Intelligent scheduling method and system based on distributed real-time rendering and medium

Abstract

The invention provides an intelligent scheduling method and system based on distributed real-time rendering and a medium. The method comprises the steps of obtaining real-time rendering task information, obtaining category rendering tasks after processing, extracting category rendering parameters, obtaining category rendering performance demand indexes through processing according to the category rendering parameters, obtaining performance parameters of all computing nodes and real-time performance reduction data, and obtaining processing performance indexes of all the computing nodes through processing. Sorting and matching the category rendering performance demand indexes and the processing performance indexes to obtain corresponding adaptive processing nodes; therefore, intelligent scheduling of rendering task classification and distributed rendering is realized through classification of real-time rendering tasks, extraction of category rendering parameters, and calculation and matching of category rendering performance demand indexes and processing performance indexes.

Description

Technical Field

[0001] The present application relates to the field of rendering processing, and in particular, to an intelligent scheduling method, system and medium based on distributed real-time rendering. Background Art

[0002] With the widespread application of computer graphics technology in many fields such as virtual reality (VR), augmented reality (AR), film and television production, game development, simulation, etc., the demand for real-time rendering of high-quality and high-complexity scenes has increased dramatically. The traditional single-machine rendering method is limited by the hardware performance of a single computing device. It is difficult to complete high-quality rendering tasks of large-scale scenes in a limited time and cannot meet real-time requirements.

[0003] Distributed real-time rendering technology greatly improves the efficiency and speed of rendering by decomposing and assigning rendering tasks to multiple computing nodes for parallel execution. However, existing distributed real-time rendering systems have many deficiencies in task scheduling. For example, unreasonable task allocation causes some nodes to be overloaded while others are idle, affecting overall performance; lack of effective management of rendering task priorities makes it impossible to respond to user interactions in a timely manner, etc. Therefore, there is an urgent need for a more intelligent and efficient scheduling method to solve these problems in order to improve the performance and user experience of distributed real-time rendering systems.

[0004] In view of the above problems, effective technical solutions are urgently needed. Summary of the invention

[0005] The purpose of this application is to provide an intelligent scheduling method, system and medium based on distributed real-time rendering, which realizes rendering task classification and intelligent scheduling of distributed rendering through the classification of real-time rendering tasks, extraction of category rendering parameters, calculation and matching of category rendering performance requirement index and processing performance index.

[0006] The present application also provides an intelligent scheduling method based on distributed real-time rendering, comprising the following steps: Acquire real-time rendering task information, process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extract corresponding category rendering parameters according to the category rendering tasks; Processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; Acquire performance parameters and real-time performance degradation data of each computing node, and obtain a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; The category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and corresponding adaptation processing nodes are obtained.

[0007] Optionally, in the intelligent scheduling method based on distributed real-time rendering described in the present application, the real-time rendering task information is obtained, the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and corresponding category rendering parameters are extracted according to the category rendering task, specifically including: Acquire real-time rendering task information, and process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; Extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; Extract lighting calculation rendering parameters according to lighting calculation, including the number of light sources and comprehensive characteristic data of light sources; Extracting shadow generation rendering parameters according to shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; Extract texture map rendering parameters based on the texture map, including texture resolution data and the number of texture types.

[0008] Optionally, in the intelligent scheduling method based on distributed real-time rendering described in the present application, the processing of the category rendering parameters to obtain the category rendering performance requirement index specifically includes: Obtaining an object model rendering performance requirement index according to the model scale data and geometric complexity data processing; Obtaining a lighting rendering performance requirement index according to the number of light sources and the comprehensive characteristic data of the light sources; Obtaining a shadow rendering performance requirement index according to the shadow precision requirement data and the occlusion relationship complexity data; The texture mapping rendering performance requirement index is obtained according to the texture resolution data and the number of texture types.

[0009] Optionally, in the intelligent scheduling method based on distributed real-time rendering described in the present application, the obtaining of performance parameters and real-time performance degradation data of each computing node, and obtaining a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data, specifically includes: Obtain the performance parameters of each computing node, including CPU performance data, GPU performance data, and storage resource data; The CPU performance data, GPU performance data, storage resource data and real-time performance degradation data are input into a preset computing node performance evaluation model for processing to obtain a processing performance index.

[0010] Optionally, in the intelligent scheduling method based on distributed real-time rendering described in the present application, the category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and the corresponding adaptation processing node is obtained, specifically including: Sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index, and the texture mapping rendering performance requirement index in descending order to obtain a sequential rendering performance requirement index; Sorting the processing performance indexes of the computing nodes in descending order to obtain a sequential processing performance index; The sequential rendering performance requirement index and the sequential processing performance index are matched one by one in sequence to obtain corresponding adaptation processing nodes.

