A performance dynamic allocation optimization method and system based on computing power

Through real-time monitoring and classification tasks, combining punishment coefficients and dynamic weight adjustment factors, dynamic hierarchical network resource optimization algorithm and adaptive competition task scheduling optimization algorithm, the problem of low resource utilization rate and inability to respond in time in the existing technology is solved. The problem of this is to achieve optimal utilization of system resources and load balancing.

CN118802801BActive Publication Date: 2024-11-29中雄科技集团股份有限公司
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411282097.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-11-29
Estimated Expiration
2044-09-13

AI Technical Summary

Technical Problem

The existing technology cannot respond to changes in the demand of complex tasks in real time, resulting in low resource utilization. Due to limited system performance, critical tasks may not be able to respond in time, resulting in deadlocks.

Method used

By classifying tasks into calculation-intensive, data-intensive and real-time tasks, the system resource utilization is monitored in real time, the penalty coefficient and dynamic weight adjustment factor are introduced, and the dynamic hierarchical network resource optimization algorithm and adaptive competition task scheduling optimization algorithm are used to achieve real-time Adjust resource allocation and task priorities.

Benefits of technology

The optimal utilization and load balancing of system resources are achieved, ensuring that high-priority tasks can quickly obtain the required resources, and improving the ability of dynamic optimization of system performance and efficient allocation of resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118802801B_ABST
    Figure CN118802801B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of system performance optimization, and specifically, to a method and system for dynamically allocating performance based on computing power. It includes the following steps: S1, classifying tasks, and calculating task priority index based on urgency, resource demand and user priority evaluation; S2, real-time monitoring of resource utilization, triggering a feedback mechanism when exceeding a preset threshold, and calculating a penalty coefficient; S3, using a dynamic hierarchical network resource optimization algorithm to dynamically adjust resource node weights and adjust resource allocation in real time; S4, using an adaptive competitive task scheduling optimization algorithm to monitor the life cycle and competitiveness index of task cells in real time, and using a feedback loop to dynamically adjust task scheduling and optimize system performance. The method and system for dynamically allocating performance based on computing power optimize resource utilization to achieve system load balancing and optimize system performance through a dynamic hierarchical network resource optimization algorithm and an adaptive competitive task scheduling optimization algorithm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of system performance optimization, and in particular to a method and system for dynamically allocating and optimizing performance based on computing power. Background Art

[0002] A performance dynamic allocation optimization method and system based on computing power aims to optimize the utilization efficiency of computing resources and improve the efficiency of task execution. By introducing penalty coefficients and dynamic weight adjustment factors, using dynamic hierarchical network resource optimization algorithms and adaptive competitive task scheduling optimization algorithms, dynamic optimization of system performance and efficient resource allocation are achieved. A feedback loop dynamic adjustment mechanism is adopted to ensure optimal utilization of system resources and load balancing.

[0003] The existing performance dynamic allocation optimization method and system cannot respond to the demand changes of complex tasks and optimize resource utilization in real time. Due to the limited system performance and the urgent task resource requirements, key tasks cannot be responded in time and cause deadlock problems. Therefore, a performance dynamic allocation optimization method and system based on computing power are provided. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for dynamic performance allocation optimization based on computing power to solve the problem raised in the above background technology that due to limited system performance and urgent task resource requirements, key tasks cannot respond in time and cause deadlock.

[0005] To achieve the above object, the present invention provides a method for optimizing dynamic performance allocation based on computing power, comprising:

[0006] S1. Classify tasks into three types, evaluate them according to their urgency, resource requirements, and user-set priorities, and derive the priority index of the tasks;

[0007] S2, real-time monitoring of the system's CPU, GPU, memory and storage. When the usage of any resource exceeds the preset threshold, the feedback mechanism is triggered and the penalty coefficient is calculated;

[0008] S3, when the feedback mechanism is triggered, the dynamic hierarchical network resource optimization algorithm is used to divide the computing resources into three levels and dynamically adjust the weight of each resource node. According to the task priority, dependency and system load, the penalty coefficient and dynamic weight adjustment factor are used to adjust the resource allocation in real time;

[0009] S4. Adopt an adaptive competitive task scheduling optimization algorithm, introduce a penalty coefficient and a task dependency matrix to monitor the life cycle and competitiveness index of task cells in real time. When the competitiveness index of a task cell exceeds the threshold, use the feedback loop dynamic adjustment mechanism to reallocate system resources.

