A multi-GPU city simulation system for large scenes

Through multi-GPU parallel real-time rendering technology and DirectX11 API optimization strategy, the problems of cross-screen pickup and full-screen operation in multi-channel large scenarios are solved, and an efficient three-dimensional urban simulation system is realized, which improves video frame synchronization performance and rendering effect.

CN112001993BInactive Publication Date: 2024-12-03深圳市规划国土房产信息中心
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202010678751.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-07-14
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing three-dimensional real-time multi-GPU urban simulation system for large scenarios lacks cross-screen pickup and full-screen operation functions in multi-channel large scenarios, and is restricted by Ethernet network communication, making video frame synchronization performance difficult to improve, and multi-channel three-dimensional urban real-time simulation performance Limited room for improvement.

Method used

Using multi-GPU parallel real-time rendering technology, the main module, processing module and rendering module are connected to each other to realize system configuration, resource management and rendering output, combined with virtual windows for multi-channel output, and through multi-threaded model, terrain loading and rendering optimization Strategy, optimize rendering performance using DirectX11 API and surface segmentation algorithm.

Benefits of technology

Realize real-time rendering of large-scale three-dimensional terrain data and huge amounts of fine building model data, supports a variety of real-time urban simulation effects, improves cross-screen pickup and full-screen operation performance, improves video frame synchronization performance, and meets high dynamic range lighting. Rendering requirements for effects such as real-time shadows and real light reflections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112001993B_ABST
    Figure CN112001993B_ABST
Patent Text Reader

Abstract

The present invention discloses a multi-GPU city simulation system for large scenes, based on multi-GPU parallel real-time rendering, a main module for completing system configuration, system initialization, thread initialization, user interoperability control, terrain management and resource management; a processing module for updating and processing resources stored in main memory and video memory; a rendering module for completing the loading and rendering of video memory resources, and performing multi-channel output of simulation scenes through virtual windows; wherein the main control module, processing module and rendering module are interconnected. The present invention is based on multi-GPU parallel real-time rendering, and can be applied to multi-GPU city three-dimensional simulation engines for ultra-large scenes, realizing the characteristics of real-time rendering of large-scale three-dimensional terrain data and massive fine building model data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a multi-GPU city simulation system for large scenes. Background Art

[0002] The application environment of the 3D real-time large-scene multi-GPU city simulation system is mainly large conference rooms, lecture halls, exhibition halls and other large places. Generally, users use curved multi-channel projection screens and multi-channel rear-projection splicing display systems for display. For most users, city-level high-precision DEM / DOM 3D terrain data and fine building model data are very huge. If you also want the simulation system to support 3D simulation technologies such as real-time lighting system, high dynamic range lighting (HDR), real-time shadow system, 3D plant system, atmospheric scattering effect, dynamic water surface with real light reflection effect and animation system for real-time rendering and display, only by using the parallel rendering method of multiple GPU-CPU can sufficient computing and rendering capabilities be provided for city simulation.

[0003] However, the existing 3D real-time large-scale multi-GPU city simulation system lacks cross-screen picking and full-screen operation functions in multi-channel large scenes. At the same time, due to the constraints of Ethernet network communication, the video frame synchronization performance between channels is difficult to improve, and the performance improvement space of multi-channel 3D city real-time simulation is limited. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-GPU city simulation system for large scenes in view of the defects of the prior art.

[0005] In order to achieve the above-mentioned purpose of the present invention, the technical solution adopted is:

[0006] A multi-GPU city simulation system for large scenes is based on multi-GPU parallel real-time rendering, including a main module for completing system configuration, system initialization, thread initialization, user interaction control, terrain management and resource management; a processing module for updating and processing resources stored in main memory and video memory; a rendering module for completing the loading and rendering of video memory resources and performing multi-channel output of simulation scenes through a virtual window; wherein the main control module, the processing module and the rendering module are interconnected.

[0007] Furthermore, the process of the main module includes:

[0008] Detect the number of GPUs in the hardware system, identify the device identification number of each GPU, and detect the computing power of each GPU; calculate the number of projectors and the size of the overlapping area of ​​the projection plane edge;

[0009] Detect the host memory size to prepare for the data resource memory management and scheduling module; detect the capabilities of various input control devices on the host and configure various input devices for the software system;

[0010] Detect software function configuration files, initialize software system function configuration data structure, and configure software operation interface;

[0011] After completing various detection and initialization operations, notify the establishment of CPU scheduling thread and GPU screen splicing thread;

[0012] According to user interaction, pass user control parameters to the CPU scheduling thread and GPU rendering thread;

[0013] If the operation is not terminated, continue to monitor user interaction.

