Method, apparatus, device, and storage medium for view rendering task scheduling

Abstract

This invention relates to the field of data processing technology and discloses a view rendering task scheduling method, apparatus, device, and storage medium, comprising: collecting and analyzing system performance and real-time load data, predicting the device's idle CPU windows, and intelligently selecting a full or core resource loading strategy accordingly. A virtual rendering engine is used to quickly generate and insert lightweight virtual views to stabilize the layout, while real views are asynchronously loaded and smoothly merged using a difference algorithm, completely eliminating visual jitter. Based on view priority and real-time load data, the system dynamically schedules rendering tasks; high-priority tasks are executed immediately, while low-priority tasks are executed in segments within idle windows and support interrupt recovery. Finally, the rendered views are efficiently loaded onto the display page. This invention improves the efficiency and stability of view rendering task scheduling.

Description

1. Description of View Rendering Task Scheduling Method, Apparatus, Device, and Storage Medium Technical Field This invention relates to the field of data processing technology, and particularly to a view rendering task scheduling method, apparatus, device, and storage medium. Background Technology With the rapid development of mobile internet technology, mobile devices (such as smartphones and tablets) have become the core carriers for people to obtain information and interact with services. The interface complexity of various mobile applications (especially e-commerce, social, and news applications) continues to increase, containing a large number of layout files, images, fonts, styles, and other view resources, which places extremely high demands on view loading efficiency, rendering smoothness, and system resource utilization. Users are increasingly sensitive to the interactive experience of applications, and problems such as page loading delays, lag, and interface jumps directly affect user retention rates. Therefore, how to dynamically optimize the view resource loading and rendering process based on the device's operating status has become a core technical pain point in the field of mobile application development. In existing mobile application view processing technologies, view resource loading and rendering usually adopt fixed strategies or simple conditional judgment mechanisms, which are difficult to adapt to the dynamically changing operating load of devices. On the one hand, traditional resource loading strategies are mostly divided into two categories: serial loading and parallel loading. Serial loading loads all view resources in a fixed order, which avoids system resource contention, but its loading efficiency is low, especially in complex page scenarios where it can easily lead to delays in the first screen display. Parallel loading loads resources concurrently through multiple threads, which can improve loading speed, but it does not consider the real-time load status of the device. When CPU and memory resources are scarce, it can easily cause problems such as thread blocking, frequent garbage collection, and UI (User Interface) thread lag, thus reducing the stability of application operation. The present invention provides a view rendering task scheduling method, apparatus, device, and storage medium to solve the problems of low efficiency and poor stability of existing view rendering task scheduling methods on the market.Firstly, a view rendering task scheduling method is provided, comprising: monitoring the number of idle CPU cores, UI queue depth, and GC pressure of a preset device in real time at preset time intervals to obtain a load data sequence; analyzing the load data sequence and the system performance index data to obtain analysis results; determining whether the preset device has idle CPU threads based on the analysis results; if so, concurrently loading the first view resources corresponding to the layout information of the page to be displayed through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files; if not, sequentially loading the second view resources corresponding to the startup phase according to the order of the view resource dependencies within the page to be displayed to obtain a view resource set, wherein the core view resources include the root layout file, the view files of the first visible area, and the critical rendering path initialization data of the startup phase; A preset virtual rendering engine is invoked to generate virtual views based on the pre-acquired layout information of the page to be displayed and the set of view resources. These virtual views are then added to the view hierarchy of the virtual view, resulting in a virtual view tree containing the view hierarchy. The identification information of each virtual view contained in the layout information of the page to be displayed is obtained. It is determined whether each virtual view is a critical rendering path view during the startup phase, whether it is located within the first screen display range, and whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold. The determination results are used to confirm the rendering priority of each virtual view based on the identification information and the determination results. View rendering tasks are executed according to the load data sequence and the rendering priority of each virtual view, and a task execution status set for the view rendering tasks is generated. Based on the task execution status set, the virtual views that have been rendered in the virtual view tree are identified, and the rendered virtual views are loaded onto the display page of the preset device for display.Secondly, a view rendering task scheduling device is provided, comprising: a data acquisition module, used to monitor the number of idle CPU cores, UI queue depth, and GC pressure of a preset device in real time at preset time intervals to obtain a load data sequence; a load analysis module, used to analyze the load data sequence and the system performance index data to obtain analysis results; and a thread judgment module, used to determine whether there are idle CPU threads on the preset device based on the analysis results. If there are, the first view resources corresponding to the layout information of the page to be displayed are loaded concurrently through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files. If there are no idle CPU threads, the second view resources corresponding to the startup phase are loaded sequentially according to the order of the view resource dependencies within the page to be displayed to obtain a view resource set. The core view resources include the root layout file, the view files of the first visible area, and the initialization data of the critical rendering path in the startup phase. The view rendering module is used to call a preset virtual rendering engine to generate virtual views based on the layout information of the page to be displayed and the view resource set obtained in advance, and add the virtual views to the view hierarchy structure of the virtual views to obtain a virtual view tree containing the view hierarchy structure of the virtual views. The view display module is used to obtain the identification information of each virtual view contained in the layout information of the page to be displayed, determine whether each virtual view is a critical rendering path view in the startup phase, whether it is located within the first screen display range, and determine whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, obtain the judgment result, combine the identification information and the judgment result to determine the rendering priority of each virtual view, execute the view rendering task according to the load data sequence and the rendering priority of each virtual view, generate the task execution status set of the view rendering task, determine the virtual views that have been rendered in the virtual view tree according to the task execution status set, and load the rendered virtual views onto the display page of the preset device for display. Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned view rendering task scheduling method. Fourthly, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by the processor, it implements the steps of the aforementioned view rendering task scheduling method. In the solutions implemented by the aforementioned view rendering task scheduling method, apparatus, device, and storage medium, real-time monitoring of device performance and load data can establish a device performance profile and load baseline, providing an accurate data foundation for subsequent intelligent prediction and avoiding blind scheduling.Based on load analysis results, the system determines whether idle CPU threads exist, providing a scientific basis for resource loading strategies and preventing resource contention during high loads. If idle threads exist, all view resources for the page to be displayed are preloaded concurrently through an asynchronous thread pool, fully utilizing idle windows to improve resource loading efficiency and significantly accelerate subsequent page rendering. If no idle threads exist, the necessary core view resources are loaded sequentially according to dependencies, minimizing IO and CPU contention and ensuring basic smoothness on low-end devices. Subsequently, the virtual rendering engine is invoked to generate virtual views and construct a virtual view tree, eliminating white screen and flickering during startup. Rendering priorities are determined based on rules such as whether each