CN117687774A - Task model training method and computing power scheduling method and system for computing power scheduling - Google Patents

Abstract

The invention provides a task model training method for computing power dispatching and a computing power dispatching method and system, wherein the computing power dispatching method comprises the following steps: acquiring a current task to be scheduled, wherein the task comprises a task type and a first resource usage amount applied for executing the task; according to the task model obtained through training, under the condition that the task performance is not influenced or within an acceptable range, carrying out dynamic telescopic adjustment on the application quota of the resource with small influence on the task performance in the first resource usage amount to obtain the resource usage amount allocated for executing the task; and performing computational scheduling on the task according to the allocated resource usage. The method and the device can complete self-adaptive resource dynamic scheduling by integrating the influence of the multidimensional resource on the task performance, and effectively improve the task deployment completion condition and the multidimensional resource utilization rate in AI computing power scheduling.

Description

Task model training method and computing power scheduling method and system for computing power scheduling

Technical field

The present invention relates to the field of cloud computing, specifically to the technical resource scheduling field of cloud computing, and in particular to a task model training method for computing power scheduling and a computing power scheduling method and system.

Background technique

With the widespread application of artificial intelligence technology in life in recent years, more and more intelligent computing power clusters have been built. In the field of cloud computing resource scheduling, the cluster scheduling problem of artificial intelligence (AI) computing power is proposed to solve the problem of utilization of AI computing power resources and deployment efficiency of machine learning tasks, and ultimately achieve the computing efficiency of AI computing power. increase. At present, AI computing cluster scheduling mainly analyzes the characteristics of machine learning (ML) tasks and optimizes the allocation of graphics processing unit (GPU) resources in AI computing to improve ML task performance or GPU Resource utilization. For example, consider the characteristics of periodicity, preemption, and location sensitivity of ML tasks during the training process to guide GPU allocation and sharing, meet the efficient scheduling of different types of training task requirements, and improve GPU resource utilization. Existing resource scheduling methods for artificial intelligence computing power usually only design scheduling strategies for a single dimension of GPU resources. However, actual task performance is not determined by considering only a single GPU resource allocation, but is determined by the joint action of multi-dimensional resources. is determined, so the design performance of existing scheduling methods cannot be fully realized in actual production clusters. Moreover, this AI computing power scheduling method cannot fully utilize the overall resources of multiple dimensions in the AI cluster, resulting in a waste of limited cluster resources.

Contents of the invention

In order to solve the above problems, the present invention proposes a task model constructed by combining machine learning methods, and uses it to scale and adjust resource allocation for tasks in the current dynamic queue to complete computing power scheduling.

According to the first aspect, embodiments of the present invention provide a training method for a task model for computing power scheduling, including: using task execution time, task GPU utilization, task average memory usage, task CPU core number, and task type as training Sample; wherein, the task GPU utilization, the average memory usage of the task, the number of CPU cores of the task, and the task type are used as the input of the task model, and the task execution time is labeled as the expected output value; according to the training sample The task model is trained based on the data in the gradient boosted regression tree; the trained task model is used as a prediction model of task execution time in the computing resource allocation and scheduling method.

Preferably, the task types in the same training sample are the same.

Preferably, the task types include computer vision tasks, natural language processing tasks, reinforcement learning tasks, graph neural network tasks, and recommendation tasks.

Preferably, the depth of the regression tree is 10, the number of base learners used is 100, and the learning rate is 0.1.

According to the second aspect, an embodiment of the present invention provides a method of computing power scheduling, including: obtaining a task currently to be scheduled, where the task includes a task type and a first resource usage amount applied for the execution of the task; according to The task model obtained by any of the methods in the first aspect dynamically adjusts the application quota for resources that have a small impact on task performance in the first resource usage without affecting task performance or within an acceptable range. , obtain the resource usage allocated for the execution of the task; perform computing power scheduling on the task according to the allocated resource usage.