[0011] Optionally, the intelligent scheduling method based on distributed real-time rendering described in the present application further includes: Obtain the urgency and importance data of each sub-rendering task of the computing node; Inputting the urgency data and the importance data into a preset rendering processing priority evaluation model for processing to obtain a rendering processing priority index; Sort the sub-rendering tasks in descending order according to the rendering processing priority index to obtain sequential sub-rendering tasks; The computing node performs rendering processing according to the sequential sub-rendering tasks.

[0012] Optionally, the intelligent scheduling method based on distributed real-time rendering described in the present application further includes: Get the processing progress data of the sub-rendering tasks in each computing node; Obtain average processing progress data of the rendering task according to the processing progress data; Dividing the processing progress data by the average processing progress data to obtain relative task processing progress data; Compare the relative task processing progress data with a preset task processing progress threshold to obtain a task processing progress status; Extracting a first threshold and a second threshold according to the task processing progress threshold, wherein the first threshold is greater than the second threshold; comparing the relative task processing progress data with the first threshold and the second threshold; If the relative task processing progress data is greater than the first threshold, the task processing progress state is an efficient operation state; If the relative task processing progress data is greater than the second threshold and less than or equal to the first threshold, the task processing progress state is a stable operation state; If the relative task processing progress data is less than or equal to the second threshold, the task processing progress status is an abnormal operation status.

[0013] In a second aspect, the present application provides an intelligent scheduling system based on distributed real-time rendering, the system comprising: a memory and a processor, wherein the memory stores a program of an intelligent scheduling method based on distributed real-time rendering, and when the program of the intelligent scheduling method based on distributed real-time rendering is executed by the processor, the following steps are implemented: Acquire real-time rendering task information, process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extract corresponding category rendering parameters according to the category rendering tasks; Processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; Acquire performance parameters and real-time performance degradation data of each computing node, and obtain a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; The category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and corresponding adaptation processing nodes are obtained.

[0014] Optionally, in the intelligent scheduling system based on distributed real-time rendering described in the present application, the real-time rendering task information is obtained, and the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and corresponding category rendering parameters are extracted according to the category rendering task, specifically including: Acquire real-time rendering task information, and process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; Extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; Extract lighting calculation rendering parameters according to lighting calculation, including the number of light sources and comprehensive characteristic data of light sources; Extracting shadow generation rendering parameters according to shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; Extract texture map rendering parameters based on the texture map, including texture resolution data and the number of texture types.

[0015] In the third aspect, the present application also provides a readable storage medium, which stores an intelligent scheduling method program based on distributed real-time rendering. When the intelligent scheduling method program based on distributed real-time rendering is executed by a processor, the steps of an intelligent scheduling method based on distributed real-time rendering as described in any one of the above items are implemented.

[0016] As can be seen from the above, the intelligent scheduling method, system and medium based on distributed real-time rendering provided by the present application. The method obtains real-time rendering task information, obtains category rendering tasks after processing, and extracts category rendering parameters, processes and obtains category rendering performance requirement indexes according to the category rendering parameters, obtains the performance parameters and real-time performance degradation data of each computing node and processes and obtains the processing performance index of each computing node, and sorts and matches the category rendering performance requirement index and the processing performance index to obtain the corresponding adaptation processing node; thereby, through the classification of real-time rendering tasks, the extraction of category rendering parameters, the calculation and matching of the category rendering performance requirement index and the processing performance index, the intelligent scheduling of rendering task classification and distributed rendering is realized.

[0017] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures particularly pointed out in the written description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 A flowchart of an intelligent scheduling method based on distributed real-time rendering provided in an embodiment of the present application; Figure 2 A flowchart of obtaining category rendering parameters of an intelligent scheduling method based on distributed real-time rendering provided in an embodiment of the present application; Figure 3 A flowchart of obtaining a category rendering performance requirement index in an intelligent scheduling method based on distributed real-time rendering provided in an embodiment of the present application; Figure 4 A flowchart of obtaining a processing performance index for an intelligent scheduling method based on distributed real-time rendering provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0021] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0022] Please refer to Figure 1 , Figure 1 1 is a flow chart of an intelligent scheduling method based on distributed real-time rendering in some embodiments of the present application. The intelligent scheduling method based on distributed real-time rendering is used in a terminal device, such as a computer, a mobile phone, etc. The intelligent scheduling method based on distributed real-time rendering includes the following steps: S11, obtaining real-time rendering task information, processing the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extracting corresponding category rendering parameters according to the category rendering tasks; S12, processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; S13, obtaining performance parameters and real-time performance degradation data of each computing node, and obtaining a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; S14, sorting the category rendering performance requirement index and the processing performance index respectively, matching them according to preset rules, and obtaining corresponding adaptation processing nodes.