view belongs to a critical rendering path, whether it is on the first screen, and the frequency of user historical operations, ensuring that high-priority views are presented first, improving the core user experience. Combined with real-time load dynamic scheduling of rendering tasks—high-priority tasks are rendered quickly, while low-priority tasks are executed in chunks during idle periods and support interrupt recovery—adaptive rendering scheduling is achieved, ensuring system smoothness. Finally, based on the task execution status set, the rendered views are loaded onto the display page, ensuring a complete and smooth interface presentation, reducing overall cold start time, and improving the efficiency and stability of view rendering task scheduling. Brief Description of the Drawings To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 is a schematic diagram of an application environment of a view rendering task scheduling method in an embodiment of the present invention; Figure 2 is a schematic flowchart of a view rendering task scheduling method in an embodiment of the present invention; Figure 3 is a schematic diagram of a structure of a view rendering task scheduling device in an embodiment of the present invention; Figure 4 is a schematic diagram of a structure of a computer device in an embodiment of the present invention; Figure 5 is another schematic diagram of a structure of a computer device in an embodiment of the present invention. Detailed Description of the Embodiments The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The view rendering task scheduling method provided by the embodiments of the present invention can be applied in the application environment shown in Figure 1, where the client communicates with the server through a network.The server can collect system performance metrics and load data sequences from the client in real time, including the number of idle CPU cores, UI queue depth (i.e., UI thread load metrics), and GC (Garbage Collection) pressure, forming a three-dimensional performance metric sequence. Then, a double exponential smoothing algorithm is used to analyze the sequence, predicting the load trend and idle CPU window within the next 600ms. Based on the prediction results, the system intelligently selects a resource loading strategy: if an idle window exists, all view resources are preloaded concurrently according to the first strategy; if not, only core resources are loaded serially according to the second strategy. During the interface rendering stage, a virtual rendering engine is introduced, which immediately generates lightweight placeholder views to stabilize the layout framework. The real views are then loaded asynchronously in the background, and a flicker-free smooth visual transition is achieved through calculating layout differences and a progressive merging algorithm. Finally, dynamic task scheduling is performed based on view priority: high-priority tasks are executed immediately, while low-priority tasks are decomposed into slices for execution within the predicted idle window, supporting interruption saving and prediction-driven recovery. Ultimately, all rendered views are efficiently and stably loaded onto the display page, thereby improving the efficiency and stability of view rendering task scheduling. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will now be described in detail through specific embodiments. Please refer to Figure 2, which is a flowchart illustrating a view rendering task scheduling method provided by an embodiment of the invention, including the following steps: S1. Real-time monitoring of the number of idle CPU cores, UI queue depth, and GC pressure of a preset device at preset time intervals to obtain a load data sequence. In this embodiment of the invention, the load data mainly includes three core metrics: First, CPU load metrics, which reflect the available computing resources of the processor by calculating the number of idle CPU cores by subtracting the number of active threads from the total number of logical cores. Second, UI thread (also called front-end thread) load metrics, which quantify the main thread pressure by detecting the backlog depth of the main thread Looper message queue (backlog depth refers to the number of unprocessed messages or the accumulated latency in the main thread Looper message queue, used to measure the congestion level of the main thread task processing), including the number of pending messages and queue waiting time. Third, GC pressure, which comprehensively assesses the pressure status of the memory subsystem by statistically analyzing the garbage collection trigger frequency, the number of global memory allocations, and the proportion of available heap memory during ART runtime. These dynamically collected time-series data together constitute the load data sequence, providing real-time input for load analysis and idle window prediction.S2. Analyze the load data sequence and the system performance index data to obtain analysis results. In this embodiment of the invention, the step of performing load analysis based on the load data sequence and the system performance index data to obtain analysis results includes: collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain a three-dimensional performance index sequence; determining whether the number of data points contained in the three-dimensional performance index sequence is greater than a preset threshold; if less, returning to the step of collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain the three-dimensional performance index sequence; if greater, using a double exponential smoothing algorithm to calculate the three-dimensional performance index of the preset device after a preset second time interval based on the three-dimensional performance index sequence, and analyzing the changing trend of the three-dimensional performance index to obtain analysis results. In detail, the step of collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain the three-dimensional performance index sequence is achieved by periodically (according to the time interval) collecting the three-dimensional index through system-level API calls and runtime status monitoring. For the number of idle CPU cores, the total number of logical cores is obtained through `Runtime.getRuntime().availableProcessors()`, and the number of active threads is obtained through `Thread.activeCount()`. The difference between the two yields the real-time number of idle cores. For the UI queue depth, the return value of the `hasMessages(0)` method is checked and the number of pending messages is calculated by obtaining the message queue object of the main thread's Looper. For GC pressure, on Android 7.0 and above, the GC count is obtained through `Debug.getRuntimeStat("art.gc.gc-count")`. In detail, the number of idle CPU cores refers to the number of currently unoccupied logical cores in the device's processor. It is calculated by subtracting the number of active threads from the total number of logical cores, and it directly reflects the processor's available computing resources at the sampling time. The UI queue depth is an indicator of the main thread's workload. It is quantified by detecting the number of pending messages (including waiting time and message count) accumulated in the Android main thread's Looper message queue. A larger depth indicates that the main thread is busier and the risk of UI response delay is higher. GC pressure represents the busyness of the memory subsystem. It is comprehensively evaluated by the garbage collection trigger frequency, global object allocation rate and heap memory usage ratio, which are statistically analyzed by ART runtime. A high pressure value indicates that there may be a memory bottleneck or lag.Specifically, the preset threshold can refer to the data length of five data collections at the specified time interval. Specifically, the preset second time interval can be 600ms. In this embodiment of the invention, the step of using a double exponential smoothing algorithm to calculate the three-dimensional performance indicators of the preset device after a preset second time interval based on the three-dimensional performance indicator sequence, and analyzing the changing trend of the three-dimensional performance indicators to obtain the analysis results includes: performing weighted recursive calculations on the performance indicators at each time point in the three-dimensional performance indicator sequence according to the time order based on preset hyperparameters to obtain smoothed estimates of the CPU sequence, UI sequence, and GC sequence; performing weighted recursive calculations on the unit-time change of the performance indicators at each time point in the three-dimensional performance indicator sequence according to the time order based on preset hyperparameters to obtain the final unit-time change of the CPU sequence, UI sequence, and GC sequence; multiplying the final unit-time change of the CPU sequence by the prediction step number in the preset hyperparameters and adding it to the smoothed estimate of the CPU sequence to obtain the final number of idle CPU cores; multiplying the final unit-time change of the UI sequence by the prediction step number in the preset hyperparameters and adding it to the smoothed estimate of the UI sequence to obtain the final UI queue depth. The final unit-time change of the GC sequence is multiplied by the prediction steps in the preset hyperparameters and then added to the smoothed estimate of the GC sequence to obtain the final GC pressure. The final unit-time change of the CPU sequence, the final unit-time change of the UI sequence, and the final unit-time change of the GC sequence are weighted