Preferably, the task model obtained according to any method in the first aspect applies for resources that have a small impact on task performance in the first resource usage without affecting task performance or within an acceptable range. The quota is dynamically scaled and adjusted to obtain the resource usage allocated for the execution of the task, including: using the task model obtained according to any method in the first aspect, and outputting the resource usage according to the task type and the first resource usage. The first expected execution time of the above task; the resource application quota that has a small impact on task performance in the resource dimension is scaled and adjusted without affecting task performance or within an acceptable range to obtain the second resource usage ; According to the second resource usage, use the task model obtained by any method in the first aspect to output the second expected execution time of the task; between the first expected execution time and the second expected execution time When the difference in execution time is below a preset difference threshold, the second resource usage is used as the allocated resource usage.

Preferably, the method further includes: if the difference between the first expected execution time and the second expected execution time exceeds a set difference threshold, using another resource scaling rate that has a small impact on task performance in the resource dimension. Apply for quotas and make resource scaling adjustments.

Preferably, the method further includes: after using all the resource scaling rates to perform resource scaling adjustment on the resource dimension, the execution time difference between the first expected execution time and the second expected execution time exceeds the difference. When the threshold is reached, the first resource usage is used as the allocated resource usage.

Preferably, using the task model to perform resource scaling adjustment under a set resource scaling rate for the resource dimension to obtain the second resource usage includes: filtering out the requirements for the task from the resource dimension. The resource dimension whose execution time impact is smaller than the set impact threshold is used as the resource dimension to be scaled; resource scaling adjustment is performed under the set resource scaling rate for the resource dimension to be scaled to obtain the second resource usage.

According to the third aspect, an embodiment of the present invention provides a computing power scheduling system, which includes: a server, a task queue, and a scheduler. It is characterized in that it also includes a task history running log database and a task as described in any one of the first aspects. Model; the task historical running log database is used to collect and save historical data of tasks, including resource usage, execution time, and task type; the task model is used to identify resources that have a small impact on task performance in tasks to be scheduled. The application quota is dynamically scaled and adjusted to obtain the resource usage allocated for the execution of the task.

According to a fourth aspect, embodiments of the present invention provide a storage medium in which computer-executable instructions are stored. When the computer-executable instructions are loaded and executed by a processor, the methods of any one of the first and second aspects are implemented. step.

Compared with the prior art, the advantages of the present invention are:

Compared with the current AI computing power scheduling method that is limited to GPU resource allocation, the scheduling method of the present invention can comprehensively integrate the impact of multi-dimensional resources on task performance to complete adaptive resource dynamic scheduling, effectively improving the completion of task deployment in AI computing power scheduling. situation and multidimensional resource utilization. This method of dynamically adjusting resource requirements can optimize the resource overapplication problem that occurs during user task application and reduce resource waste. At the same time, this method also optimizes the inefficient use of some dimensional resources in ML tasks due to the mismatch between resources in various dimensions, and improves the overall resource utilization efficiency of AI computing power.

Description of the drawings

The embodiments of the present invention will be further described below with reference to the accompanying drawings, wherein:

Figure 1 is a schematic diagram of the architecture of a computing power scheduling system according to an embodiment of the present invention;

Figure 2 is a schematic flowchart of a computing power scheduling method based on a task model according to an embodiment of the present invention;

Figure 3 is a flow chart of multi-dimensional resource scaling and scheduling steps according to an embodiment of the present invention;

Figure 4 is a schematic diagram of the hardware device architecture equipped with a scheduling system according to an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below through specific embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not intended to limit the present invention.