[0010] As a further improvement of the technical solution, in S1, the three types of tasks are computationally intensive tasks, data intensive tasks and real-time tasks;

[0011] Among them, the computation-intensive tasks are used to consume the computing resources of the CPU and GPU; the data-intensive tasks are used to consume memory and storage resources; the real-time tasks are delay-sensitive and time-limited tasks;

[0012] The task priority index is obtained by evaluating the task urgency, resource requirements and priority set by the user. The specific steps are as follows:

[0013] S1.1. User sets task priority , set the urgency of the task , the amount of resources required for the task ;

[0014] S1.2. Define dynamic weight coefficients , and , and satisfies ;

[0015] S1.3. Calculate the priority index of each task :

[0016]

[0017] in, is the weight of the task priority; The weight of the amount of resources required for the task; is the weight of the task urgency;

[0018] As a further improvement of the technical solution, in S2, the feedback mechanism is specifically as follows:

[0019] Preset Threshold : Set a preset usage threshold for CPU, GPU, memory, and storage resources ;

[0020] Detection logic: After each resource usage data collection, the system checks whether the usage of all resources exceeds the preset threshold. If the usage of any resource exceeds , then the feedback mechanism is triggered;

[0021] Among them, the penalty coefficient of calculating the current moment , the specific method is as follows:

[0022]

[0023] in, For CPU usage at the moment; For GPU usage at the moment; For Memory usage at the moment; For Storage disk usage at the moment; is the penalty coefficient gain, which is used to control the growth rate of the penalty coefficient; For The maximum value of all resource usage at the moment.

[0024] As a further improvement of the technical solution, in S3, the dynamic hierarchical network resource optimization algorithm is implemented based on the hierarchical network model, and resource allocation is adjusted in real time through hierarchical management, dynamic weight adjustment and task priority calculation. The specific steps are as follows:

[0025] S3.1. Divide system computing resources into three levels, including basic level, intermediate level and advanced level;

[0026] S3.2, classify all tasks into three types of tasks, including computation-intensive, data-intensive, and real-time tasks;

[0027] S3.3. Calculate the priority index of each task ;

[0028] S3.4、Monitor each task Success rate at all times and total resource consumption , calculate the dynamic weight adjustment factor ;

[0029] S3.5. Build a task dependency matrix , and temporarily update the resource node Weight ;

[0030] S3.6. Based on the updated resource node weight Allocate resources in real time.

[0031] As a further improvement of the technical solution, in S3.4, the dynamic weight adjustment factor is calculated , the specific method is as follows:

[0032]

[0033] in, For Time Mission success rate; For Time Mission The total resource consumption.

[0034] As a further improvement of the present technical solution, in S3.5, temporarily updating the resource node Weight , the specific method is as follows:

[0035]

[0036] in, For Time Mission In the resource node The weight of For The amount of resources available at any given time; For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies of For Moment Resource Node The sum of resource requirements on For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies.

[0037] As a further improvement of the technical solution, in S4, the adaptive competitive task scheduling optimization algorithm is based on the adaptive competitive model and the task cell model, monitors the life cycle of the task cell and calculates the competitiveness index of the task cell, dynamically schedules tasks and optimizes resource allocation, and the specific method is as follows:

[0038] S4.1.1. Each task Treated as an independent task cell ;

[0039] S4.1.2. According to the progress of the task and resource consumption Dynamically adjust the life cycle of task cells ;

[0040] S4.1.3. Priority index of combined tasks , resource requirements, dependencies and system load to calculate the competitiveness index of task cells .