and summed to obtain the trend of the three-dimensional performance indicators. The final number of idle CPU cores, the final UI queue depth, the final GC pressure, and the trend are integrated to obtain the analysis results. In detail, the performance indicators of each time point in the three-dimensional performance indicator sequence are weighted and recursively calculated in chronological order based on preset hyperparameters to obtain smoothed estimates of the CPU sequence, UI sequence, and GC sequence. This is achieved by iteratively processing each time point in the sequence using a horizontal smoothing coefficient α (usually set to 0.2): For time point t, the horizontal component is updated according to the formula L_t = α × Y_t + (1-α) × (L_{t-1} + T_{t-1}), where Y_t is the actual observed value at time point t, L_{t-1} is the horizontal component of the previous time point, and T_{t-1} is the unit time change of the previous time point.In detail, based on preset hyperparameters, the unit-time change of the performance indicators at each time point in the three-dimensional performance indicator sequence is calculated recursively and weighted in chronological order to obtain the final unit-time change of the CPU sequence, the final unit-time change of the UI sequence, and the final unit-time change of the GC sequence. A trend smoothing coefficient β (usually set to 0.1) is used to iteratively process each time point in the sequence: at each time point t, the unit-time change is updated according to the formula T_t = β×(L_t-L_{t-1})+(1-β)×T_{t-1}, where L_t is the horizontal component just calculated at the current time point, L_{t-1} is the horizontal component at the previous time point, and T_{t-1} is the unit-time change at the previous time point. S3. Based on the analysis results, determine whether the preset device has idle CPU threads. In this embodiment of the invention, determining whether the preset device has idle CPU threads based on the analysis results includes: obtaining the three-dimensional performance indicators and the trend direction of the three-dimensional performance indicators contained in the analysis results; weighting and summing the number of idle CPU cores, UI queue depth, and GC pressure in the future three-dimensional indicators according to preset weight coefficients to obtain the load score of the preset device; if the load score is less than a preset score threshold and the trend direction of the three-dimensional performance indicators is decreasing, then it is determined that the preset device has idle CPU threads; if the load score is greater than or equal to the preset score threshold or the trend direction of the three-dimensional performance indicators is increasing, then it is determined that the preset device does not have idle CPU threads. If they exist, then S4 is executed, concurrently loading the first view resources corresponding to the layout information of the page to be displayed pre-acquired through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files. In this embodiment of the invention, the first view resources corresponding to the layout information of the page to be displayed, which are obtained in advance, are concurrently loaded through a preset asynchronous thread pool to obtain a view resource set. The view resources include layout files, images, fonts, and style files. This is achieved by parsing all view resources (including layout files, images, fonts, styles, etc.) required by the current page layout request and classifying these resources by type and priority. Then, all secondary resources (such as non-first-screen images, prefetched data, cache modules, etc.) are concurrently loaded through the preset asynchronous thread pool. If these resources do not exist, step S5 is executed, and the second view resources corresponding to the startup phase are sequentially loaded according to the order of the view resource dependencies within the page to be displayed to obtain a view resource set. The core view resources include the root layout file, the view files of the first-screen visible area, and the critical rendering path initialization data of the startup phase.In this embodiment of the invention, the second view resources corresponding to the startup phase are loaded sequentially according to the order of the view resource dependencies within the page to be displayed, resulting in a view resource set. The core view resources include the root layout file, the view files for the first visible area, and the initialization data for the critical rendering path during the startup phase. Essential core resources for the startup phase (such as the root layout file of the main Activity, the view XML for the first visible area, and the initialization data for the critical rendering path during the startup phase) are selected from all view resources. These resources are then loaded in a background thread using a sequential or very low-concurrency approach to minimize I / O and CPU contention. S6: A preset virtual rendering engine is invoked to generate a virtual view based on the pre-acquired layout information of the page to be displayed and the view resource set. The virtual view is added to the view hierarchy structure of the virtual view, resulting in a virtual view tree containing the view hierarchy structure of the virtual view. In this embodiment of the invention, the step of invoking a preset virtual rendering engine to generate a virtual view based on the pre-acquired layout information of the page to be displayed and the view resource set includes: parsing the layout resource identifier and the reference information of the target view container contained in the view layout request to obtain a virtual rendering task, wherein the target view container is a view component in the virtual view tree used to carry the virtual view; generating a lightweight view for temporary placement based on the virtual rendering task to obtain a placeholder view; parsing and constructing the actual view object from the corresponding layout file by invoking an asynchronous thread based on the layout resource identifier in the virtual rendering task to obtain a real view; calculating the layout difference between the placeholder view and the real view; and merging the placeholder view and the real view based on the layout difference to obtain a virtual view. In detail, the virtual rendering task is obtained by parsing the layout resource identifier and target view container reference information contained in the view layout request. The target view container is the view component in the virtual view tree used to host the virtual view. It is obtained by acquiring the layout resource ID and target container information (such as a reference or ID of the parent ViewGroup) from the view layout request. Then, according to preset business logic, the request is split into independent rendering tasks. For example, a complex page may contain multiple child layouts that need to be loaded asynchronously. Each rendering task is encapsulated into a task object containing information such as resource ID, container reference, priority, and callback listener.In detail, the step of generating a lightweight view for temporary placeholders based on the virtual rendering task, to obtain the placeholder view, uses ViewStub technology. A placeholder object is instantiated based on the virtual rendering task using the `new ViewStub(context)` method, and then the `setLayoutResource()` method is called to bind the resource ID from the virtual rendering task to the ViewStub, while simultaneously setting the layout parameters of the ViewStub (width and height are usually set to 0). In detail, the step of parsing and constructing the actual view object from the corresponding layout file based on the layout resource identifier in the virtual rendering task by calling an asynchronous thread, as described in HK 20135853 A 8, involves submitting an asynchronous task for each virtual rendering task and calling the `LayoutInflater.from(context).inflate(resId, null)` method in the thread to parse the virtual rendering task and construct the actual view. In detail, the step of merging the placeholder view and the actual view based on the layout differences to obtain the virtual view involves finding the position index of the placeholder view and then gradually replacing or updating view nodes based on the layout differences. During the merging process, a gradual visual transition is achieved using the Transition animation to avoid interface jumps. Specifically, the layout differences refer to the differences between the placeholder view and the actual view in terms of layout attributes such as size, position, margins, and visibility. Specifically, this manifests as differences in the area occupied on the screen, relative coordinates, and view hierarchy structure. In this embodiment of the invention, adding the virtual view to the view hierarchy structure of the virtual view to obtain a virtual view tree containing the virtual view is achieved by inserting the generated virtual view into the target position in a preset view tree framework (usually the root container of the Activity, such as DecorView or a specified target view container) in the main thread after receiving the view layout request, and setting its layout parameters (width and height set to 0 to avoid participating in layout calculations), thereby quickly constructing a complete view hierarchy structure containing placeholder nodes, i.e., a virtual view tree.In this embodiment of the invention, calculating the layout difference between the placeholder view and the actual view includes: extracting view structure information from the placeholder view to obtain a