15results about How to "Improve loading speed" patented technology

The embodiment of the invention provides a node graph rendering method and related equipment. The node graph rendering method comprises the following steps: monitoring the number of nodes contained in a node graph; if the number of the nodes does not exceed a preset node number threshold value, using a real-time rendering mode to render the node graph; and if the number of the nodes exceeds a preset node number threshold value, rendering the node graph by using a background rendering mode. According to the technical scheme, the number of the nodes in the node graph is monitored, real-time rendering is forbidden when the number of the nodes exceeds the preset node number threshold value, background rendering is used instead, the final layout of all the nodes is calculated through a background program, and after calculation is completed, the layouts of all the nodes are rendered into the page at a time, so that the layout of all the nodes is rendered into the page at a time. According to the method, frequent calculation and rendering are reduced, and the performance bottleneck of real-time rendering is avoided, so that the page loading speed and the user experience are remarkably improved.

The embodiment of the application provides a kind of data processing unit of sparse convolutional neural network, it is related to data processing field.Data processing unit includes weight storage module, activation cache module, weight decoding module, activation decoding module, weight loading queue, activation loading queue and operation and output cache module;Weight storage module and activation cache module are used to temporarily store data, weight decoding module and activation decoding module are used to decode sparse coded data;Weight loading queue and activation loading queue are used to load data into operation and output cache module;Operation and output cache module are used to multiply and accumulate operation to weight and activation data and temporarily store multiply and accumulate operation result in output cache.The embodiment of the application can read multiple activation data and convolution weight in parallel, greatly improve the loading speed;Most of the clock cycle, multiplier can be effectively applied;Convolution result cache addressing logic is simple, avoids addressing congestion problem.

The invention discloses an animation data compression method, apparatus and device, and a computer readable storage medium. The method comprises the steps of obtaining to-be-compressed animation data and network state information; performing data compression on the to-be-compressed animation data to obtain initial compressed animation data; determining a target compression parameter based on the initial compression animation data and the network state information; and performing data compression on the initial compressed animation data based on the target compression parameter to obtain target animation data. By adopting the method provided by the invention, the data volume of the animation data to be compressed can be greatly reduced, and the visual fidelity of the core element is ensured.

The application discloses a game scene change control method, which comprises the following steps: creating at least two prefabricated puzzle elements corresponding to a target scene; when a virtual character controlled by a player enters the target scene, providing the prefabricated puzzle elements for the player to select, and generating the prefabricated puzzle elements selected by the player in the target scene; when the player changes the state of the prefabricated puzzle elements generated in the target scene through operation, storing final state information of a final state of the prefabricated puzzle elements, and deleting intermediate state information of the prefabricated puzzle elements in the state change process; when the virtual character controlled by the player reenters the target scene, loading necessary data capable of restoring the final state of the corresponding prefabricated puzzle elements from the corresponding prefabricated puzzle elements according to the final state information. The application improves the flexibility of the fusion of scene data and puzzle solving data in the game, reduces the memory pressure of the scene running, and improves the game experience of the player.

The invention provides a method and system for enhancing the model reasoning speed, a storage medium and electronic equipment. The method comprises the following steps: in each iteration of each functional layer in a plurality of functional layers of a target model, executing an operation comprising the following steps: executing arithmetic logic of a current functional layer in a current iteration based on at least one processor, reading key value KV data generated by a next functional layer in a previous iteration before the current iteration in advance from a storage device, and loading the key value KV data into a memory; writing KV data generated by the current function layer in the current iteration into a buffer area based on the operation logic of the current function layer in the current iteration executed by the at least one processor; and writing all the KV data in the buffer area into the storage device based on the fact that the buffer area is full or the resource overhead is smaller than a reference value.

This invention provides a lightweight method for BIM operation and maintenance (O&M) models. The invention splits and reorganizes the pipeline model within the O&M BIM model into a new BIM model based on the system. Instead of loading the entire model and then displaying hidden components, it loads only the pipeline model for the corresponding system at a time based on O&M client commands. Simultaneously, merging pipelines reduces the number of components, accelerating model loading. By abstracting the model's geometry and O&M information, it simplifies the model without losing its O&M data. This invention significantly reduces the number of components in the O&M model, reduces the number of Drawcalls in the graphics engine, accelerates model loading and rendering, and ensures that only a portion of the model is loaded at a time, reducing memory burden while retaining the model's O&M information. Compared to traditional model loading methods, it improves model loading speed by 40% and reduces memory usage by 20%, achieving the goal of lightweighting the O&M model.

This specification provides embodiments of a distributed system computing node, a data loading method for a learning model, and a corresponding system. The data loading method for the machine learning model includes: determining the data access sequence for each iteration of the machine learning model; in a first processing step, preloading sample data corresponding to the data access sequence of each iteration into a cache pool according to the order of the iterations; in a second processing step, reading the corresponding sample data from the cache pool according to the order of the iterations; wherein the first and second processing steps are executed in parallel; and in a third processing step, inputting the sample data into the machine learning model for iterative training according to the order of the iterations; wherein the third processing step is executed in parallel with the second processing step.

