Tr max Indicates the maximum number of device read and write times, Tr min Indicates the minimum number of device read and write times, Tn i represents the number of network protocol interactions in the i-th time period, Tn max Indicates the maximum number of network protocol interactions, Tn min Indicates the minimum number of network protocol interactions, Tc i Indicates the amount of data transmitted in the i-th time period, Tc max Indicates the maximum data transmission rate, Tc min Indicates the minimum data transmission volume, T i Indicates the resource mutex waiting time in the i-th time period, T max Indicates the maximum resource mutex lock waiting time.

10. The method for intelligently allocating computing power based on big data processing according to claim 1, characterized in that: The mathematical model of the computing power allocation rationality index is specifically:

Images