[0014] Furthermore, the process of CPU scheduling threads includes:

[0015] According to the main program notification, an independent resource scheduling main thread is established for each CPU;

[0016] Calculate the viewing area of ​​each frame based on user interaction;

[0017] Select different parallel rendering strategies and enable different scheduling processes. For example, logically subdivide the entire projection screen into multiple independent graphics rendering windows and allocate them to each GPU according to its computing power;

[0018] Process terrain data and terrain texture according to the viewing area;

[0019] According to different scheduling strategies, the update loads the hard disk resources into the memory buffer, or loads the resources from the memory into the video memory;

[0020] Updates the display memory according to different scheduling strategies.

[0021] Furthermore, the process of the GPU screen splicing thread includes:

[0022] Select a GPU and create a screen stitching thread;

[0023] Before rendering each frame, create and clear the virtual window;

[0024] Receive rendering results from multiple GPUs and synthesize the segmented images into a completed frame image in a virtual window;

[0025] According to the number of projectors and the size of the overlapping area at the edge of the projection plane, the image is divided into multiple channels for output.

[0026] The GPU rendering thread includes an immediate rendering thread and a delayed rendering thread.

[0027] Furthermore, the immediate rendering thread includes:

[0028] Create an immediate rendering thread for each GPU;

[0029] Calculate rendering parameters and shadow effects;

[0030] Perform frustum culling on the current GPU-allocated visible area;

[0031] Perform an immediate rendering operation.

[0032] Furthermore, the delayed rendering thread includes:

[0033] Create multiple delayed rendering threads for each GPU;

[0034] Calculate the delay rendering constant;

[0035] Perform frustum culling on the current GPU-allocated visible area;

[0036] Perform parallel rendering operations on objects or special effects;

[0037] The rendering results are stored in the GPU Buffer.

[0038] A multi-threaded model optimization strategy for the multi-GPU city simulation system for large scenes is as follows:

[0039] Let Tf be the data acquisition time, Tp be the data preprocessing time, Ts be the data transmission time to GPU, and Tr be the GPU rendering time;

[0040] Assuming that there are m available processors, the best performance can be obtained when Tf+Tp=Ts; when Ts≥Tr, a balance can be achieved between data loading and rendering processes by increasing the number of processors.

[0041] A terrain loading and rendering optimization strategy for the multi-GPU city simulation system for large scenes is specifically to use a quadtree refinement level LOD processing algorithm in conjunction with the surface subdivision of DirectX11 API to optimize terrain rendering.

[0042] Use DirectX11's texture array to load four-level terrain LOD data;

[0043] Receive a primitive data obtained by calculating the full quad mesh through the DirectX11 shell shader, and calculate various transformations of the control points and the subdivision configuration of each edge of the input primitive to perform surface subdivision;

[0044] The calculation control points are used to define the graphic parameters of the desired graphic.

[0045] A parallel rendering optimization strategy for the multi-GPU city simulation system for large scenes is specifically to optimize the engine rendering strategy by utilizing the multi-threaded processing characteristics of DirectX11, combining CPU resource scheduling, using screen subdivision and delayed rendering.

[0046] Compared with the prior art, the present invention has outstanding substantive features and significant progress: the present invention is based on multi-GPU parallel real-time rendering, is a multi-GPU urban 3D simulation engine suitable for ultra-large scenes, realizes real-time rendering of large-scale 3D terrain data and massive fine building model data, and supports a variety of real-time urban simulation effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0048] Figure 1 A schematic diagram of a multi-GPU city simulation system for large scenes according to the present invention;

[0049] Figure 2 A schematic diagram of the core operation of the present invention;

[0050] Figure 3 It is the DEM digital elevation replacement map of the present invention;

[0051] Figure 4 It is a schematic diagram of the demonstration of the DOM and DEM quadtree algorithms of different resolutions of the present invention;

[0052] Figure 5 This is the effect diagram of the curved projection screen;

[0053] Figure 6 is a schematic diagram of the performance analysis interface of the present invention;

[0054] Figure 7 It is the effect diagram of the multi-channel three-dimensional scene of the present invention;

[0055] Figure 8 A schematic diagram of multi-GPU performance test results. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present application.

[0057] Embodiment 1:

[0058] like Figures 1 to 8 As shown in the figure, a multi-GPU city simulation system for large scenes is based on multi-GPU parallel real-time rendering, including:

[0059] The main module is used to complete system configuration, system initialization, thread initialization, user interaction control, terrain management and resource management; the processing module is used to update and process the resources stored in the main memory and video memory; the rendering module is used to complete the loading and rendering of video memory resources and perform multi-channel output of simulation scenes through virtual windows; wherein the main control module, processing module and rendering module are interconnected.