[0041] As a further improvement of this technical solution, in S4.1.2, according to the progress of the task and resource consumption Dynamically adjust the life cycle of task cells , the specific method is as follows:

[0042]

[0043] in, is the initial life cycle; For Time Mission The percentage of progress completed; For Time Mission resource consumption;

[0044] Calculate the competitiveness index of the task cells as described in S4.1.3 , the specific method is as follows:

[0045]

[0046] in, For Time Mission The priority index of For Time Mission The amount of resources required; is the weighted sum of the task dependency matrix and the dependent task status; For The amount of resources available at any given time.

[0047] As a further improvement of the technical solution, in S4, the feedback loop dynamic adjustment mechanism is based on feedback control principle and system status monitoring technology, and is used to adjust the priority and resource allocation strategy of the task in real time. The specific steps are as follows:

[0048] S4.2.1. Initialize the feedback loop mechanism and define the primary feedback loop and the secondary feedback loop;

[0049] S4.2.2. Setting the competitiveness index threshold of task cells ;

[0050] S4.2.3. Competitiveness index of task cells Exceeding the threshold When the primary feedback loop is triggered, the task is adjusted dynamically. Priority index and resource allocation strategies, feeding back the adjusted task priorities and resource allocation results to the task cells;

[0051] S4.2.4, through the secondary feedback loop, comprehensively evaluate the status of all task cells and dynamically adjust the global resource allocation strategy;

[0052] S4.2.5. Continuously perform primary and secondary feedback loop adjustments, monitor and adjust task priorities and resource allocation in real time, and dynamically adjust feedback loop parameters and thresholds.

[0053] On the other hand, the present invention provides a performance dynamic allocation optimization system based on computing power, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned performance dynamic allocation optimization method based on computing power.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. The performance dynamic allocation optimization method and system based on computing power introduces a penalty coefficient and a dynamic weight adjustment factor. By using a dynamic hierarchical network resource optimization algorithm, the system computing resources can be divided into a basic layer, an intermediate layer, and an advanced layer. Refined resource allocation is performed for tasks with different priorities and computing requirements, thereby optimizing resource utilization efficiency.

[0056] 2. In this computing power-based performance dynamic allocation optimization method and system, an adaptive competitive task scheduling optimization algorithm and a feedback loop dynamic adjustment mechanism are used to achieve real-time adjustment of task priority and resource allocation strategies, ensuring that high-priority and urgent tasks can quickly obtain the required resources. The system resource allocation strategy is adjusted globally, and the status of all tasks is comprehensively considered to ensure optimal utilization of system resources and load balancing. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 The figure is a flow chart of the overall method of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention. Example 1

[0059] See also Figure 1As shown, this embodiment provides a performance dynamic allocation optimization method based on computing power, including the following steps:

[0060] S1. Classify tasks into three types, evaluate them according to their urgency, resource requirements, and user-set priorities, and derive the priority index of the tasks;

[0061] The three types of tasks are compute-intensive tasks, data-intensive tasks, and real-time tasks;

[0062] Among them, computationally intensive tasks are used to consume CPU and GPU computing resources; data-intensive tasks are used to consume memory and storage resources; real-time tasks are tasks that are sensitive to delays and have time limits;

[0063] The priority index of the task is obtained by evaluating the urgency of the task, resource requirements and the priority set by the user. The specific steps are as follows:

[0064] S1.1. User sets task priority , set the urgency of the task , the amount of resources required for each task ;

[0065] S1.2. Define dynamic weight coefficients , and , and satisfies ;

[0066] Among them, these weight coefficients can be dynamically adjusted according to the real-time status and strategy of the system. When system resources are tight, the weight of resource demand can be increased; when the task backlog is serious, the weight of urgency can be increased;

[0067] S1.3. Calculate the priority index of each task :

[0068] .