virtual view structure description object; parsing the view structure of the actual view to obtain a real view tree structure description object; comparing the differences between the placeholder view and the actual view in view class name, layout parameters, visibility state, and background style based on the real view tree structure description object and the virtual view structure description object to obtain a root view attribute comparison result; comparing the topology of the placeholder view and the actual view based on the real view tree structure description object and the virtual view structure description object to obtain a structure comparison result; and calculating the layout position difference between the placeholder view and the actual view based on the root view attribute comparison result and the structure comparison result to obtain the layout difference. In detail, the step of extracting the view structure information from the virtual placeholder image to obtain the virtual view structure description object is achieved by obtaining the class name, resource ID, and basic attributes (such as width, height, visibility, and layout parameters) of the virtual view, then using getLayoutParams() to obtain the LayoutParams object and extracting layout attributes such as margin and gravity; obtaining the view identification information through getTag() and getId(); and simultaneously recording its index position in the target view container. In detail, the class name, ID, layout parameters, and all style attributes of the real view are obtained; the actual measured size is obtained through getMeasuredWidth() and getMeasuredHeight(); all custom attributes, event listener settings, and animation states of the view are recorded. Specifically, parsing the view structure of the real view to obtain a real view tree structure description object involves comparing the placeholder view and the real view in terms of view class name, layout parameters, visibility state, and background style based on the real view tree structure description object and the virtual view structure description object. The root view attribute comparison result is obtained by extracting the root node's class name, resource ID, layout parameters (width, height, mode, and size), visibility, background, and other core attributes from the virtual and real view structure description objects respectively; and using an equality comparison algorithm to compare each item, calculating the difference for numerical attributes (such as width and height).In detail, the step of comparing the topology of the placeholder view and the real view based on the real view tree structure description object and the virtual view structure description object to obtain the structure comparison result is achieved by analyzing whether the position indices of the two views are the same in the same target view container, and by checking their relative positional relationship in the layout flow (such as the order in LinearLayout, the relative constraints of RelativeLayout, and the constraint relationship of ConstraintLayout). In detail, the step of calculating the layout position difference between the placeholder view and the real view based on the root view attribute comparison result and the structure comparison result to obtain the layout difference, and calculating the actual screen area (position coordinates and size) occupied by the placeholder view based on the current layout state of the target view container; then calculating the expected screen area of ​​the real view under the same layout conditions; next, comparing the spatial characteristics of the two areas, including the top-left corner coordinate offset, width change value, height change value, area change rate, etc.; finally generating quantified layout difference data. S7. Obtain the identification information of each virtual view contained in the layout information of the page to be displayed, determine whether each virtual view is a key rendering path view in the startup phase, whether it is located within the first screen display range, and determine whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, obtain the judgment result, combine the identification information and the judgment result to confirm the rendering priority of each virtual view, execute the view rendering task according to the load data sequence and the rendering priority of each virtual view, and generate the task execution status set of the view rendering task. In this embodiment of the invention, the process of obtaining the identification information of each virtual view contained in the layout information of the page to be displayed, determining whether each virtual view is a critical rendering path view in the startup phase, whether it is within the first screen display range, and determining whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, and obtaining the judgment result, and combining the identification information and the judgment result to determine the rendering priority of each virtual view, is achieved by parsing the business attributes (such as whether it is a critical rendering path view in the startup phase, user interaction frequency, data dependency relationship) and visual attributes (such as view size, position, whether it is within the first screen display range) of each virtual view in the request, and combining them with preset priority rules (such as priority of core views on the first screen, priority of user operation hotspots, priority of data update frequency) to dynamically assign a rendering priority level (such as CRITICAL, HIGH, IDLE_LOW) to each virtual view. In this embodiment of the invention, the critical rendering path view refers to the part of the view that directly affects the time required for the user to see the usable interface for the first time during the application startup process. Such views typically include the root layout, core UI components of the first screen visible area (such as the title bar, main content area, navigation bar, etc.), and controls directly related to user interaction.Their rendering must be prioritized and cannot be delayed, fragmented, or interrupted; otherwise, it will lead to longer startup times, blank screens, or incomplete interfaces perceived by the user. In this embodiment of the invention, the first-screen rendering range refers to the view area that a user can directly see within the visible area of ​​the screen without any scrolling when opening the application page. It corresponds to the physical display window of the device screen and typically contains the core content of the application, such as the title bar, main information display area, and key operation entry points. In mobile development, the boundaries of the first-screen rendering range depend on the width and height parameters of the device screen (such as resolution and pixel density) and the layout structure of the application. HK 20135853 A 10 In this embodiment of the invention, the step of executing view rendering tasks according to the load data sequence and the rendering priority of each virtual view, and generating a set of task execution states for the view rendering tasks, includes: sorting all virtual views according to the dependency relationships between virtual views in the virtual view tree, in the order in which the dependent view is executed before the dependent view, to obtain a set of virtual views; converting each virtual view in the ordered set of virtual views into a rendering task, using the rendering priority of each virtual view as the priority of the corresponding rendering task, and associating each rendering task with a corresponding view object identifier and the parameters required for rendering, to generate a rendering task queue; executing the rendering tasks in the rendering task queue with preset first priority and preset second priority, and marking the execution results to obtain a first marking result; determining whether the preset device has an idle CPU window; if not, serializing and storing the task states of the rendering tasks in the rendering task queue with preset third priority, and returning to the step of determining whether the preset device has an idle CPU window; if so, decomposing the rendering tasks in the rendering task queue with preset third priority into multiple execution fragments. The execution slices are executed sequentially. After all execution slices are completed, the execution results of the rendering tasks with a preset third task priority in the rendering task queue are marked to obtain a second marking result. The first marking result and the second marking result are summarized to obtain a task execution state set. Specifically, the step of sorting all virtual views according to the dependency relationships between virtual views in the virtual view tree, in the order that the dependent view executes before the dependent view, to obtain a virtual view set, is achieved by traversing the virtual view tree to identify parent-child relationships and business dependencies between views (such as the rendering of a view depending on the size information of another view); then, a topological sorting algorithm is used to process these dependencies to ensure that the dependent view always executes after the dependent view.In detail, the step of converting each virtual view in the ordered virtual view set into a rendering task, using the rendering priority of each virtual view as the priority of the corresponding rendering task, and associating each rendering task with a corresponding view object identifier and rendering parameters to generate a rendering task queue, involves converting each virtual view into a rendering task object, including