[0023] It should be noted that when rendering a scene, in order to improve rendering efficiency, it is necessary to decompose the real-time rendering task through a preset task classification model to obtain a category rendering task. The task classification model is a model obtained by training a large amount of historical data and can classify rendering objects according to user settings and needs. After the rendering objects are classified, the category rendering tasks, i.e., subtasks, are obtained; according to the category rendering tasks, the corresponding category rendering parameters can be extracted. The category rendering parameters refer to parameters that can reflect the difficulty of different category rendering tasks. Different category rendering tasks and different rendering difficulties have different operating performance requirements. According to the category rendering parameters, the category rendering performance requirement index can be calculated, i.e., the performance required for the rendering task of this category. During the operation of the computing node, the operating performance often decreases with the increase of the running time. Therefore, the performance parameters and real-time performance degradation data of each computing node are obtained, and then the processing performance index of each computing node is calculated. The category rendering performance requirement index is first sorted to intuitively understand the performance requirements of each subtask, and then the processing performance index is sorted to intuitively understand the processing performance of each computing node, so as to facilitate matching processing.

[0024] Please refer to Figure 2 , Figure 2 This is a flowchart of obtaining category rendering parameters of an intelligent scheduling method based on distributed real-time rendering provided by an embodiment of the present application. According to an embodiment of the present invention, the real-time rendering task information is obtained, and the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and the corresponding category rendering parameters are extracted according to the category rendering task, specifically including: S21, obtaining real-time rendering task information, and processing the real-time rendering task information through a preset task classification model to obtain a category rendering task, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; S22, extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; S23, extracting lighting calculation rendering parameters according to the lighting calculation, including the number of light sources and comprehensive characteristic data of the light sources; S24, extracting shadow generation rendering parameters according to the shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; S25. Extracting texture mapping rendering parameters according to the texture mapping, including texture resolution data and the number of texture types.

[0025] It should be noted that object model rendering tasks, lighting calculations, shadow generation, and texture mapping are four aspects of rendering. The overall rendering is achieved after rendering these four aspects separately. The model scale data is a comprehensive reflection of the number of model vertices, the number of cotton pieces, etc. The larger the model scale, the more complex the processing difficulty and the higher the performance requirements. The geometric complexity data refers to the complexity of the model's geometric shape; the number of light sources refers to the number of light sources, and the comprehensive characteristic data of the light source refers to data that reflects the degree of the comprehensive characteristics of the light source. The comprehensive characteristics comprehensively consider the intensity, color and range. The number of light sources and the comprehensive characteristics of the light source will affect the complexity of the lighting calculation; high-resolution textures will take up a lot of memory. Texture types include different types of textures such as diffuse textures, normal textures, and height textures. The more types of textures, the more complex the processing.

[0026] Please refer to Figure 3 , Figure 3 1 is a flow chart of obtaining a category rendering performance requirement index of an intelligent scheduling method based on distributed real-time rendering provided by an embodiment of the present application. According to an embodiment of the present invention, the category rendering parameters are processed separately to obtain the category rendering performance requirement index, specifically including: S31, obtaining an object model rendering performance requirement index according to the model scale data and geometric complexity data; S32, obtaining a lighting rendering performance requirement index according to the number of light sources and the comprehensive characteristic data of the light sources; S33, obtaining a shadow rendering performance requirement index according to the shadow precision requirement data and the occlusion relationship complexity data; S34. Obtain a texture mapping rendering performance requirement index according to the texture resolution data and the number of texture types.

[0027] It should be noted that when processing the category rendering tasks separately, it is necessary to consider their requirements for computing node performance separately. After calculation, it is convenient to select the appropriate computing node; the model scale data and geometric complexity data are input into the preset object model performance requirement evaluation model for processing to obtain the object model rendering performance requirement index. The object model performance requirement evaluation model is obtained through training with a large amount of historical data. The object model rendering performance requirement index refers to the requirements for processor performance when rendering the object model; the calculation formula of the object model rendering performance requirement index in the object model performance evaluation model is: ; in, The performance requirement index for object model rendering, , are model scale data and geometric complexity data, respectively. , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The lighting rendering performance requirement index is obtained by processing a preset lighting rendering performance requirement evaluation model. The lighting rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The lighting rendering performance requirement index refers to the requirements for processor performance when performing lighting rendering. The calculation formula of the lighting rendering performance requirement index in the lighting rendering performance evaluation model is: ; in, is the lighting rendering performance requirement index, , They are the number of light sources and the comprehensive characteristic data of light sources, , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The shadow rendering performance requirement index is obtained by processing a preset shadow rendering performance requirement evaluation model. The shadow rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The shadow rendering performance requirement index refers to the requirements for processor performance when performing shadow rendering. The calculation formula of the shadow rendering performance requirement index in the shadow rendering performance requirement evaluation model is: ; in, is the shadow rendering performance requirement index, , They are shadow accuracy requirement data and occlusion relationship complexity data, , , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The texture mapping rendering performance requirement index is obtained by processing a preset texture mapping rendering performance requirement evaluation model. The texture mapping rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The texture mapping rendering performance requirement index refers to the requirements for processor performance when performing texture mapping rendering. The calculation formula of the texture mapping rendering performance requirement index in the texture mapping rendering performance requirement evaluation model is: ; in, The performance requirement index for texture mapping rendering, , They are texture resolution data and number of texture types respectively. , They are respectively preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering).