When conducting research on AI computing power scheduling, the inventor found that the reason for the low utilization rate of existing computing power resources and even waste of resources is that only GPU resources are considered during the scheduling process and the overall impact of multiple dimensions of resources on actual task performance is ignored. Impact. In order to improve the task running performance and resource utilization of AI computing power, and overcome the shortcomings of existing AI computing power scheduling methods that only design scheduling strategies for the dimension of GPU resources, the inventor combined the analysis of existing multi-dimensional resources and the actual production-oriented The analysis of the cluster proposes to use machine learning to model the impact of multi-dimensional resource requirements of machine learning (ML) tasks in different types of AI computing clusters on task execution, and complete adaptive resource scaling allocation and scheduling based on the status of computing resources to guide Resource scheduling in AI computing power scenarios to improve task deployment execution efficiency on the user task side and resource utilization on the hardware resource side. In the present invention, the multi-dimensional resources usually include the number of CPU cores, GPU usage, and memory usage; in some embodiments, they may also include the capacity to apply for network resources.

Figure 1 is a schematic diagram of the architecture of a computing power scheduling system according to an embodiment of the present invention. The computing power scheduling system is composed of users, different types of servers, task queues, task historical running log databases, task resource performance models (hereinafter referred to as task models) and schedulers. Among them, the user is the subject who submits machine learning tasks and can submit tasks online or offline. The task queue of the current scheduling round can be updated according to the tasks submitted by the user. The server is the main body of running tasks, and the process of computing power scheduling is to deploy tasks to the server; server types include core computing resources and edge computing resources, including physical machines, virtual machines, edge computing nodes, etc. The task queue is responsible for maintaining a dynamic task list composed of online submitted user tasks and saving all tasks waiting for scheduling. In the process of computing power scheduling, tasks that can be executed under the current server status need to be selected from the task queue for computing power scheduling. . The task historical running log database is responsible for collecting and saving historical data of tasks, including resource usage, execution time, task type, etc.; in some embodiments, historical data includes task start time, task end time, task GPU utilization, task Average memory usage, number of task CPU cores, and task type (i.e., the application type of the task load). The task model is responsible for guiding the adaptive resource scaling and adjustment process of tasks during the scheduling process.

The operation process of the computing power scheduling system is as follows: the scheduler updates the dynamic task queue according to the tasks dynamically submitted by the user, as well as specific task attributes such as task type and multi-dimensional resource requirements. For the current scheduling of each round: at the task level, the scheduler combines the task model to adaptively adjust the resource requirements of the current user tasks in the task queue, and returns the updated task demand resource amount to the scheduler; cluster At the lower level, the scheduler combines the task demand scaling results and the multi-dimensional resource usage status of the cluster to deploy tasks to server nodes that can meet the resource needs to achieve dynamic optimization of resource allocation for cluster jobs. Finally, the scheduling system updates the execution status at the task level and the resource status at the cluster level, and prepares for the next round of scheduling.