[0060] The main modules include hardware monitor, interoperability controller, resource manager and terrain manager modules.

[0061] The processing module has an I / O thread and a processing thread. The processing module is a CPU processing module.

[0062] Rendering module, including resource thread and rendering thread.

[0063] In some embodiments, the process of the main module includes: detecting the number of GPUs in the hardware system, identifying the device identification number of each GPU, and detecting the computing power of each GPU; calculating the number of projectors and the size of the overlapping area of ​​the edge of the projection plane; detecting the host memory size to prepare for the data resource memory management and scheduling module; detecting the capabilities of various input control devices of the host, and configuring various input devices for the software system; detecting the software function configuration file, initializing the software system function configuration data structure, and configuring the software operation interface; after completing various detection and initialization operations, notifying the establishment of a CPU scheduling thread and a GPU screen splicing thread; based on user interaction, passing user control parameters to the CPU scheduling thread and the GPU rendering thread; if the operation is not terminated, continuously listening to user interaction.

[0064] In some embodiments, the process of CPU scheduling thread includes: scheduling the main thread for independent resources established for each CPU according to the notification of the main program; calculating the view area of ​​each frame according to user interaction; selecting different parallel rendering strategies and enabling different scheduling processes. For example, the entire projection screen is logically subdivided into multiple independent graphics rendering windows, and each GPU is reasonably allocated according to the computing power of each GPU; processing terrain data and terrain texture according to the view area; updating and loading hard disk resources into memory buffers or loading resources from memory into video memory according to different scheduling strategies; updating display memory according to different scheduling strategies.

[0065] In some embodiments, the process of the GPU screen splicing thread includes: selecting a GPU and establishing a screen splicing thread; before rendering each frame, establishing and clearing a virtual window; receiving rendering results from multiple GPUs, and synthesizing the segmented images into a completed frame image in the virtual window; and segmenting the image into multiple channels for output according to the number of projectors and the size of the overlapping area of ​​the projection plane edge.

[0066] In some embodiments, the GPU rendering thread includes an immediate rendering thread and a delayed rendering thread.

[0067] The immediate rendering thread includes: establishing an immediate rendering thread for each GPU; calculating rendering parameters and shadow effects;

[0068] Assign the current GPU a visible area for frustum culling; perform immediate rendering operations.

[0069] The delayed rendering thread includes: establishing multiple delayed rendering threads for each GPU; calculating delayed rendering constants; allocating a visible area to the current GPU for frustum pruning; performing parallel rendering operations on objects or special effects; and storing rendering results in GPU Buffer.

[0070] In some embodiments, a multi-threaded model optimization strategy according to the above-mentioned multi-GPU city simulation system for large scenes is specifically:

[0071] Let Tf be the data acquisition time, Tp be the data preprocessing time, Ts be the data transmission time to GPU, and Tr be the 6PU rendering time;

[0072] Assuming that there are m available processors, the best performance can be obtained when Tf+Tp=Ts; when Ts≥Tr, a balance can be achieved between data loading and rendering processes by increasing the number of processors.

[0073] The multi-threaded model adopts an on-demand allocation mechanism. The main thread manages the task queue and establishes a sub-thread for each available processor. When the sub-thread completes the processing task, it obtains the next processing task from the task queue. The resource scheduling of each GPU is controlled by an independent CPU subroutine, and each subroutine starts a main thread and several sub-threads. The I / O thread processes requests from the rendering thread. It not only loads data from the disk into the memory buffer, but also updates the data in the memory buffer to the GPU video memory; the I / O thread communicates with various data processing threads and processes the "locking" or "unlocking" request queues "consumed" by the rendering thread. The multi-threaded model is based on the characteristics of the Fermi architecture, making full use of the computing power of multiple CPUs to ensure that the rendering time of multiple GPUs is not limited by resource processing and transmission time. As shown in Table 1.

[0074] Table 1. Thread classification and usage

[0075]

[0076] In some embodiments, a terrain loading and rendering optimization strategy based on the above-mentioned large-scene oriented multi-GPU city simulation system is specifically to use a quadtree refinement level LOD processing algorithm in conjunction with the surface subdivision of the DirectX11 API to optimize terrain rendering.