[0028] Please refer to Figure 4 , Figure 4 The flowchart of obtaining a processing performance index of an intelligent scheduling method based on distributed real-time rendering provided by an embodiment of the present application. According to an embodiment of the present invention, the obtaining of performance parameters and real-time performance degradation data of each computing node, and obtaining the processing performance index of each computing node according to the performance parameters and real-time performance degradation data processing specifically includes: S41, obtaining performance parameters of each computing node, including CPU performance data, GPU performance data, and storage resource data; S42: Input the CPU performance data, GPU performance data, storage resource data and real-time performance degradation data into a preset computing node performance evaluation model for processing to obtain a processing performance index.

[0029] It should be noted that in the system for processing rendering tasks, there are multiple computing nodes. A computing node refers to a processor in the system that can process rendering tasks. The preset computing node performance evaluation model is obtained through historical data training. The calculation formula for the processing performance index in the computing node performance evaluation model is: ; in, To process the performance index, , , They are CPU performance data, GPU performance data and storage resource data. For real-time performance degradation data, , , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); CPU performance data and GPU performance data refer to the data reflecting CPU performance and GPU performance respectively obtained through the preset performance evaluation system; the processing performance index reflects the ability of the computing node to process rendering tasks, and the real-time performance degradation data refers to the performance degradation of the computing node due to long-term operation, reduced memory, etc.

[0030] According to an embodiment of the present invention, sorting the category rendering performance requirement index and the processing performance index respectively, matching them according to preset rules, and obtaining corresponding adaptation processing nodes specifically includes: Sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index, and the texture mapping rendering performance requirement index in descending order to obtain a sequential rendering performance requirement index; Sorting the processing performance indexes of the computing nodes in descending order to obtain a sequential processing performance index; The sequential rendering performance requirement index and the sequential processing performance index are matched one by one in sequence to obtain corresponding adaptation processing nodes.

[0031] It should be noted that after sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index and the texture mapping rendering performance requirement index respectively, the status and size order of the performance requirements can be quickly seen, and then matched with the sequential processing performance index corresponding to the computing node. The matching principle is that the performance requirement index with a large processing performance index is matched. In this embodiment, it is matched in order, that is, the first sequential rendering performance requirement index is matched with the first sequential processing performance index, and so on.

[0032] According to an embodiment of the present invention, it also includes: Obtain the urgency and importance data of each sub-rendering task of the computing node; Inputting the urgency data and the importance data into a preset rendering processing priority evaluation model for processing to obtain a rendering processing priority index; Sort the sub-rendering tasks in descending order according to the rendering processing priority index to obtain sequential sub-rendering tasks; The computing node performs rendering processing according to the sequential sub-rendering tasks.

[0033] It should be noted that sub-rendering tasks are relative to the overall rendering task. A full-day rendering task can be divided into several sub-rendering tasks. In a computing node, different sub-rendering tasks may exist at the same time, but the urgency and importance of each rendering task are different. Therefore, in order to better protect urgent and important tasks, it is necessary to evaluate the priority. The rendering processing priority evaluation model obtained by training based on historical data is used to obtain the rendering processing priority index. The calculation formula is: ; in, The rendering process priority index, , They are urgency data and importance data, , They are respectively preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering).

[0034] According to an embodiment of the present invention, it also includes: Get the processing progress data of the sub-rendering tasks in each computing node; Obtain average processing progress data of the rendering task according to the processing progress data; Dividing the processing progress data by the average processing progress data to obtain relative task processing progress data; Compare the relative task processing progress data with a preset task processing progress threshold to obtain a task processing progress status; Extracting a first threshold and a second threshold according to the task processing progress threshold, wherein the first threshold is greater than the second threshold; comparing the relative task processing progress data with the first threshold and the second threshold; If the relative task processing progress data is greater than the first threshold, the task processing progress state is an efficient operation state; If the relative task processing progress data is greater than the second threshold and less than or equal to the first threshold, the task processing progress state is a stable operation state; If the relative task processing progress data is less than or equal to the second threshold, the task processing progress status is an abnormal operation status.