According to an embodiment of the present invention, a task model training method is proposed. Based on the historical data in the task historical operation log database, the machine learning regression analysis method is used to complete the construction of the task model. Specifically, the modeling process is based on the machine learning method of extreme gradient boosting (XGBoost) to analyze the multi-dimensional resource requirements of different types of tasks, that is, the impact of multi-dimensional resources on task execution time. XGBoost is based on the boosting (Boosting) method to train a set of decision trees, and uses the gradient boosting method to iteratively fit the residuals of the decision trees. The output result obtained by the XGBoost model is the weighted sum of the predicted values of all residual iterative decision trees. The objective function of the task resource demand prediction model based on XGBoost consists of two parts: a loss function that measures the model's fitting degree and a regularization term that increases the complexity of the model. The process of establishing the model structure of the task resource demand prediction model requires minimizing the objective function value. The specific method is to select the feature value segmentation points of leaf nodes by maximizing the branch gain, and sequentially build new branches of the tree model for each feature in the model, and finally complete the XGBoost model of modeling task resource requirements. Each time, the feature with the largest branch gain is selected as the segmentation feature of the leaf node, and the resource features are selected in turn to complete the construction of the task resource demand prediction model. In this way, the total objective function of the model is the smallest, and the task resource demand prediction model for data fitting prediction is completed. Preferably, the training method includes: taking task execution time, task GPU utilization, task average memory occupancy, task CPU core number, and task type as training samples; wherein, task GPU utilization, task average memory occupancy, task CPU core The number and task type are used as the input of the task model, and the task execution time (that is, the start time of task execution to the execution end time) is the label as the expected output value. The features (i.e. input) in the training sample are task GPU utilization, task average memory usage, task CPU core number, and task type. The label (that is, the output) in the training sample is the task execution time; the task execution time is the start time of task execution to the execution end time. Since the content of the original tracking log record is the task start and end time, the task execution time in the training sample needs to be calculated through data preprocessing. The task types involved in the training samples can be different, but in order to achieve more accurate computing power scheduling, in some embodiments, when training a task model, the task types in the training samples are the same, that is, for different tasks. Type trains the corresponding task model. The task types may include computer vision tasks, natural language processing tasks, reinforcement learning tasks, graph neural network tasks, recommendation tasks and other common machine learning task types in various fields. The task model is trained according to the data in the training sample; in some embodiments, modeling is performed based on the gradient boosting tree regression method, that is, the regression method provided in the XGBoost library, and the task regression model corresponding to the current task type uses task resources Usage regression task execution time, that is, the input is resource status and task type, and the output is the predicted execution time of the task. In some embodiments, during the training process, the depth of the regression tree is 10, the number of base learners used in Boosting ensemble learning is 100, and the learning rate is 0.1. The trained task model is used as a prediction model for task execution time in the computing power scheduling method.

The embodiment of the present invention constructs task models of different task types based on machine learning methods, which are used to guide dynamic resource scaling in the online scheduling process, which can solve the problem of resource overapplication for tasks in the resource scheduling and allocation process, and improve the utilization rate and efficiency of cluster resources. Task deployment completion status.

The computing power scheduling system of the present invention is based on task performance modeling to complete the scaling adjustment of online task resource demand allocation. The system starts from two levels: user tasks and cluster resources to optimize the resource scheduling of AI computing power. Specifically, it includes: at the task level, based on the expected task performance and task resource performance model, adaptive scaling adjustment of task requirements to improve task deployment execution efficiency; and at the cluster level, based on the multi-dimensional resource utilization status, resource utilization status Allocation and scheduling were optimized, and the scheduling strategy for ML tasks after resource scaling was completed, completing the improvement of resource utilization in AI computing power scheduling.

Figure 2 is a schematic flowchart of a computing power scheduling method based on a task model according to an embodiment of the present invention. The method includes: receiving a task submitted by a user, where the task includes a task type and a quantity of multi-dimensional resources applied for execution of the task (ie, first resource usage). Resource usage includes the number of CPU cores, GPU utilization percentage, average memory usage, and task load type. In some embodiments, the resource usage may also include the capacity requested to occupy network resources. According to the user tasks to be executed in the current scheduling round, the task queue of the scheduling system is updated, and the scheduler is periodically notified to perform scheduling. The period can be reasonably adjusted according to the load level; preferably, the period can be set to 5 minutes. The period of the scheduling round can be set according to the system operation requirements, and the scheduling queue maintains the tasks submitted by the current user for execution and their specific resource demand information through a list. The scheduling queue sorts the tasks currently waiting for scheduling according to the scheduling policy. The scheduling policy includes priority scheduling of tasks submitted first (first come, first served), shortest remaining time priority and main resource fairness, etc., and supports further modification and customization of tasks waiting to be scheduled. Sorting; among them, the shortest time remaining priority means that every time a new task arrives in the scheduling queue, the task that will end in the shortest time is given priority; the main resource fairness means that due to the application of resources in multiple dimensions, it is necessary to ensure that the resources in the task scheduling are balanced. Inter-task fairness, the resource requested by the task with the smallest proportion in the total resource pool is prioritized in the scheduling queue to achieve the effect of fairness in resource allocation between tasks. The scheduler requests the current task in the scheduling queue. The scheduling queue sets the first task from the initial waiting state to the scheduling state according to the task sorting result under the currently specified scheduling policy, and then executes the scheduling process of the current task.