The invention provides a front-end resource dynamic loading and caching optimization method based on user behavior prediction. According to the dynamic resource loading and caching strategy, a resource loading and caching elimination mechanism based on user behavior prediction is achieved through four core modules, and the incidence relation between user behaviors and resource access is established by recording the route access history, the page dwell time and the element click behavior; a priority scoring model is provided for different resources, resource priority scores are dynamically calculated to decide whether to be preloaded or not, preloading trigger mechanism judgment is added, and intelligent cache management is achieved. Finally, the problems of performance bottleneck and resource redundancy of traditional static loading are solved, user experience, resource efficiency and system stability are balanced, and the method is particularly suitable for large-scale application and complex user scenes.

The invention provides a page loading method and device, electronic equipment and a readable storage medium. The method comprises the steps of receiving an access request for a page; judging whether a target offline resource packet pre-downloaded for the page exists in a local cache or not; wherein the target offline resource bundle is pre-constructed based on version resource information of an application to which the page belongs; if yes, determining a target resource located in a resource directory of the target offline resource packet from the resources needing to be loaded on the page; otherwise, the resources are loaded through the network request; and reading the target resource from the local, and loading a non-target resource required to be loaded by the page through a network request. The page loading speed of the application program can be increased, and the user experience is improved.

The application discloses a 3D model file generation method, comprising the following steps: obtaining a preset structure of a model file to be generated; obtaining inputted multi-type data from a preset path; wherein the multi-type data comprises initial file information, mesh information and texture information; writing the initial file information into a file header in the preset structure to obtain a first initial file; writing the mesh information into the first initial file to obtain a second initial file; and writing the texture information into the second initial file to obtain a model file. The data definition fully considers the storage layout of the data type in the memory, greatly improves the loading speed, improves the operation efficiency, and reduces the memory occupation of the model file. The application also discloses a 3D model file generation device, a terminal equipment and a computer readable storage medium, which have the above beneficial effects.

This disclosure relates to the field of data processing technology, and provides a method, apparatus, electronic device, and readable storage medium for generating panoramic engineering data. The method includes: fusing a collection of multi-view engineering images to obtain panoramic index data; extracting data from the panoramic index data based on preset state requirements to obtain an index subset and corresponding image block data; combining the index subset and image block data to obtain lightweight panoramic engineering data; performing local caching and stitching processing on the lightweight panoramic engineering data to obtain data to be rendered; and performing hierarchical rendering processing on the data to be rendered to obtain target panoramic engineering data. This improves data loading speed, enhances terminal display smoothness, increases panoramic engineering visualization efficiency, improves the comprehensiveness of information display, enhances interactivity, and increases the richness and accuracy of annotation information.

The application discloses a kind of based on parameter updating's component rendering method, comprising: in response to the parameter classification set of the component of page label, determine the listening parameter set of component;Judge whether the configuration operation is carried out to listening parameter set;If yes, in response to the configuration operation of target parameter in listening parameter set, judge whether the parameter value of target parameter before and after configuration operation is consistent;If not consistent, determine target component according to listening parameter set, and store configuration parameter value in local storage;According to configuration parameter value, render target component in local cache, and cache rendered target component in local cache.The application effectively alleviates the pressure of client network bandwidth and interface resources, and the server does not need to frequently process parameter query and data return logic, reduces the concurrent processing pressure of server.The application also discloses a device, electronic equipment and storage medium for implementing the above method.

This application relates to a method, apparatus, electronic device, and storage medium for merging nodes in a layered and segmented 3D model. The method includes obtaining the root nodes to be merged; determining the overall bounding box to be merged and its major axis, width axis, and height axis based on the root nodes; dividing the overall bounding box along the major and width axes to obtain segmented grids; obtaining the segmentation vertices formed by the segmented grids along the major and width axes; constructing a 3D mesh scene including a preset number of root nodes; extending vertex rays from the segmented vertices along the height axis; and determining the target multi-level of detail based on the segmented vertices, the 3D mesh scene, and the vertex rays, and merging the root nodes to be merged into the target level root nodes. This application solves the problem of maintaining rendering quality while reducing the number of nodes required to initially display the entire layered and segmented 3D model, improving the loading speed and display smoothness of layered and segmented 3D models such as oblique photogrammetry models.

The application provides a position adjustment method and device of an interface element, equipment, a medium and a program product, and relates to the technical field of Internet. The method comprises the following steps: obtaining an initial position of an interface element in an interface according to a pre-trained deep learning model and an interface element sequence, wherein the initial position comprises a stacking attribute, and the stacking attribute is used for indicating a stacking order of the interface element in the interface; and in the case that the interface is in an editing mode, the initial position is adjusted through a position adjustment event to obtain a target position of a target interface element in the interface. According to the scheme, an accurate initial page layout can be obtained, so that accurate element position adjustment is realized based on the initial page layout, and user experience is improved.