[0035] It should be noted that during the operation of the computing node, there may be operation failures or equipment failures. Therefore, it is necessary to dynamically monitor the operation status to facilitate timely adjustments. The subtasks of an overall task should try to keep the progress consistent, so that the overall task can be better completed. The average processing progress can better reflect the stable progress of the subtasks. In this embodiment, the task processing progress threshold is set to: [0, 0.8], the task processing progress state is an abnormal operation state; (0.8, 1], the task processing progress state is a stable operation state; greater than 1; the task processing progress state is an efficient operation state; when the relative task processing progress data is 0.85, the task processing progress state is a stable operation state.

[0036] It is worth mentioning that it also includes: Get the problematic sub-rendering task corresponding to the abnormal running status, and process the progress data Sort the corresponding problematic sub-rendering tasks in descending order to obtain a problematic sub-rendering task queue; Obtain the utilization rate of each computing node, sort the utilization rates of each computing node in ascending order, and obtain a computing node utilization rate queue; Task allocation is performed after matching the problematic sub-rendering task queue with the computing node usage queue.

[0037] It should be noted that after the problematic sub-rendering task is detected, its computing node needs to be adjusted in time. The utilization rate refers to the ratio of the current utilization performance of the computing node to the total operating performance. The matching method between the problematic sub-rendering task queue and the computing node utilization queue is that the first one in the problematic sub-rendering task queue is matched with the first one in the computing node utilization queue, and so on.

[0038] The present invention also discloses an intelligent scheduling system based on distributed real-time rendering, comprising a memory and a processor, wherein the memory stores an intelligent scheduling method program based on distributed real-time rendering, and when the intelligent scheduling method program based on distributed real-time rendering is executed by the processor, the following steps are implemented: Acquire real-time rendering task information, process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extract corresponding category rendering parameters according to the category rendering tasks; Processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; Acquire performance parameters and real-time performance degradation data of each computing node, and obtain a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; The category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and corresponding adaptation processing nodes are obtained.

[0039] It should be noted that when rendering a scene, in order to improve rendering efficiency, it is necessary to decompose the real-time rendering task through a preset task classification model to obtain a category rendering task. The task classification model is a model obtained by training a large amount of historical data and can classify rendering objects according to user settings and needs. After the rendering objects are classified, the category rendering tasks, i.e., subtasks, are obtained; according to the category rendering tasks, the corresponding category rendering parameters can be extracted. The category rendering parameters refer to parameters that can reflect the difficulty of different category rendering tasks. Different category rendering tasks and different rendering difficulties have different operating performance requirements. According to the category rendering parameters, the category rendering performance requirement index can be calculated, i.e., the performance required for the rendering task of this category. During the operation of the computing node, the operating performance often decreases with the increase of the running time. Therefore, the performance parameters and real-time performance degradation data of each computing node are obtained, and then the processing performance index of each computing node is calculated. The category rendering performance requirement index is first sorted to intuitively understand the performance requirements of each subtask, and then the processing performance index is sorted to intuitively understand the processing performance of each computing node, so as to facilitate matching processing.

[0040] According to an embodiment of the present invention, the real-time rendering task information is obtained, and the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and corresponding category rendering parameters are extracted according to the category rendering task, specifically including: Acquire real-time rendering task information, and process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; Extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; Extract lighting calculation rendering parameters according to lighting calculation, including the number of light sources and comprehensive characteristic data of light sources; Extracting shadow generation rendering parameters according to shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; Extract texture map rendering parameters based on the texture map, including texture resolution data and the number of texture types.

[0041] It should be noted that object model rendering tasks, lighting calculations, shadow generation, and texture mapping are four aspects of rendering. The overall rendering is achieved after rendering these four aspects separately. The model scale data is a comprehensive reflection of the number of model vertices, the number of cotton pieces, etc. The larger the model scale, the more complex the processing difficulty and the higher the performance requirements. The geometric complexity data refers to the complexity of the model's geometric shape; the number of light sources refers to the number of light sources, and the comprehensive characteristic data of the light source refers to data that reflects the degree of the comprehensive characteristics of the light source. The comprehensive characteristics comprehensively consider the intensity, color and range. The number of light sources and the comprehensive characteristics of the light source will affect the complexity of the lighting calculation; high-resolution textures will take up a lot of memory. Texture types include different types of textures such as diffuse textures, normal textures, and height textures. The more types of textures, the more complex the processing.

[0042] According to an embodiment of the present invention, the step of processing the category rendering parameters separately to obtain the category rendering performance requirement index specifically includes: Obtaining an object model rendering performance requirement index according to the model scale data and geometric complexity data processing; Obtaining a lighting rendering performance requirement index according to the number of light sources and the comprehensive characteristic data of the light sources; Obtaining a shadow rendering performance requirement index according to the shadow precision requirement data and the occlusion relationship complexity data; The texture mapping rendering performance requirement index is obtained according to the texture resolution data and the number of texture types.