For the resource application of the current task in the scheduling state, the adaptive resource scaling process based on the task model prediction is executed to adjust the resource application demand of the task. Figure 3 is a flow chart of multi-dimensional resource scaling and scheduling steps. First, determine the corresponding trained task model according to the task type of the current task, and predict the execution time of the current task under different multi-dimensional resource application amounts based on the task model. The construction input of the task model includes the usage of different dimensions of resources in historical log data and the corresponding task execution time. A regression model is obtained based on the machine learning method, which can predict the expected task execution time under the resource application amount. According to an embodiment of the present invention, one model is trained for each task type, so that the accuracy of the model obtained is higher. Combined with existing model feature importance analysis methods, the degree of impact of different dimensional resources in the current task model on the final task execution time is obtained. Combined with the original resource application results of the current task, the resource dimensions that have a small impact on the task execution time will be implemented to perform resource scaling adjustments under the set resource scaling rate to obtain the allocated resource usage. The resource scaling rate is one of the parameters of the present invention, which represents the percentage range of resource usage adjustment. By selecting appropriate parameters as the resource scaling rate, resource waste can be reduced while ensuring actual task execution efficiency. Since different tasks have different sensitivity to resources, the present invention adaptively searches for a resource scaling rate that meets the performance requirements of the current task. Among them, the resource scaling rate search method is to combine the task model and conduct a binary search under different resource scaling rates to ensure the task performance requirements while maximizing the saving of resources requested by the task. Specifically, the importance of multi-dimensional resource features is sorted according to the degree of impact of each dimensional resource feature in the task model on task execution time. Among them, the degree of influence is determined by randomly shuffling the characteristic value of each dimension resource, and calculating the drop value of the task model prediction accuracy at this time compared with the time when the task model is not randomly shuffled. The greater the drop value, the corresponding dimension resource pair. The greater the impact on task execution time, the higher its importance. The quotas of resources with low feature importance (little impact on task performance) are scaled and adjusted, that is, adaptive dynamic scaling is performed on them without affecting task performance or within a range acceptable to users. For example, the task resource performance model of the current task A points out that CPU and GPU are important features that affect task execution time, and memory has a small impact on task performance. In the resource scaling phase, the memory applied by the user is scaled. The scaling ratio is It is determined by combining the task model and performing a binary search under different resource scaling rates, which is the resource scaling rate that meets the current task performance requirements. Use the current task model to predict the task execution time corresponding to the resource demand after resource scaling adjustment again; by comparing the task execution time before and after resource scaling adjustment, determine whether the resource scaling adjustment process has significant impact on task performance that exceeds the performance adjustment threshold. If the impact is large, the scheduling method will try to adjust the resource scaling at other resource scaling rates; eventually, if the impact is within the threshold range, the scheduling method will receive the current resource application adjustment result (that is, the allocated resource usage As the second resource usage), it is applied in the scheduling process; if all set resource scaling rates will have a greater impact on the task execution time, this scheduling method will give up the resource scaling adjustment for the task and replace the original resource application results with Directly applied to the scheduling process. In the process of resource scaling of the memory, the task model of task A is used to perform the expected execution time of the task before and after scaling (the difference in execution time between the first expected execution time and the second expected execution time). If the conditions before and after scaling are satisfied If the execution time difference is less than the set difference threshold, memory scaling is performed. If not, the scaling size is adjusted until the conditions are met. If the current resource scaling affects task performance or the execution time difference is not within the acceptable range, the original resource application of the current task will be directly returned (that is, the resource applied for the task when the user submitted the task, that is, the first resource will be usage as the second resource usage), no resource scaling adjustments will be made to avoid affecting the service quality of the system.

