High-parallelism rendering method based on AI algorithm capability

Abstract

The invention provides a high-parallelism rendering method based on AI (artificial intelligence) algorithm capability, which comprises the following steps: learning a user task and rendering software through AI, splitting the user task into a plurality of subtasks according to a splitting principle, and splitting the rendering software into a plurality of granularities according to the requirements of the user task; a learning model is input, the completion time and the resource demand of each subtask are predicted through AI, the minimum waiting time and the maximum resource utilization rate set value are output, and an efficient parallel task scheduling execution plan is generated; distributing corresponding multi-granularity software for each subtask according to the execution plan, and determining an execution sequence of the parallel rendering tasks according to the dependency relationship, priority and resource requirements of the subtasks; and collecting execution data as feedback information, inputting the feedback information into the learning model, and carrying out iterative optimization on the learning model. According to the method, the AI algorithm capability is adopted to improve the degree of parallelism of rendering software, and the execution efficiency and quality of cloud rendering tasks can be improved.

Description

Technical Field

[0001] The present invention relates to the field of cloud rendering technology, and in particular to a high-parallelism rendering method based on AI algorithm capabilities. Background Art

[0002] With the widespread application of artificial intelligence technology, cloud rendering is closely integrated with AI technology and has penetrated into all aspects of cloud rendering. For example, the Chinese patent with publication number CN115373762B discloses a method for dynamic loading of tool resources in real-time cloud rendering based on AI. The method includes taking tools as the core, constructing an open cloud rendering knowledge system through ontology data, and iteratively generating domain knowledge and application rules to realize the dynamic loading of tool resources. For another example, the Chinese patent application with publication number CN115460254A discloses a method for dynamic loading of environmental resources in real-time cloud rendering based on AI, including focusing on the environment to realize dynamic loading of resources. For another example, the Chinese patent application with publication number CN115373761A discloses a method for dynamic loading of scene resources in real-time cloud rendering based on AI, including automatically identifying rendering scenes and dynamically loading required scene resources. These patented technologies realize the intelligent convergence of resources to business in real-time cloud rendering through AI technology, and finally achieve automatic push of resources.

[0003] However, with the rapid development of AI technology, the complexity and computing power requirements of rendering tasks have increased significantly. This has brought a series of new challenges in dynamic resource loading, especially for rendering software resources. Different rendering software have differences in parallel processing capabilities, resulting in uneven performance during cloud execution. This difference not only affects the overall efficiency and quality of cloud rendering, but also puts higher requirements on resource scheduling. Therefore, how to effectively solve these execution efficiency and quality problems caused by the performance differences of different software has become one of the problems that need to be overcome in the current development of cloud rendering technology. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the prior art and provide a high-parallel rendering method based on AI algorithm capabilities. The AI ​​algorithm capabilities are used to improve the parallelism of the rendering software, which can improve the execution efficiency and quality of cloud rendering tasks.

[0005] In order to achieve the above objectives and other objectives, the present invention is implemented by including the following technical solutions: As a first aspect, the present invention proposes a high-parallelism rendering method based on AI algorithm capability, comprising the steps of:

[0006] S1. Learning user tasks and rendering software through AI, splitting the user tasks into multiple subtasks according to the splitting principle, and splitting the rendering software into multiple granularities according to the requirements of the user tasks;

[0007] S2. Input the learning model, use AI to predict the completion time and resource requirements of each subtask, output the minimum waiting time and maximum resource utilization setting value, and generate an execution plan for efficient parallel task scheduling;

[0008] S3, assigning corresponding multi-granularity software to each of the subtasks according to the execution plan, and determining the execution order of the parallel rendering tasks according to the dependency, priority and resource requirements of the subtasks;

[0009] S4. Collect execution data as feedback information and input it into the learning model to perform iterative optimization of the learning model.

[0010] In one embodiment, an operation guarantee mechanism is established in step S3, and key nodes are monitored by an AI algorithm.

[0011] In one embodiment, the operation guarantee mechanism includes a real-time supervision module, an abnormality monitoring module and a dynamic adjustment module. The real-time supervision module is used to continuously track the task execution status and resource usage; the abnormality monitoring module is used to monitor abnormal situations in real time through AI algorithms; the dynamic adjustment module is used to enable the system to automatically adjust the task execution plan or resource allocation when an abnormal situation is detected.