[0043] It should be noted that when processing the category rendering tasks separately, it is necessary to consider their requirements for computing node performance separately. After calculation, it is convenient to select the appropriate computing node; the model scale data and geometric complexity data are input into the preset object model performance requirement evaluation model for processing to obtain the object model rendering performance requirement index. The object model performance requirement evaluation model is obtained through training with a large amount of historical data. The object model rendering performance requirement index refers to the requirements for processor performance when rendering the object model; the calculation formula of the object model rendering performance requirement index in the object model performance evaluation model is: ; in, The performance requirement index for object model rendering, , are model scale data and geometric complexity data, respectively. , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The lighting rendering performance requirement index is obtained by processing a preset lighting rendering performance requirement evaluation model. The lighting rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The lighting rendering performance requirement index refers to the requirements for processor performance when performing lighting rendering. The calculation formula of the lighting rendering performance requirement index in the lighting rendering performance evaluation model is: ; in, is the lighting rendering performance requirement index, , They are the number of light sources and the comprehensive characteristic data of light sources, , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The shadow rendering performance requirement index is obtained by processing a preset shadow rendering performance requirement evaluation model. The shadow rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The shadow rendering performance requirement index refers to the requirements for processor performance when performing shadow rendering. The calculation formula of the shadow rendering performance requirement index in the shadow rendering performance requirement evaluation model is: ; in, is the shadow rendering performance requirement index, , They are shadow accuracy requirement data and occlusion relationship complexity data, , , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); The texture mapping rendering performance requirement index is obtained by processing a preset texture mapping rendering performance requirement evaluation model. The texture mapping rendering performance requirement evaluation model is a model obtained by processing a large amount of historical data. The texture mapping rendering performance requirement index refers to the requirements for processor performance when performing texture mapping rendering. The calculation formula of the texture mapping rendering performance requirement index in the texture mapping rendering performance requirement evaluation model is: ; in, The performance requirement index for texture mapping rendering, , They are texture resolution data and number of texture types respectively. , They are respectively preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering).

[0044] According to an embodiment of the present invention, the step of obtaining the performance parameters and real-time performance degradation data of each computing node, and obtaining the processing performance index of each computing node according to the performance parameters and the real-time performance degradation data, specifically includes: Obtain the performance parameters of each computing node, including CPU performance data, GPU performance data, and storage resource data; The CPU performance data, GPU performance data, storage resource data and real-time performance degradation data are input into a preset computing node performance evaluation model for processing to obtain a processing performance index.

[0045] It should be noted that in the system for processing rendering tasks, there are multiple computing nodes. A computing node refers to a processor in the system that can process rendering tasks. The preset computing node performance evaluation model is obtained through historical data training. The calculation formula for the processing performance index in the computing node performance evaluation model is: ; in, To process the performance index, , , They are CPU performance data, GPU performance data and storage resource data. For real-time performance degradation data, , , They are preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering); CPU performance data and GPU performance data refer to the data reflecting CPU performance and GPU performance respectively obtained through the preset performance evaluation system; the processing performance index reflects the ability of the computing node to process rendering tasks, and the real-time performance degradation data refers to the performance degradation of the computing node due to long-term operation, reduced memory, etc.

[0046] According to an embodiment of the present invention, sorting the category rendering performance requirement index and the processing performance index respectively, matching them according to preset rules, and obtaining corresponding adaptation processing nodes specifically includes: Sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index, and the texture mapping rendering performance requirement index in descending order to obtain a sequential rendering performance requirement index; Sorting the processing performance indexes of the computing nodes in descending order to obtain a sequential processing performance index; The sequential rendering performance requirement index and the sequential processing performance index are matched one by one in sequence to obtain corresponding adaptation processing nodes.

[0047] It should be noted that after sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index and the texture mapping rendering performance requirement index respectively, the status and size order of the performance requirements can be quickly seen, and then matched with the sequential processing performance index corresponding to the computing node. The matching principle is that the performance requirement index with a large processing performance index is matched. In this embodiment, it is matched in order, that is, the first sequential rendering performance requirement index is matched with the first sequential processing performance index, and so on.

[0048] According to an embodiment of the present invention, it also includes: Obtain the urgency and importance data of each sub-rendering task of the computing node; Inputting the urgency data and the importance data into a preset rendering processing priority evaluation model for processing to obtain a rendering processing priority index; Sort the sub-rendering tasks in descending order according to the rendering processing priority index to obtain sequential sub-rendering tasks; The computing node performs rendering processing according to the sequential sub-rendering tasks.

[0049] It should be noted that sub-rendering tasks are relative to the overall rendering task. A full-day rendering task can be divided into several sub-rendering tasks. In a computing node, different sub-rendering tasks may exist at the same time, but the urgency and importance of each rendering task are different. Therefore, in order to better protect urgent and important tasks, it is necessary to evaluate the priority. The rendering processing priority evaluation model obtained by training based on historical data is used to obtain the rendering processing priority index. The calculation formula is: ; in, The rendering process priority index. , They are urgency data and importance data, , They are respectively preset characteristic coefficients (the characteristic coefficients are obtained by querying the preset intelligent scheduling platform based on distributed real-time rendering).