After resource scaling is performed, the final resource request amount of the current scheduling status task is submitted to the scheduler and waits for the scheduler to allocate corresponding available resources. The scheduler allocates corresponding server resources based on the resource availability status of the servers in the current AI computing cluster, including the current usage of multi-dimensional resources and the load balancing situation between servers. If the current server status cannot satisfy the execution of the current task, the scheduling fails and the current task is resubmitted to the task queue to wait for the next round of scheduling. User tasks that are about to fail to be scheduled are automatically resubmitted until they can be scheduled.

Execute task deployment, schedule the task to the corresponding server for execution, update the resource usage status of the server currently executing the task, and the completion status of the current task, and wait for the next round of scheduling to perform a new round of online scheduling. Adaptive resource scaling scheduling method based on task prediction.

The computing power scheduling method of the embodiment of the present invention guides resource scheduling in AI computing power scenarios, comprehensively considers the scheduling and allocation of resources in multiple dimensions, and is used to improve the task deployment and execution efficiency of user tasks. At the user task level, by analyzing and modeling the impact of multi-dimensional resource factors on tasks, adaptive and dynamic resource scaling adjustments are realized, optimizing task deployment and completion, and improving task efficiency. Since the actual task execution efficiency is determined by the entirety of multi-dimensional resources, the present invention considers the resource demand characteristics of different types of tasks and the cluster utilization of multi-dimensional resources, and can solve the problem of multi-dimensional resources allocated at the task level. Mismatching problems can also improve the deployment efficiency of cluster-level tasks and the utilization of multi-dimensional resources.

Figure 4 is a hardware device equipped with the adaptive task resource scaling and scheduling system, including the functional implementation of the above scheduling steps and the corresponding architecture module implementation. The scheduling architecture module can be implemented through multiple hardware or processors required for corresponding steps, and the scheduling process is implemented through code execution of the scheduling system process stored in a computer-readable medium.

The device includes a processor of a central processing unit, an internal bus, a network interface, and a computer-readable storage medium. The processor and computer communicate with each other via the bus. The readable medium stores program codes for implementing all or part of the functions of the adaptive task resource expansion and contraction scheduling system in the present invention. When the program is executed by the processor, the adaptive task resource scaling and scheduling function can be implemented.

The present invention proposes an adaptive task resource expansion and contraction scheduling algorithm based on a machine learning task prediction method. This method combines the task performance model under multi-dimensional resources to optimize the resource over-application and mismatch between multi-dimensional resources when users submit tasks. Through machine learning methods, the present invention constructs the importance of the impact of different resource dimensions on the final performance of the task, and performs adaptive scaling optimization of the excess application part. This resource scheduling method optimizes the resource requirements of user tasks so that resources in all dimensions of the computing cluster can be efficiently used. In particular, it can make full use of scarce and expensive GPU resources, effectively improving the overall utilization efficiency of the cluster.

Each embodiment of this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments or implementation modes. The same or similar parts among the various embodiments of the present invention are mutually exclusive. Please refer to the above, the implementation principles surrounding the inventive concept and the technical effects produced can be referred to each other, and will not be described again here. Various embodiments or implementations in the present invention may be combined with each other without conflict.

It should be noted that although the steps are described in a specific order above, it does not mean that the steps must be performed in the above-mentioned specific order. In fact, some of these steps can be performed concurrently or even change the order, as long as it can be achieved Just the functions you need.

The invention may be a system, method and/or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of the invention.

Computer-readable storage media may be tangible devices that retain and store instructions for use by an instruction execution device. Computer-readable storage media may include, for example, but are not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or Flash memory), Static Random Access Memory (SRAM), Compact Disk Read Only Memory (CD-ROM), Digital Versatile Disk (DVD), Memory Stick, Floppy Disk, Mechanical Coding Device, such as a printer with instructions stored on it. Protruding structures in hole cards or grooves, and any suitable combination of the above.