[0069] in, A priority set for the user, usually a constant or discrete value; The amount of resources required for each task is weighted summed according to the task's demand for different resources; The urgency level is set directly according to the importance level of the task; The specific calculation method is as follows:

[0070]

[0071] in, The CPU requirement for the task; GPU requirements for the task; The memory requirement of the task; Storage requirements for tasks; The maximum CPU requirement for the task; The maximum value of the GPU required by the task; is the maximum memory requirement of the task; is the maximum storage requirement of the task;

[0072] Among them, the priority index of the task is obtained The role and purpose of is mainly to guide resource allocation and task scheduling to ensure the optimal use of system resources. By calculating the priority index of the task, the system can ensure that high-priority and urgent tasks are completed within the specified time to meet the needs of users and applications. Resource allocation and task scheduling based on the priority index can maximize resource utilization and avoid resource idleness or overload, thereby improving the operating efficiency of the system. It can respond to changes in resource usage and task requirements in real time and flexibly adjust resource allocation strategies.

[0073] S2, real-time monitoring of the system's CPU, GPU, memory and storage. When the usage of any resource exceeds the preset threshold, the feedback mechanism is triggered and the penalty coefficient is calculated;

[0074] The feedback mechanism is as follows:

[0075] Preset Threshold : Set a preset usage threshold for CPU, GPU, memory, and storage resources ;

[0076] Detection logic: After each resource usage data collection, the system checks whether the usage of all resources exceeds the preset threshold. If the usage of any resource exceeds , then the feedback mechanism is triggered;

[0077] The feedback mechanism can monitor the usage of system resources in real time and trigger adjustments when the resource usage rate exceeds the preset threshold to ensure that resource allocation is more reasonable and efficient. By introducing a penalty coefficient, the feedback mechanism can increase the resource allocation of high-priority tasks and reduce the resource usage of low-priority tasks when resources are tight. The feedback mechanism balances the resource load, avoids resource overload and system bottlenecks, maintains the stable operation of the system, and adjusts the resource allocation strategy in time to prevent the system from crashing or performance degradation due to excessive resource use. The feedback mechanism can respond to changes in system status in a timely manner, adjust the resource allocation strategy in real time, and adapt to dynamically changing workloads and task requirements.

[0078] Among them, calculate the penalty coefficient at the current moment , the specific method is as follows:

[0079]

[0080] in, For CPU usage at the moment; For GPU usage at the moment; For Memory usage at the moment; For Storage disk usage at the moment; is the penalty coefficient gain, which is used to control the growth rate of the penalty coefficient; For The maximum value of all resource usage at the moment;

[0081] By introducing the penalty coefficient , can avoid excessive concentration of resources on certain tasks. When the utilization rate of some resources is too high, Increase and reduce the priority of these tasks to make resource allocation more balanced; balance the system load and reduce the risk of system overload or resource exhaustion. The introduction of penalty coefficients enables the system to dynamically adjust resource allocation strategies and respond flexibly according to real-time resource usage and task requirements, thereby improving the system's elasticity and adaptability.

[0082] S3, when the feedback mechanism is triggered, the dynamic hierarchical network resource optimization algorithm is used to divide the computing resources into three levels and dynamically adjust the weight of each resource node. According to the task priority, dependency and system load, the penalty coefficient and dynamic weight adjustment factor are used to adjust the resource allocation in real time;

[0083] In S3, the dynamic hierarchical network resource optimization algorithm is implemented based on the hierarchical network model. Through hierarchical management, dynamic weight adjustment and task priority calculation, resource allocation is adjusted in real time. The specific steps are as follows:

[0084] S3.1. Divide system computing resources into three levels, including basic level, intermediate level and advanced level;

[0085] Among them, the basic layer: handles low-priority and low-computing tasks; the intermediate layer: handles medium-priority and moderate-computing tasks; the advanced layer: handles high-priority and high-computing tasks;

[0086] The benefits of dividing system computing resources into basic, intermediate and advanced layers are: through hierarchical management, refined resource allocation can be performed for tasks with different priorities and computing requirements, optimizing resource utilization efficiency and ensuring that high-priority and high-demand tasks are fully supported by resources;

[0087] S3.2, classify all tasks into three types of tasks, including computation-intensive, data-intensive, and real-time tasks;

[0088] S3.3. Calculate the priority index of each task ;

[0089] S3.4、Monitor each task and total resource consumption , calculate the dynamic weight adjustment factor ;

[0090] Each task is at time Success rate The ratio of completed workload to total workload; total resource consumption It is a dynamic indicator used to measure the task at the moment resource usage;

[0091] S3.5. Build a task dependency matrix , and temporarily update the resource node Weight ;

[0092] S3.6. Based on the updated resource node weight Allocate resources in real time.