[0050] According to an embodiment of the present invention, it also includes: Obtain the processing progress data of the sub-rendering tasks in each computing node; Obtain average processing progress data of the rendering task according to the processing progress data; Dividing the processing progress data by the average processing progress data to obtain relative task processing progress data; Compare the relative task processing progress data with a preset task processing progress threshold to obtain a task processing progress status; Extracting a first threshold and a second threshold according to the task processing progress threshold, wherein the first threshold is greater than the second threshold; comparing the relative task processing progress data with the first threshold and the second threshold; If the relative task processing progress data is greater than the first threshold, the task processing progress state is an efficient operation state; If the relative task processing progress data is greater than the second threshold and less than or equal to the first threshold, the task processing progress state is a stable operation state; If the relative task processing progress data is less than or equal to the second threshold, the task processing progress status is an abnormal operation status.

[0051] It should be noted that during the operation of the computing node, there may be operation failures or equipment failures. Therefore, it is necessary to dynamically monitor the operation status to facilitate timely adjustments. The subtasks of an overall task should try to keep the progress consistent, so that the overall task can be better completed. The average processing progress can better reflect the stable progress of the subtasks. In this embodiment, the task processing progress threshold is set to: [0, 0.8], the task processing progress state is an abnormal operation state; (0.8, 1], the task processing progress state is a stable operation state; greater than 1; the task processing progress state is an efficient operation state; when the relative task processing progress data is 0.85, the task processing progress state is a stable operation state.

[0052] It is worth mentioning that it also includes: Get the problematic sub-rendering task corresponding to the abnormal running status, and process the progress data Sort the corresponding problematic sub-rendering tasks in descending order to obtain a problematic sub-rendering task queue; Obtain the utilization rate of each computing node, sort the utilization rates of each computing node in ascending order, and obtain a computing node utilization rate queue; Task allocation is performed after matching the problematic sub-rendering task queue with the computing node usage queue.

[0053] It should be noted that after the problematic sub-rendering task is detected, its computing node needs to be adjusted in time. The utilization rate refers to the ratio of the current utilization performance of the computing node to the total operating performance. The matching method between the problematic sub-rendering task queue and the computing node utilization queue is that the first one in the problematic sub-rendering task queue is matched with the first one in the computing node utilization queue, and so on.

[0054] A third aspect of the present invention provides a readable storage medium, which includes a program for an intelligent scheduling method based on distributed real-time rendering. When the program for an intelligent scheduling method based on distributed real-time rendering is executed by a processor, the steps of an intelligent scheduling method based on distributed real-time rendering as described in any one of the above items are implemented.

[0055] The present invention discloses an intelligent scheduling method, system and medium based on distributed real-time rendering, which obtains real-time rendering task information, processes it to obtain category rendering tasks, extracts category rendering parameters, processes them respectively according to the category rendering parameters to obtain category rendering performance requirement indexes, obtains performance parameters and real-time performance degradation data of each computing node and processes them to obtain processing performance indexes of each computing node, sorts and matches the category rendering performance requirement indexes and processing performance indexes respectively to obtain corresponding adaptation processing nodes; thereby, through the classification of real-time rendering tasks, the extraction of category rendering parameters, the calculation and matching of category rendering performance requirement indexes and processing performance indexes, the classification of rendering tasks and the intelligent scheduling of distributed rendering are realized.

[0056] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0057] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0058] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0059] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, the aforementioned program can be stored in a readable storage medium, and when the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0060] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

Claims

1. An intelligent scheduling method based on distributed real-time rendering, characterized in that: include: Acquire real-time rendering task information, process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extract corresponding category rendering parameters according to the category rendering tasks; Processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; Acquire performance parameters and real-time performance degradation data of each computing node, and obtain a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; The category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and corresponding adaptation processing nodes are obtained.

2. The intelligent scheduling method based on distributed real-time rendering according to claim 1 is characterized in that: The real-time rendering task information is obtained, and the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and corresponding category rendering parameters are extracted according to the category rendering task, specifically including: Acquire real-time rendering task information, and process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; Extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; Extract lighting calculation rendering parameters according to lighting calculation, including the number of light sources and comprehensive characteristic data of light sources; Extracting shadow generation rendering parameters according to shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; Extract texture map rendering parameters based on the texture map, including texture resolution data and the number of texture types.