[0012] In one embodiment, "monitoring key nodes through AI algorithms" specifically involves using AI algorithm capabilities to conduct detailed analysis and evaluation of the interim results when the subtask is completed, while predicting the development trend of the task by learning historical data and real-time feedback information, and making fine adjustments when necessary.

[0013] In one embodiment, in the step S1, the splitting principles include the minimum coupling principle, the material non-single-share principle, and the subtask execution time being the same or multiple principle.

[0014] In one embodiment, in step S1, the granularity includes functional modules, third-party plug-ins, and extended functions.

[0015] In one embodiment, "inputting a learning model" in step S2 includes collecting historical data and extracting solution elements.

[0016] In one embodiment, the historical data includes subtask type and complexity, subtask execution time, resource usage, subtask status, subtask dependencies, and the number and type of concurrent subtasks; the solution elements include subtask estimated execution time, actual completion time, subtask waiting time, resource requirements, software resource utilization of different granularities, task priority and scheduling strategy.

[0017] As a second aspect, the present invention proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements any method described in the first aspect when loaded into the processor.

[0018] As a third aspect, the present invention proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any of the methods described in the first aspect.

[0019] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0020] 1. The present invention uses the powerful learning ability of AI to learn user tasks and rendering software, and splits user tasks according to specific splitting principles. At the same time, different software is granularized according to subtask requirements, which can weaken the differences in parallel processing capabilities of different rendering software, improve the level of software resource scheduling, and ultimately increase the number of concurrent tasks and software resource utilization;

[0021] 2. The present invention utilizes the powerful algorithmic capabilities of AI, and by inputting historical data and program elements in the learning phase and feedback information in the training phase for AI to continuously optimize and adjust model parameters, the execution efficiency of future tasks can be improved;

[0022] 3. The present invention utilizes the powerful algorithmic capabilities of AI to establish an operation guarantee mechanism to monitor key nodes, which can realize intelligent supervision and regulation of task and resource scheduling, ensure that tasks proceed smoothly as planned, and improve the quality of task completion. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Shown is a flow chart of a high-parallelism rendering method based on AI algorithm capability of the present invention.

[0024] Figure 2 Shown is an overall design schematic diagram of a high-parallel rendering method based on AI algorithm capabilities of the present invention.

[0025] Figure 3 Shown is a schematic diagram of the process of executing plan formation in a high-parallel rendering method based on AI algorithm capabilities of the present invention. DETAILED DESCRIPTION

[0026] The embodiments of the present invention provide a high-parallel rendering method based on AI algorithm capabilities, thereby solving the problem in the prior art that different rendering software have differences in parallel processing capabilities, resulting in uneven performance during cloud execution, affecting the overall efficiency and quality of cloud rendering.

[0027] Below, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. 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. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0028] Embodiment 1

[0029] Please see attached Figure 1-Figure 3 , an embodiment of the present invention provides a high-parallel rendering method based on AI algorithm capability, the method comprising the steps of:

[0030] S1. Perception: AI is used to learn user tasks and decompose user tasks into multiple independent subtasks according to certain splitting principles. At the same time, AI is used to learn rendering software and different rendering software is split into different granularities according to task requirements to increase the number of parallel software.

[0031] Among them, the splitting principles include the principle of low coupling, the principle of non-single-shared materials, and the principle of equal or multiple execution times of subtasks. The principle of low coupling refers to reducing the correlation between subtasks and reducing or avoiding the need to frequently call subtask results; the principle of non-single-shared materials refers to the materials used in subtasks being public resources to avoid locking and restricted use; the principle of equal or multiple execution times of subtasks refers to the execution times of different subtasks being the same or multiples to increase the number of parallel tasks and efficiency. The granularity includes functional modules, third-party plug-ins, and extended functions.

[0032] S2, Plan: First collect historical data, then extract plan elements as input to the learning model, and train the model to predict the completion time and resource requirements of each subtask, then output the minimum waiting time and maximum resource utilization setting value, and finally generate an execution plan for efficient parallel task scheduling;

[0033] The historical data includes subtask type and complexity, subtask execution time, resource usage, subtask status, subtask dependency, number and type of concurrent subtasks, etc. The solution elements include subtask estimated execution time, resource requirements, actual completion time, subtask waiting time, software resource utilization of different granularities, task priority and scheduling strategy, etc.