[0093] Among them, the task dependency matrix is ​​constructed in S3.5 , the specific method is as follows:

[0094] S3.5.1. List all tasks that need to be performed and record the dependencies of each task;

[0095] S3.5.2. Assign weights based on the importance of dependencies. The weight is a value in the range [0,1], indicating the strength of the dependency.

[0096] S3.5.3. Create a The matrix ,in is the total number of tasks. The initial values ​​of the matrix are all 0. The matrix is ​​filled according to the dependencies between tasks and the assigned weights. Elements Indicates the task Dependent tasks the extent of;

[0097] Calculate dynamic weight adjustment factor , the specific method is as follows:

[0098]

[0099] in, For Time Mission success rate; For Time Mission The total resource consumption.

[0100] Introducing dynamic weight adjustment factors The task priority can be adjusted in real time according to the task success rate and resource consumption, optimizing system resource utilization, improving the timeliness of task completion and overall system performance;

[0101] Temporary update resource node Weight , the specific method is as follows:

[0102]

[0103] in, For Time Mission In the resource node The weight of For The amount of resources available at any given time; For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies of For Moment Resource Node The sum of resource requirements on For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies.

[0104] S4. Adopt an adaptive competitive task scheduling optimization algorithm, introduce a penalty coefficient and a task dependency matrix to monitor the life cycle and competitiveness index of task cells in real time. When the competitiveness index of a task cell exceeds the threshold, use the feedback loop dynamic adjustment mechanism to reallocate system resources.

[0105] The adaptive competitive task scheduling optimization algorithm is based on the adaptive competitive model and the task cell model. It monitors the life cycle of the task cell and calculates the competitiveness index of the task cell, dynamically schedules tasks and optimizes resource allocation. The specific methods are as follows:

[0106] S4.1.1. Each task Treated as an independent task cell ;

[0107] S4.1.2. According to the progress of the task and resource consumption Dynamically adjust the life cycle of task cells ;

[0108] S4.1.3. Priority index of combined tasks , resource requirements, dependencies and system load to calculate the competitiveness index of task cells .

[0109] The introduction of the adaptive competition model and task cell model can effectively solve the problems of resource competition and task delays. For example, in a high-performance computing cluster, multiple scientific research tasks need to be calculated at the same time. Each task has different requirements and urgency for computing resources. Through the adaptive competition model and task cell model, the progress and resource usage of each task can be monitored in real time, and urgent and high-priority scientific research tasks can be prioritized and the system load can be balanced.

[0110] According to the task progress and resource consumption Dynamically adjust the life cycle of task cells , the specific method is as follows:

[0111]

[0112] in, is the initial life cycle; For Time Mission The percentage of progress completed; For Time Mission resource consumption;

[0113] Calculation of the competitiveness index of task cells in S4.1.3 , the specific method is as follows:

[0114]

[0115] in, For Time Mission The priority index of For Time Mission The amount of resources required; is the weighted sum of the task dependency matrix and the dependent task status; For The amount of resources available at any given time.