a view reference, rendering parameters, priority settings, and callback functions. Specifically, the view object identifier is identification information used to uniquely identify a view object, such as the view's resource ID or its address in memory. Specifically, the preset first priority refers to critical priority, the preset second priority refers to high priority, and the preset third priority refers to low priority. Specifically, the step of executing rendering tasks with preset first and second priorities in the rendering task queue and marking the execution results to obtain a first marking result involves traversing the task queue to filter out tasks with first and second priorities. The first priority tasks are executed synchronously on the main thread, while the second priority tasks are submitted to a dedicated thread pool for parallel execution. After each task is completed, its execution status (success / failure), execution time, resource consumption, and rendering output are recorded. HK 20135853 A 11 Specifically, determining whether the preset device has an idle CPU window is done by monitoring the system load using the SystemLoadMonitor method to determine if there is a suitable CPU window for executing low-priority tasks. Specifically, decomposing the rendering tasks with a preset third task priority in the rendering task queue into multiple execution fragments is done by analyzing the execution flow of each low-priority task, splitting it into multiple independent fragment units based on the operation type and estimated time consumption, with each fragment's execution time controlled to ≤50ms. S8. Based on the task execution status set, confirm the virtual views that have been rendered in the virtual view tree, and load the rendered virtual views onto the display page of the preset device for display. In this embodiment of the invention, the step of confirming the rendered views in the virtual view tree based on the task execution status set involves traversing each virtual view node in the virtual view tree according to the task execution status set, matching the unique identifier of the virtual view (such as view ID or task ID) with the task identifier in the task execution status set, and finding the rendering task execution result corresponding to each view. If the task status is marked as "completed" and the rendering result is valid, the view is marked as rendered and its rendering output (such as bitmap, view reference, etc.) is generated, while the status attributes of the corresponding node in the virtual view tree are updated. Finally, a confirmation list containing all rendered views and their metadata (such as rendering completion time, resource usage) is generated.In this embodiment of the invention, loading the rendered view onto the display page of the preset device refers to dynamically attaching the rendered view to a predetermined position in the view hierarchy (such as DecorView or a specified content container) of the current Activity or Fragment of the preset device through UI operations (such as ViewGroup.addView() or replacing ViewStub). As can be seen, in the above scheme, a three-dimensional performance indicator sequence is constructed by real-time collection of the device's system performance indicators and load data sequences, including the number of idle CPU cores, UI queue depth, and GC pressure. Then, a double exponential smoothing algorithm is used to analyze the sequence to predict the load trend and idle CPU windows within the next 600ms. Based on the prediction results, the system intelligently selects a resource loading strategy: if an idle window exists, all view resources are preloaded concurrently according to the first strategy; if not, only core resources are loaded serially according to the second strategy. During the interface rendering stage, the scheme introduces a virtual rendering engine, which immediately generates a lightweight placeholder view to stabilize the layout framework, and then asynchronously loads the real view in the background, achieving a flicker-free smooth visual transition through calculating layout differences and a progressive merging algorithm. Finally, dynamic task scheduling is performed based on view priority: high-priority tasks are executed immediately, while low-priority tasks are decomposed into slices for execution within the predicted idle window. This supports interruption saving and prediction-driven recovery, ultimately loading all rendered views efficiently and stably onto the display page, thereby improving the efficiency and stability of view rendering task scheduling. It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this invention. In one embodiment, a view rendering task scheduling device is provided, which corresponds one-to-one with the view rendering task scheduling method in the above embodiments. As shown in Figure 3, the view rendering task scheduling device includes a data acquisition module 101, a load analysis module 102, a thread judgment module 103, a view rendering module 104, and a view display module 105.The functional modules are described in detail below: Data acquisition module 101 is used for memory and GC pressure indicators; Load analysis module 102 is used to analyze the load data sequence and the system performance indicator data to obtain analysis results; Thread judgment module 103 is used to determine whether there are idle CPU threads in the preset device based on the analysis results. If there are, the first view resources corresponding to the layout information of the page to be displayed are loaded concurrently through the preset asynchronous thread pool to obtain a view resource set. The view resources include layout files, images, fonts, and style files. If there are no idle CPU threads, the second view resources corresponding to the startup phase are loaded sequentially according to the order of the view resource dependencies in the page to be displayed to obtain a view resource set. The core view resources include the root layout file, the view file of the first visible area, and the initialization data of the key rendering path in the startup phase; View rendering module 104 is used to call the preset virtual rendering engine to generate a virtual view based on the layout information of the page to be displayed and the view resource set, and add the virtual view to the view hierarchy structure of the virtual view to obtain a virtual view tree containing the view hierarchy structure of the virtual view. The view display module 105 is used to obtain the identification information of each virtual view contained in the layout information of the page to be displayed, determine whether each virtual view is a key rendering path view in the startup phase, whether it is located within the first screen display range, and determine whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, obtain the judgment result, combine the identification information and the judgment result to determine the rendering priority of each virtual view, execute the view rendering task according to the load data sequence and the rendering priority of each virtual view, generate the task execution status set of the view rendering task, determine the virtual views that have been rendered in the virtual view tree according to the task execution status set, and load the rendered virtual views onto the display page of the preset device for display. In one embodiment, the data acquisition module 101, when performing load analysis based on the load data sequence and the system performance index data to obtain analysis results, specifically performs the following steps: Collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain a three-dimensional performance index sequence; determining whether the number of data points contained in the three-dimensional performance index sequence is greater than a preset threshold; if less, returning to the step of collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain the three-dimensional performance index sequence; if greater, using a double exponential smoothing algorithm to calculate the three-dimensional performance index of the preset device after a preset second time interval based on the three-dimensional performance index sequence, and analyzing the changing trend of the three-dimensional performance index to obtain analysis results.In one embodiment, when the data acquisition module 101 executes the calculation of the three-dimensional performance indicators of the preset device based on the three-dimensional performance indicator sequence using the double exponential smoothing algorithm after a preset second time interval, and analyzes the changing trend of the three-dimensional performance indicators to obtain the analysis results, it specifically performs the following: Based on preset hyperparameters, it performs weighted recursive calculations on the performance indicators at each time point in the three-dimensional performance indicator sequence in chronological order to obtain smoothed estimates of the CPU sequence, UI sequence, and GC sequence; Based on preset hyperparameters, it performs weighted recursive calculations on the unit-time change of the performance indicators at each time point in the three-dimensional performance indicator sequence in chronological order to obtain the final unit-time change of the CPU sequence, UI sequence, and GC sequence; It multiplies the final unit-time change of the CPU sequence by the prediction step number in the preset hyperparameters and adds it to the smoothed estimate of the CPU sequence to obtain the final number of idle CPU cores; The final UI queue depth is obtained by multiplying the final unit-time change of the UI sequence by