The embodiments of the present invention have been described above. The above description is illustrative, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The selection of terms used in the embodiments of the present invention is intended to best explain the principles, practical applications or technical improvements in the market of each embodiment, or to enable other persons of ordinary skill in the art to understand the disclosure of the embodiments of the present invention. Various Examples.

Claims (11)

1. A method of training a task model for computational power scheduling, comprising:

taking task execution time, task GPU utilization rate, task average memory occupation, task CPU core number and task type as training samples; the task GPU utilization rate, the task average memory occupation, the task CPU core number and the task type are used as inputs of a task model, and the task execution time is a label and is used as an expected output value;

training the task model based on a regression tree with gradient lifting according to data in the training sample;

and taking the trained task model as a prediction model of task execution time in the computational power scheduling method.

2. The method of claim 1, wherein the task types in the same training sample are the same.

3. The method of claim 2, wherein the task types include computer vision tasks, natural language processing tasks, reinforcement learning tasks, graphic neural network tasks, recommendation tasks.

4. The method of claim 1, wherein the regression tree has a depth of 10, the number of base learners employed is 100, and the learning rate is 0.1.

5. A method of power dispatch, comprising:

acquiring a current task to be scheduled, wherein the task comprises a task type and a first resource usage amount applied for executing the task;

the task model obtained by any one of the methods according to claims 1 to 4, under the condition that the task performance is not affected or within an acceptable range, performing dynamic expansion adjustment on an application quota of a resource with small influence on the task performance in the first resource usage amount, so as to obtain a resource usage amount allocated for execution of the task;

and performing computational scheduling on the task according to the allocated resource usage.

6. The method according to claim 5, wherein the task model obtained according to any one of claims 1 to 4 dynamically scaling the resource application quota having a small effect on the task performance in the first resource usage without affecting the task performance or within an acceptable range, to obtain the resource usage allocated for the execution of the task, includes:

outputting a first expected execution time of the task according to the task type and the first resource usage amount by using the task model obtained according to any one of the methods of claims 1 to 4;

the resource application quota with small influence on the task performance in the resource dimension is subjected to telescopic adjustment under the condition that the task performance is not influenced or within an acceptable range, so that a second resource usage amount is obtained;

outputting a second expected execution time of the task according to the second resource usage amount by using the task model obtained by the method of any one of claims 1 to 4;

and when the difference between the first expected execution time and the second expected execution time is below a preset difference threshold, taking the second resource usage as the allocated resource usage.

7. The method as recited in claim 6, further comprising:

if the difference between the first expected execution time and the second expected execution time exceeds a set difference threshold, resource application quota with small influence on task performance in the resource dimension by using another resource expansion rate is utilized to carry out resource expansion adjustment.

8. The method as recited in claim 6, further comprising:

and when the difference between the execution time of the first expected execution time and the execution time of the second expected execution time exceeds the difference threshold after the resource expansion and contraction adjustment is carried out on the resource dimension by utilizing all the resource expansion and contraction rates, taking the first resource usage amount as the allocated resource usage amount.

9. The method of claim 6, wherein the performing, with the task model, resource scaling adjustments at a set resource scaling rate for the resource dimension to obtain a second resource usage amount comprises:

screening out the resource dimension with the influence degree on the execution time of the task smaller than a set influence degree threshold from the resource dimension, and taking the resource dimension as the resource dimension to be stretched;

and executing resource expansion adjustment under the set resource expansion rate aiming at the dimension of the resource to be expanded, so as to obtain a second resource usage amount.

10. A computing power scheduling system, comprising: a server, a task queue, a scheduler, further comprising a task history log database and a task model according to any one of claims 1 to 4;

the task history running log database is used for collecting and storing the history data of tasks, including resource use conditions, execution time and task types;

the task model is used for dynamically adjusting the application quota of the resource with small influence on the task performance in the task to be scheduled, so as to obtain the resource usage amount allocated for the execution of the task.

11. A storage medium having stored therein computer executable instructions which when loaded and executed by a processor perform the steps of the method according to any of claims 1 to 9.