[0116] The feedback loop dynamic adjustment mechanism is based on feedback control principles and system status monitoring technology, and is used to adjust task priorities and resource allocation strategies in real time. The specific steps are as follows:

[0117] S4.2.1. Initialize the feedback loop mechanism and define the primary feedback loop and the secondary feedback loop;

[0118] Among them, the primary feedback loop is used to adjust the priority and resource allocation of a single task in real time; the secondary feedback loop is used to globally adjust the system resource allocation strategy, taking into account the status of all tasks;

[0119] S4.2.2. Setting the competitiveness index threshold of task cells ;

[0120] S4.2.3. Competitiveness index of task cells Exceeding the threshold When the primary feedback loop is triggered, the task is adjusted dynamically. Priority index and resource allocation strategies, feeding back the adjusted task priorities and resource allocation results to the task cells;

[0121]

[0122]

[0123] in, The adjustment step size is used to control the magnitude of the priority adjustment; The adjustment step size is used to control the magnitude of resource allocation adjustment. and The value principle is: and The value should ensure the stability of the system adjustment process and avoid system fluctuations caused by excessive adjustment. The value should ensure that the system can respond to changes in resource requirements and priorities of tasks in a timely manner. and The specific value usually needs to be determined through experiments and tuning; = 0.1 and =0.05 is the starting point for tuning; gradually adjust through experiments and monitoring system operation. and Value; can be adjusted dynamically according to actual load and task requirements and Get value;

[0124] S4.2.4, through the secondary feedback loop, comprehensively evaluate the status of all task cells and dynamically adjust the global resource allocation strategy;

[0125] S4.2.5. Continuously perform primary and secondary feedback loop adjustments, monitor and adjust task priorities and resource allocation in real time, and dynamically adjust feedback loop parameters and thresholds. Example 2

[0126] This embodiment provides a performance dynamic allocation optimization system based on computing power, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the above-mentioned performance dynamic allocation optimization method based on computing power.

[0127] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A performance dynamic allocation optimization method based on computing power, characterized in that: The following steps are involved: S1. Classify tasks into three types, evaluate them according to their urgency, resource requirements, and user-set priorities, and derive the priority index of the tasks; In S1, the three types of tasks are computationally intensive tasks, data intensive tasks, and real-time tasks; Among them, the computation-intensive tasks are used to consume the computing resources of the CPU and GPU; the data-intensive tasks are used to consume memory and storage resources; the real-time tasks are delay-sensitive and time-limited tasks; The task priority index is obtained by evaluating the task urgency, resource requirements and priority set by the user. The specific steps are as follows: S1.

1. User sets task priority level , set the urgency of the task , the amount of resources required for the task ; S1.

2. Define dynamic weight coefficients , and , and satisfies ; S1.

3. Calculate the priority index of each task : ; in, is the weight of the task priority; The weight of the amount of resources required for the task; is the weight of the task urgency; S2, real-time monitoring of the system's CPU, GPU, memory and storage. When the usage of any resource exceeds the preset threshold, the feedback mechanism is triggered and the penalty coefficient is calculated; In S2, the feedback mechanism is as follows: Preset Threshold : Set a preset usage threshold for CPU, GPU, memory, and storage resources ; Detection logic: After each resource usage data collection, the system checks whether the usage of all resources exceeds the preset threshold. If the usage of any resource exceeds , then the feedback mechanism is triggered; Among them, the penalty coefficient of calculating the current moment , the specific method is as follows: ; in, For CPU usage at the moment; For GPU usage at the moment; For Memory usage at the moment; For Storage disk usage at the moment; is the penalty coefficient gain, which is used to control the growth rate of the penalty coefficient; For The maximum value of all resource usage at the moment; S3, when the feedback mechanism is triggered, the dynamic hierarchical network resource optimization algorithm is used to divide the computing resources into three levels and dynamically adjust the weight of each resource node. According to the task priority, dependency and system load, the penalty coefficient and dynamic weight adjustment factor are used to adjust the resource allocation in real time; In S3, the dynamic hierarchical network resource optimization algorithm is implemented based on the hierarchical network model, and adjusts resource allocation in real time through hierarchical management, dynamic weight adjustment and task priority calculation. The specific steps are as follows: S3.

1. Divide system computing resources into three levels, including basic level, intermediate level and advanced level; S3.2, classify all tasks into three types of tasks, including computation-intensive, data-intensive, and real-time tasks; S3.