3. The intelligent scheduling method based on distributed real-time rendering according to claim 2, characterized in that: The step of processing the category rendering parameters separately to obtain the category rendering performance requirement index specifically includes: Obtaining an object model rendering performance requirement index according to the model scale data and geometric complexity data processing; Obtaining a lighting rendering performance requirement index according to the number of light sources and the comprehensive characteristic data of the light sources; Obtaining a shadow rendering performance requirement index according to the shadow precision requirement data and the occlusion relationship complexity data; The texture mapping rendering performance requirement index is obtained according to the texture resolution data and the number of texture types.

4. The intelligent scheduling method based on distributed real-time rendering according to claim 3 is characterized in that: The obtaining of the performance parameters and real-time performance degradation data of each computing node, and obtaining the processing performance index of each computing node according to the performance parameters and the real-time performance degradation data, specifically includes: Obtain the performance parameters of each computing node, including CPU performance data, GPU performance data, and storage resource data; The CPU performance data, GPU performance data, storage resource data and real-time performance degradation data are input into a preset computing node performance evaluation model for processing to obtain a processing performance index.

5. The intelligent scheduling method based on distributed real-time rendering according to claim 4 is characterized in that: The step of sorting the category rendering performance requirement index and the processing performance index respectively, matching them according to preset rules, and obtaining corresponding adaptation processing nodes specifically includes: Sorting the object model rendering performance requirement index, the lighting rendering performance requirement index, the shadow rendering performance requirement index, and the texture mapping rendering performance requirement index in descending order to obtain a sequential rendering performance requirement index; Sorting the processing performance indexes of the computing nodes in descending order to obtain a sequential processing performance index; The sequential rendering performance requirement index and the sequential processing performance index are matched one by one in sequence to obtain corresponding adaptation processing nodes.

6. The intelligent scheduling method based on distributed real-time rendering according to claim 5, characterized in that: Also includes: Obtain the urgency and importance data of each sub-rendering task of the computing node; Inputting the urgency data and the importance data into a preset rendering processing priority evaluation model for processing to obtain a rendering processing priority index; Sort the sub-rendering tasks in descending order according to the rendering processing priority index to obtain sequential sub-rendering tasks; The computing node performs rendering processing according to the sequential sub-rendering tasks.

7. The intelligent scheduling method based on distributed real-time rendering according to claim 6, characterized in that: Also includes: Get the processing progress data of the sub-rendering tasks in each computing node; Obtain average processing progress data of the rendering task according to the processing progress data; Dividing the processing progress data by the average processing progress data to obtain relative task processing progress data; Compare the relative task processing progress data with a preset task processing progress threshold to obtain a task processing progress status; Extracting a first threshold and a second threshold according to the task processing progress threshold, wherein the first threshold is greater than the second threshold; comparing the relative task processing progress data with the first threshold and the second threshold; If the relative task processing progress data is greater than the first threshold, the task processing progress state is an efficient operation state; If the relative task processing progress data is greater than the second threshold and less than or equal to the first threshold, the task processing progress state is a stable operation state; If the relative task processing progress data is less than or equal to the second threshold, the task processing progress status is an abnormal operation status.

8. An intelligent scheduling system based on distributed real-time rendering, characterized in that: The system comprises a memory and a processor, wherein the memory comprises an intelligent scheduling method program based on distributed real-time rendering, and the intelligent scheduling method program based on distributed real-time rendering is executed by the processor to implement the following steps: Acquire real-time rendering task information, process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, and extract corresponding category rendering parameters according to the category rendering tasks; Processing the category rendering parameters respectively to obtain category rendering performance requirement indexes; Acquire performance parameters and real-time performance degradation data of each computing node, and obtain a processing performance index of each computing node according to the performance parameters and the real-time performance degradation data; The category rendering performance requirement index and the processing performance index are sorted respectively, matched according to preset rules, and corresponding adaptation processing nodes are obtained.

9. The intelligent scheduling system based on distributed real-time rendering according to claim 8, characterized in that: The real-time rendering task information is obtained, and the real-time rendering task information is processed by a preset task classification model to obtain a category rendering task, and corresponding category rendering parameters are extracted according to the category rendering task, specifically including: Acquire real-time rendering task information, and process the real-time rendering task information through a preset task classification model to obtain category rendering tasks, including object model rendering tasks, lighting calculation, shadow generation, and texture mapping; Extracting object model rendering parameters according to the object model rendering task, including model scale data and geometric complexity data; Extract lighting calculation rendering parameters according to lighting calculation, including the number of light sources and comprehensive characteristic data of light sources; Extracting shadow generation rendering parameters according to shadow generation, including shadow accuracy requirement data and occlusion relationship complexity data; Extract texture map rendering parameters based on the texture map, including texture resolution data and the number of texture types.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes an intelligent scheduling method program based on distributed real-time rendering. When the intelligent scheduling method program based on distributed real-time rendering is executed by a processor, the steps of an intelligent scheduling method based on distributed real-time rendering as described in any one of claims 1 to 7 are implemented.

Images