the prediction steps in the preset hyperparameters and then adding it to the smoothed estimate of the UI sequence. The final GC pressure is obtained by multiplying the final unit-time change of the GC sequence by the prediction steps in the preset hyperparameters and then adding it to the smoothed estimate of the GC sequence. The final unit-time change of the CPU sequence, the UI sequence, and the GC sequence are weighted and summed to obtain the trend of the three-dimensional performance indicators. The final number of idle CPU cores, the final UI queue depth, the final GC pressure, and the trend are integrated to obtain the analysis results. In one embodiment, the load analysis module 102, when performing the step of determining whether the preset device has idle CPU threads based on the analysis results, is specifically configured to: obtain the three-dimensional performance indicators and the trend direction of the three-dimensional performance indicators contained in the analysis results; perform a weighted summation of the number of idle CPU cores, UI queue depth, and GC pressure in the future three-dimensional indicators according to preset weight coefficients to obtain the load score of the preset device; if the load score is less than a preset score threshold and the trend direction of the three-dimensional performance indicators is decreasing, then it is determined that the preset device has idle CPU threads; if the load score is greater than or equal to the preset score threshold or the trend direction of the three-dimensional performance indicators is increasing, then it is determined that the preset device does not have idle CPU threads.In one embodiment, when the view rendering module 104 executes the call to a preset virtual rendering engine to generate a virtual view based on the pre-acquired layout information of the page to be displayed and the view resource set, it specifically performs the following steps: Parsing the layout resource identifier and the reference information of the target view container contained in the view layout request to obtain a virtual rendering task, wherein the target view container is a view component in the virtual view tree used to carry the virtual view; Generating a lightweight view for temporary placement based on the virtual rendering task to obtain a placeholder view; Parsing and constructing the actual view object from the corresponding layout file based on the layout resource identifier in the virtual rendering task by calling an asynchronous thread to obtain a real view; Calculating the layout difference between the placeholder view and the real view; Merging the placeholder view and the real view based on the layout difference to obtain a virtual view. In one embodiment, the view rendering module 104, when performing the calculation of the layout difference between the placeholder view and the real view, specifically performs the following steps: Extracting view structure information from the placeholder image to obtain a virtual view structure description object; Parsing the view structure of the real view to obtain a real view tree structure description object; Comparing the differences between the placeholder view and the real view in view class name, layout parameters, visibility state, and background style based on the real view tree structure description object and the virtual view structure description object to obtain a root view attribute comparison result; Comparing the topology of the placeholder view and the real view based on the real view tree structure description object and the virtual view structure description object to obtain a structure comparison result; Calculating the layout position difference between the placeholder view and the real view based on the root view attribute comparison result and the structure comparison result to obtain the layout difference.In one embodiment, the view display module 105, when executing the view rendering task based on the load data sequence and the rendering priority of each virtual view, and generating the task execution status set of the view rendering task, specifically performs the following steps: Based on the dependencies between virtual views in the virtual view tree, sort all virtual views according to the order in which the dependent view is executed before the dependent view, to obtain a virtual view set; convert each virtual view in the ordered virtual view set into a rendering task, using the rendering priority of each virtual view as the priority of the corresponding rendering task, and associating each rendering task with a corresponding view object identifier and the parameters required for rendering, to generate a rendering task queue; execute the rendering tasks in the rendering task queue with preset first priority and preset second priority, and mark the execution results to obtain a first marking result; determine whether the preset device has an idle CPU window; if not, serialize and store the task status of the rendering tasks with preset third priority in the rendering task queue, and then return to the step of determining whether the preset device has an idle CPU window; if so, decompose the rendering tasks with preset third priority in the rendering task queue into multiple execution fragments; The execution slices are executed sequentially. After all execution slices are completed, the execution results of the rendering tasks with a preset third task priority in the rendering task queue are marked to obtain a second marking result. The first marking result and the second marking result are summarized to obtain a task execution status set. This invention provides a view rendering task scheduling device that collects system performance indicators and load data sequences of the device in real time, including the number of idle CPU cores, UI queue depth, and GC pressure, to form a three-dimensional performance indicator sequence. Then, the sequence is analyzed using a double exponential smoothing algorithm to predict the load trend and idle CPU window within the next 600ms. Based on the prediction results, the system intelligently selects a resource loading strategy: if there is an idle window, all view resources are preloaded concurrently according to the first strategy; if not, only core resources are loaded serially according to the second strategy. In the interface rendering stage, the solution introduces a virtual rendering engine, which immediately generates a lightweight placeholder view to stabilize the layout framework, and then asynchronously loads the real view in the background, and achieves a smooth visual transition without screen flicker by calculating layout differences and using a progressive merging algorithm. Finally, dynamic task scheduling is performed based on view priority: high-priority tasks are executed immediately, while low-priority tasks are broken down into slices for execution within the predicted idle window. This supports interruption saving and HK 20135853 A 15 prediction-driven recovery, ultimately loading all rendered views onto the display page efficiently and stably, thereby improving the efficiency and stability of view rendering task scheduling.Specific limitations regarding the view rendering task scheduling device can be found in the limitations of the view rendering task scheduling method described above, and will not be repeated here. Each module in the aforementioned view rendering task scheduling device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module. In one embodiment, a computer device is provided, which can be a server, and its internal structure diagram can be shown in Figure 4. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a view rendering task scheduling method on the server side. In one embodiment, a computer device is provided, which can be a client, and its internal structure can be as shown in Figure 5. The computer device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a view rendering task scheduling method on the client side.In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps: Memory and GC pressure metrics; Load analysis is performed based on the load data sequence and the system performance metrics data to obtain analysis results; Based on the analysis results, it is determined whether the preset device has idle CPU threads; If so, a first view resource corresponding to the layout information of the page to be displayed is concurrently loaded through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files; If not, a second view resource corresponding to the startup phase is sequentially loaded according to the order of view resource dependencies within the page to be displayed to obtain a view resource set, wherein the core view resource includes a root layout file, view files for the first visible area, and initialization data for the critical rendering path during the startup phase; A preset virtual rendering engine is invoked to generate a virtual view based on the layout information of the page to be displayed and the HK 20135853 A 16 view resource set, and the virtual view is added to the view hierarchy structure of the virtual view to obtain a virtual view tree containing the view hierarchy structure of the virtual view; The system obtains the identification information of each virtual view contained in the layout information of the page to be displayed, determines whether each virtual view is a critical rendering path view in the startup phase, whether it is located within the first screen display range, and whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, obtains the judgment result, combines the identification information and the judgment result to determine the rendering priority of each virtual