[0077] like Figure 3 and 4 As shown, the texture array of DirectX11 is used to load four levels of terrain LOD data, among which, the finest layer: 1 piece of terrain → 1 texture (4096*4096), the second level: 1 piece of terrain → 4 textures (2048*2048 texture array), the third level: 1 piece of terrain → 16 textures (1024*1024 texture array), the coarsest layer: 1 piece of terrain → 64 textures (512*512 texture array);

[0078] Receive a primitive data obtained by calculating the full quad mesh through the DirectX11 shell shader, and calculate various transformations of the control points and the subdivision configuration of each edge of the input primitive to perform surface subdivision;

[0079] The calculation control points are used to define the graphic parameters of the desired graphic.

[0080] Each layer of fine-scale layered DEM digital elevation map is loaded into the GPU as a texture displacement map resource, and introduced into the primitive as the intermediate vertex data of the vertex at this level, which can produce very delicate and realistic three-dimensional terrain at a very low cost, and the accuracy of DEM digital elevation is not limited in any way. The Tessellator unit can divide some larger primitives into many smaller primitives, and combine these small primitives together to form an orderly geometric figure, which is more complex and of course closer to reality.

[0081] For DOM aerial film texture data with LOD, we use a multi-CPU threaded producer / consumer software model for real-time data update. The coarsest first-level DOM aerial film texture is loaded into the DOM memory buffer at once, and the remaining three-level DOM aerial film texture data is dynamically loaded into the DOM memory buffer according to the viewpoint position and viewing distance. According to the need for viewpoint changes, starting from the periphery of the DOM aerial film texture data, the DOM aerial film texture data in the disk is dynamically updated to the DOM memory buffer; according to the viewpoint position of different GPUs, the texture data in the DOM memory buffer is dynamically updated to the video memory of multiple graphics cards.

[0082] In some embodiments, a parallel rendering optimization strategy for the above-mentioned large-scene multi-GPU city simulation system is specifically to optimize the engine rendering strategy by utilizing the multi-threaded processing characteristics of DirectX11, combining CPU resource scheduling, using screen subdivision and delayed rendering.

[0083] The purpose of screen segmentation is to ensure that all GPUs have sufficient rendering tasks. The entire screen is divided according to the number of GPUs, and each GPU performs real-time rendering and pixel depth testing on all scene contents in a segmented screen. Then, all intermediate image results are output to a unified pixel intermediate window for image frame buffer stitching, and finally multi-channel output.

[0084] Delayed rendering is a new graphics software development technology added by DirectX 11. It can create multiple rendering threads in a GPU to perform various rendering tasks respectively. Finally, the immediate rendering thread executes GPU operations, which can effectively improve the real-time rendering performance of the engine. The present invention has multiple delayed rendering threads, as shown in Table 2. In the hybrid rendering mode, the cone is cut for an immediate rendering thread and multiple delayed rendering threads to minimize the number of real-time rendering of GPU triangles. The viewpoint position performs collision detection in real time to avoid wall penetration and ground contact operations.

[0085] Table 2 Delayed rendering thread usage

[0086]

[0087] The present invention compares task subdivision rendering and screen subdivision rendering in an actual environment. The task subdivision rendering method is to divide the rendering objects in the three-dimensional rendering scene by type, load the rendering objects into different GPUs for real-time rendering, and then output all intermediate image results into a unified intermediate window for image synthesis according to pixel depth.

[0088] Table 3 Comparison of task segmentation and window segmentation rendering

[0089]

[0090] The real-time rendering of a 3D city requires that its video output frame rate is ≥25 frames / s, that is, the time for all real-time rendering must be ≤40ms. For task subdivision rendering, the pixel color data and pixel depth data of multiple 1400*2730 windows need to be copied to a unified 1400*2730 pixel intermediate window through the north bridge chip and PCI-E×16 bus. The data transmission volume is large, and half of the hardware overhead time is consumed by bus data transmission. Screen subdivision rendering requires the pixel color data of 6 1400*390 subdivision windows to be copied to a unified 1400*2730 pixel intermediate window through the north bridge chip and PCI_E x16 bus. Compared with task subdivision rendering, the biggest feature of screen subdivision rendering is that the data transmission volume in the bus is reduced, and the hardware overhead is reduced.

[0091] like Figure 7 As shown in the figure, the engine loads the DEM and DOM data and 3D model of Shenzhen City, selects a large scene with a viewing distance of 20 kilometers for testing, and turns on all special effects. This scene renders about 80 million polygons per frame at the same time, which consumes a lot of CPU and GPU resources.

[0092] like Figure 8 The figure shows the comparison between the default task subdivision rendering strategy of DirectX11 and the rendering strategy optimized in this paper under different numbers of GPUs. It can be clearly seen from the figure that for the method in this paper, when the number of GPUs increases, the number of rendered frames of the scene increases steadily. However, the default rendering strategy cannot give full play to the performance of multiple GPUs because it is not optimized.