3. Calculate the priority index of each task ; S3.4、Monitor each task Success rate at all times and total resource consumption , calculate the dynamic weight adjustment factor ; S3.

5. Build a task dependency matrix , and temporarily update the resource node Weight ; S3.

6. Based on the updated resource node weight Allocate resources in real time; S4, using adaptive competitive task scheduling optimization algorithm, introducing penalty coefficient and task dependency matrix to monitor the life cycle and competitiveness index of task cells in real time. When the competitiveness index of task cells exceeds the threshold, the feedback loop dynamic adjustment mechanism is used to reallocate system resources. In S4, the adaptive competitive task scheduling optimization algorithm is based on the adaptive competitive model and the task cell model, monitors the life cycle of the task cell and calculates the competitiveness index of the task cell, dynamically schedules tasks and optimizes resource allocation. The specific method is as follows: S4.1.

1. Each task Treated as an independent task cell ; S4.1.

2. According to the progress of the task and resource consumption Dynamically adjust the life cycle of task cells ; S4.1.

3. Priority index of combined tasks , resource requirements, dependencies and system load to calculate the competitiveness index of task cells .

2. The method for optimizing the dynamic performance allocation based on computing power according to claim 1, characterized in that: In S3.4, the dynamic weight adjustment factor is calculated , the specific method is as follows: ; in, For Time Mission success rate; For Time Mission The total resource consumption.

3. The method for optimizing the dynamic performance allocation based on computing power according to claim 2, characterized in that: In S3.5, temporarily updating the resource node Weight , the specific method is as follows: ; in, For Time Mission In the resource node The weight of For The amount of resources available at any given time; For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies of For Moment Resource Node The sum of resource requirements on For Time Mission The priority index of For Time Mission Dynamic weight adjustment factor of For the task The sum of dependencies.

4. The method for optimizing the dynamic performance allocation based on computing power according to claim 3 is characterized in that: According to S4.1.2, according to the task progress and resource consumption Dynamically adjust the life cycle of task cells , the specific method is as follows: ; in, is the initial life cycle; For Time Mission The percentage of progress completed; For Time Mission resource consumption; Calculate the competitiveness index of the task cells as described in S4.1.3 , the specific method is as follows: ; in, For Time Mission The priority index of For Time Mission The amount of resources required; is the weighted sum of the task dependency matrix and the dependent task status; For The amount of resources available at any given time.

5. The method for optimizing the dynamic performance allocation based on computing power according to claim 4, characterized in that: In S4, the feedback loop dynamic adjustment mechanism is based on feedback control principle and system status monitoring technology, and is used to adjust the priority of tasks and resource allocation strategy in real time. The specific steps are as follows: S4.2.

1. Initialize the feedback loop mechanism and define the primary feedback loop and the secondary feedback loop; S4.2.

2. Setting the competitiveness index threshold of task cells ; S4.2.

3. Competitiveness index of task cells Exceeding the threshold When the primary feedback loop is triggered, the task is adjusted dynamically. Priority index and resource allocation strategies, feeding back the adjusted task priorities and resource allocation results to the task cells; S4.2.4, through the secondary feedback loop, comprehensively evaluate the status of all task cells and dynamically adjust the global resource allocation strategy; S4.2.

5. Continuously perform primary and secondary feedback loop adjustments, monitor and adjust task priorities and resource allocation in real time, and dynamically adjust feedback loop parameters and thresholds.

6. A performance dynamic allocation optimization system based on computing power, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes a computer program to implement the computing power-based performance dynamic allocation optimization method as described in any one of claims 1-5.

Citation Information

Patent Citations

  • Non-IT resource allocation method and system at data center

    CN108712480A

  • Implementation method of satellite-ground convergence network resource allocation strategy based on Q learning

    CN113055489A

  • Method for realizing high-energy-efficiency calculation unloading through strategy gradient algorithm in multi-unmanned aerial vehicle assisted mobile edge calculation

    CN117499867A

  • Network integrated dynamic resource routing management system and method

    CN118590430A

  • Method and apparatus for allocation of resources

    US20040136379A1