view, executes the view rendering task according to the load data sequence and the rendering priority of each virtual view, and generates a task execution status set of the view rendering task; determines the virtual views that have been rendered in the virtual view tree according to the task execution status set, and loads the rendered virtual views onto the display page of the preset device for display.In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it performs the following steps: Memory and GC pressure indicators; Load analysis is performed based on the load data sequence and the system performance indicator data to obtain analysis results; Based on the analysis results, it is determined whether the preset device has idle CPU threads; If so, a first view resource corresponding to the layout information of the page to be displayed is concurrently loaded through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files; If not, a second view resource corresponding to the startup phase is sequentially loaded according to the order of the view resource dependencies within the page to be displayed to obtain a view resource set, wherein the core view resources include the root layout file, the view file of the first visible area, and the critical rendering path initialization data of the startup phase; A preset virtual rendering engine is invoked to generate a virtual view based on the layout information of the page to be displayed and the view resource set, and the virtual view is added to the view hierarchy structure of the virtual view to obtain a virtual view tree containing the view hierarchy structure of the virtual view; The process involves obtaining the identification information of each virtual view contained in the layout information of the page to be displayed, determining whether each virtual view is a critical rendering path view during the startup phase, whether it is located within the first screen display range, and whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold. The determination results are then used to confirm the rendering priority of each virtual view based on the identification information and the determination results. A view rendering task is executed according to the load data sequence and the rendering priority of each virtual view, and a task execution status set for the view rendering task is generated. Based on the task execution status set, the virtual views that have been rendered in the virtual view tree are identified, and the rendered virtual views are loaded onto the display page of the preset device for display. It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be found in the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described in detail here. Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory.Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. Finally, it should be noted that if any software tools or components not belonging to this company appear in the embodiments of the application, they are merely illustrative examples and do not represent actual use. The embodiments described above are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.HK 20135853 A 1 Claim 1. A view rendering task scheduling method, characterized in that it includes: real-time monitoring of the number of idle CPU cores, UI queue depth, and GC pressure of a preset device at preset time intervals to obtain a load data sequence; analyzing the load data sequence and the system performance index data to obtain analysis results; determining whether the preset device has idle CPU threads based on the analysis results; if so, concurrently loading the first view resources corresponding to the layout information of the page to be displayed through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files; if not, sequentially loading the second view resources corresponding to the startup phase according to the order of the view resource dependencies within the page to be displayed to obtain a view resource set, wherein the core view resources include the root layout file, the view files of the first visible area, and the critical rendering path initialization data of the startup phase; A preset virtual rendering engine is invoked to generate virtual views based on the pre-acquired layout information of the page to be displayed and the set of view resources. These virtual views are then added to the view hierarchy of the virtual view, resulting in a virtual view tree containing the view hierarchy. The identification information of each virtual view contained in the layout information of the page to be displayed is obtained. It is determined whether each virtual view is a critical rendering path view during the startup phase, whether it is located within the first screen display range, and whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold. The determination results are used to confirm the rendering priority of each virtual view based on the identification information and the determination results. View rendering tasks are executed according to the load data sequence and the rendering priority of each virtual view, and a task execution status set for the view rendering tasks is generated. Based on the task execution status set, the virtual views that have been rendered in the virtual view tree are identified, and the rendered virtual views are loaded onto the display page of the preset device for display.2. The view rendering task scheduling method as described in claim 1, characterized in that the step of performing load analysis based on the load data sequence and the system performance index data to obtain analysis results includes: collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain a three-dimensional performance index sequence; determining whether the number of data points contained in the three-dimensional performance index sequence is greater than a preset threshold; if less than a threshold, returning to the step of collecting the number of idle CPU cores, UI queue depth, and GC pressure contained in the load data sequence to obtain the three-dimensional performance index sequence; if greater than a threshold, using a double exponential smoothing algorithm to calculate the three-dimensional performance index of the preset device after a preset second time interval based on the three-dimensional performance index sequence, and analyzing the changing trend of the three-dimensional performance index to obtain analysis results. 3. The view rendering task scheduling method as described in claim 2, characterized in that the step of using a double exponential smoothing algorithm to calculate the three-dimensional performance index of the preset device after a preset second time interval based on the three-dimensional performance index sequence, and analyzing the changing trend of the three-dimensional performance index to obtain the analysis results, includes: performing weighted recursive calculation on the performance index at each time point in the three-dimensional performance index sequence according to the time sequence based on preset hyperparameters to obtain the smoothed estimate of the CPU sequence, the smoothed estimate of the UI sequence, and the smoothed estimate of the GC sequence; performing weighted recursive calculation on the unit time change of the performance index at each time point in the three-dimensional performance index sequence according to the time sequence based on preset hyperparameters to obtain the final unit time change of the CPU sequence, the final unit time change of the UI sequence, and the final unit time change of the GC sequence; multiplying the final unit time change of the CPU sequence by the prediction step number in the preset hyperparameters and adding it to the smoothed estimate of the CPU sequence to obtain the final number of idle CPU cores; The final UI queue depth is obtained by multiplying the final unit-time change of the UI sequence by the prediction steps in the preset hyperparameters and then adding it to the smoothed estimate of the UI sequence. The final GC pressure is obtained by multiplying the final unit-time change of the GC sequence by the prediction steps in the preset hyperparameters and then adding it to the smoothed estimate of the GC sequence. The final unit-time change of the CPU sequence, the UI sequence, and the GC sequence are weighted and summed to obtain the trend of the three-dimensional performance indicators. The final number of idle CPU cores, the final UI queue depth, the final GC pressure, and the trend are integrated to obtain the analysis results.4. The view rendering task scheduling method as described in claim 1, characterized in that, determining whether the preset device has idle CPU threads based on the analysis results includes: obtaining the three-dimensional performance indicators and the trend direction of the three-dimensional performance indicators contained in the analysis results; weighting and summing the number of idle CPU cores, UI queue depth, and GC pressure in the future three-dimensional indicators according to preset weight coefficients to obtain the load score of the preset device; if the load score is less than a preset score threshold and the trend direction of the three-dimensional performance indicators is decreasing, then it is determined that the preset device has idle CPU threads; if the load score is greater than or equal to the preset score threshold or the trend direction of the three-dimensional performance indicators is increasing, then it is determined that the preset device does not have idle CPU threads. 