[0034] S3, execution: allocate corresponding software resources to each subtask according to the execution plan, and determine the execution order of parallel rendering tasks according to the dependencies, priorities and resource requirements of the subtasks; in addition, establish an operation guarantee mechanism, monitor key nodes through AI algorithms, so as to realize intelligent supervision and regulation of task and resource scheduling;

[0035] Among them, the operation guarantee mechanism includes a real-time supervision module, an abnormality monitoring module and a dynamic adjustment module. The real-time supervision module is used to continuously track the task execution status and resource usage; the abnormality monitoring module uses AI algorithms to monitor abnormal situations in real time, such as insufficient resources, task execution delays, etc.; the dynamic adjustment module is used to automatically adjust the task execution plan or resource allocation to solve the problem when an abnormality is detected. The key nodes are the interim results when the subtasks are completed. The AI ​​algorithm capabilities are used to conduct detailed analysis and evaluation of these nodes. At the same time, by learning historical data and real-time feedback, it can predict the development trend of the task and make fine adjustments when necessary to ensure that the task proceeds smoothly as planned.

[0036] S4, Iteration: The system collects execution data, including actual waiting time, resource utilization, etc., and inputs it into the learning model as feedback information to continuously optimize and adjust model parameters and improve the execution efficiency of future tasks.

[0037] Embodiment 2

[0038] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the method of the first embodiment when loaded into the processor.

[0039] Embodiment 3

[0040] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method of the first embodiment is implemented.

[0041] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the same technology of the present invention, the present invention is also intended to include these modifications and variations.

Claims

1. A high-parallel rendering method based on AI algorithm capabilities, characterized in that: Includes steps: S1. Learning user tasks and rendering software through AI, splitting the user tasks into multiple subtasks according to the splitting principle, and splitting the rendering software into multiple granularities according to the requirements of the user tasks; S2. Input the learning model, use AI to predict the completion time and resource requirements of each subtask, output the minimum waiting time and maximum resource utilization setting value, and generate an execution plan for efficient parallel task scheduling; S3, assigning corresponding multi-granularity software to each of the subtasks according to the execution plan, and determining the execution order of the parallel rendering tasks according to the dependency, priority and resource requirements of the subtasks; S4. Collect execution data as feedback information and input it into the learning model to perform iterative optimization of the learning model.

2. The method according to claim 1, characterized in that In step S3, an operation guarantee mechanism is established to monitor key nodes through AI algorithms.

3. The method according to claim 2, characterized in that The operation guarantee mechanism includes a real-time supervision module, an abnormality monitoring module and a dynamic adjustment module. The real-time supervision module is used to continuously track the task execution status and resource usage; the abnormality monitoring module is used to monitor abnormal situations in real time through AI algorithms; and the dynamic adjustment module is used to enable the system to automatically adjust the task execution plan or resource allocation when an abnormal situation is detected.

4. The method according to claim 2, characterized in that: "Monitoring key nodes through AI algorithms" specifically means using AI algorithm capabilities to conduct detailed analysis and evaluation of the interim results when the subtasks are completed. At the same time, by learning historical data and real-time feedback information, the development trend of the task can be predicted, and fine-tuning can be performed when necessary.

5. The method according to claim 1, characterized in that In the step S1, the splitting principles include the principle of low coupling, the principle of non-single-shared materials, and the principle of equal or multiple execution times of subtasks.

6. The method according to claim 1, characterized in that In the step S1, the granularity includes functional modules, third-party plug-ins and extended functions.

7. The method according to claim 1, characterized in that In step S2, "inputting a learning model" includes collecting historical data and extracting solution elements.

8. The method according to claim 7, characterized in that The historical data includes subtask type and complexity, subtask execution time, resource usage, subtask status, subtask dependencies, and the number and type of concurrent subtasks; the solution elements include subtask estimated execution time, actual completion time, subtask waiting time, resource requirements, software resource utilization of different granularities, task priority and scheduling strategy.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is loaded into the processor, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images