5. The view rendering task scheduling method as described in claim 1, characterized in that, the step of calling a preset virtual rendering engine to generate a virtual view based on the layout information of the page to be displayed and the view resource set obtained in advance includes: parsing the layout resource identifier and the reference information of the target view container contained in the view layout request to obtain a virtual rendering task, wherein the target view container is a view component in the virtual view tree used to carry the virtual view; generating a lightweight view for temporary placeholder based on the virtual rendering task to obtain a placeholder view; parsing and constructing the actual view object from the corresponding layout file based on the layout resource identifier in the virtual rendering task by calling an asynchronous thread to obtain a real view; calculating the layout difference between the placeholder view and the real view; and merging the placeholder view and the real view based on the layout difference to obtain a virtual view. 6. The view rendering task scheduling method as described in claim 5, characterized in that, calculating the layout difference between the placeholder view and the real view includes: extracting view structure information from the virtual placeholder image to obtain a virtual view structure description object; parsing the view structure of the real view to obtain a real view tree structure description object; comparing the differences between the placeholder view and the real view in view class name, layout parameters, visibility state, and background style according to the real view tree structure description object and the virtual view structure description object to obtain a root view attribute comparison result; comparing the topology of the placeholder view and the real view according to the real view tree structure description object and the virtual view structure description object to obtain a structure comparison result; calculating the layout position difference between the placeholder view and the real view according to the root view attribute comparison result and the structure comparison result to obtain the layout difference.7. The view rendering task scheduling method as described in claim 1, characterized in that, the step of executing view rendering tasks according to the load data sequence and the rendering priority of each virtual view, and generating a set of task execution states for the view rendering tasks, includes: sorting all virtual views according to the dependency relationship between virtual views in the virtual view tree, in the order in which the dependent view is executed before the dependent view, to obtain a set of virtual views; converting each virtual view in the ordered set of virtual views into a rendering task, using the rendering priority of each virtual view as the priority of the corresponding rendering task, and associating each rendering task with a corresponding view object identifier and the parameters required for rendering, to generate a rendering task queue; executing the rendering tasks in the rendering task queue that have a preset first priority and a preset second priority, and marking the execution results to obtain a first marking result; determining whether the preset device has an idle CPU window; if not, serializing and storing the task states of the rendering tasks in the rendering task queue that have a preset third priority, and then returning to the step of determining whether the preset device has an idle CPU window; if so, decomposing the rendering tasks in the rendering task queue that have a preset third priority into multiple execution fragments; The execution slices are executed sequentially. After all execution slices are executed, the execution results of the rendering tasks with a preset third task priority in the rendering task queue are marked to obtain a second marking result. The first marking result and the second marking result are summarized to obtain a task execution status set.8. A view rendering task scheduling device, characterized in that it comprises: a data acquisition module, used to monitor the number of idle CPU cores, UI queue depth, and GC pressure of a preset device in real time at preset time intervals to obtain a load data sequence; a load analysis module, used to analyze the load data sequence and the system performance index data to obtain analysis results; and a thread judgment module, used to determine whether there are idle CPU threads in the preset device according to the analysis results. If there are, the first view resources corresponding to the layout information of the page to be displayed are loaded concurrently through a preset asynchronous thread pool to obtain a view resource set, wherein the view resources include layout files, images, fonts, and style files. If there are no idle CPU threads, the second view resources corresponding to the startup phase are loaded sequentially according to the order of the view resource dependencies in the page to be displayed to obtain a view resource set. The core view resources include the root layout file, the view file of the first visible area, and the key rendering path initialization data of the startup phase. The view rendering module is used to call a preset virtual rendering engine to generate virtual views based on the layout information of the page to be displayed and the view resource set obtained in advance, and add the virtual views to the view hierarchy structure of the virtual views to obtain a virtual view tree containing the view hierarchy structure of the virtual views. The view display module is used to obtain the identification information of each virtual view contained in the layout information of the page to be displayed, determine whether each virtual view is a critical rendering path view in the startup phase, whether it is located within the first screen display range, and determine whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold, obtain the judgment result, combine the identification information and the judgment result to determine the rendering priority of each virtual view, execute the view rendering task according to the load data sequence and the rendering priority of each virtual view, generate the task execution status set of the view rendering task, determine the virtual views that have been rendered in the virtual view tree according to the task execution status set, and load the rendered virtual views onto the display page of the preset device for display. 9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the steps of the view rendering task scheduling method as described in any one of claims 1 to 7. 10. A computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the steps of the view rendering task scheduling method as described in any one of claims 1 to 7.HK 20135853 A 1 Instruction Manual Appendix: Network Client / Server Diagram 1 HK 20135853 A 2 Real-time monitoring of the number of idle CPU cores, UI queue depth, and GC pressure of the preset device at preset time intervals to obtain a load data sequence. Analysis of the load data sequence and the system performance index data yields analysis results. S1 Based on the analysis results, it is determined whether the preset device has idle CPU threads. Based on the task execution status set, it is confirmed that the virtual views in the virtual view tree have been rendered, and the rendered virtual views are loaded onto the display page of the preset device for display. S2 S3 S4 The system concurrently loads the first view resources corresponding to the layout information of the page to be displayed, obtained from a pre-set asynchronous thread pool, to obtain a set of view resources. These view resources include layout files, images, fonts, and style files. A pre-set virtual rendering engine is then invoked to generate virtual views based on the pre-acquired layout information of the page to be displayed and the set of view resources. These virtual views are added to the view hierarchy of the virtual view, resulting in a virtual view tree containing the view hierarchy. The second view resources corresponding to the startup phase are then sequentially loaded according to the order of view resource dependencies within the page to be displayed, thus obtaining the set of view resources. The core view resources include the root layout file, the view file of the first visible area, and the initialization data of the critical rendering path during the startup phase. The system obtains the identification information of each virtual view contained in the layout information of the page to be displayed, determines whether each virtual view is a critical rendering path view during the startup phase, whether it is located within the first display area, and whether the user's historical operation frequency on the corresponding area of ​​the view is greater than a preset frequency threshold. The system obtains the judgment result, combines the identification information and the judgment result to confirm the rendering priority of each virtual view, executes the view rendering task according to the load data sequence and the rendering priority of each virtual view, and generates the task execution state set S6 S7 S8 S5 of the view rendering task. Figure 2 HK 20135853 A 3 Load analysis module Thread judgment module View rendering module Data acquisition module View rendering task scheduling device View display module 101 102 103 104 105 Figure 3 Computer equipment Non-volatile storage medium Operating system Computer program Internal memory Processor Network interface System bus Database Figure 4 HK 20135853 A 4 Computer equipment, non-volatile storage media, operating system, computer program, internal memory, processor, network interface, system bus, input device, display screen. Figure 5 HK 20135853 A.

Images