[0093] The present invention aims at the requirements of real-time rendering of massive fine building model data, cross-screen picking of multi-channel large scenes, full-screen operation and other functions, optimizes screen refinement rendering, real-time data transmission management, editable urban terrain, etc., and realizes effects such as real-time lighting system, high dynamic range lighting (HDR), real-time shadow system, three-dimensional plant system, atmospheric scattering effect, dynamic water surface with real light reflection effect, etc. in a large-scale urban three-dimensional simulation scene.

[0094] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the protection scope of the invention.

Claims

1. A multi-GPU city simulation system for large scenes, characterized by: Based on multi-GPU parallel real-time rendering, it includes a main module for completing system configuration, system initialization, thread initialization, user interaction control, terrain management and resource management; A processing module, used for updating and processing resources stored in the main memory and the video memory; The rendering module is used to load and render video memory resources and perform multi-channel output of simulation scenes through virtual windows; Wherein, the main module, processing module and rendering module are interconnected; The process of the main module includes: detecting the number of GPUs in the hardware system, identifying the device identification number of each GPU, and detecting the computing power of each GPU; calculating the number of projectors and the size of the overlapping area of ​​the projection plane edge; Detect the host memory size to prepare for the data resource memory management and scheduling module; Detect the capabilities of various input control devices on the host and configure various input devices for the software system; Detect software function configuration files, initialize software system function configuration data structure, and configure software operation interface; After completing various detection and initialization operations, notify the establishment of CPU scheduling thread and GPU screen splicing thread; According to user interaction, pass user control parameters to the CPU scheduling thread and GPU rendering thread; If the operation is not terminated, continue to monitor the user's interaction; The process of the GPU screen splicing thread includes: selecting a GPU and establishing a screen splicing thread; before rendering each frame, establishing and clearing a virtual window; receiving rendering results of multiple GPUs, and synthesizing the segmented images into a completed frame image in the virtual window; and segmenting the image into multiple channels for output according to the number of projectors and the size of the overlapping area of ​​the projection plane edge; The GPU rendering thread includes an immediate rendering thread and a delayed rendering thread; The immediate rendering thread includes: establishing an immediate rendering thread for each GPU; calculating rendering parameters and shadow effects; Perform frustum culling on the visible area allocated to the current GPU; perform immediate rendering operations; The delayed rendering thread includes: establishing multiple delayed rendering threads for each GPU; calculating a delayed rendering constant; Assign the visible area to the current GPU for frustum culling; perform parallel rendering operations on objects or special effects; and store the rendering results in the GPU Buffer.

2. The multi-GPU city simulation system for large scenes according to claim 1 is characterized in that: The process of CPU scheduling thread includes: According to the main program notification, an independent resource scheduling main thread is established for each CPU; Calculate the viewing area of ​​each frame based on user interaction; Select different parallel rendering strategies and enable different scheduling processes: specifically, logically subdivide the entire projection screen into multiple independent graphics rendering windows, and allocate them to each GPU according to the computing power of each GPU; process terrain data and terrain textures according to the viewing area; update and load hard disk resources into the memory buffer, or load resources from the memory to the video memory, according to different scheduling strategies; update the display memory according to different scheduling strategies.

3. The multi-GPU city simulation system for large scenes according to claim 1 is characterized in that: The process of the GPU screen splicing thread includes: selecting a GPU and establishing a screen splicing thread; before rendering each frame, establishing and clearing a virtual window; receiving rendering results of multiple GPUs, and synthesizing the segmented images into a completed frame image in the virtual window; and segmenting the image into multiple channels for output according to the number of projectors and the size of the overlapping area of ​​the projection plane edge.

4. The multi-GPU city simulation system for large scenes according to claim 1 is characterized in that: The GPU rendering thread includes an immediate rendering thread and a delayed rendering thread.

5. The multi-GPU city simulation system for large scenes according to claim 4 is characterized in that: The immediate rendering thread includes: establishing an immediate rendering thread for each GPU; calculating rendering parameters and shadow effects; allocating a visible area to the current GPU to perform cone cutting; and executing an immediate rendering operation.

6. The multi-GPU city simulation system for large scenes according to claim 4 is characterized in that: The delayed rendering thread includes: establishing multiple delayed rendering threads for each GPU; calculating delayed rendering constants; allocating a visible area to the current GPU for frustum pruning; performing parallel rendering operations on objects or special effects; and storing rendering results in GPU Buffer.

7. A multi-threaded model optimization method for a large-scale multi-GPU city simulation system according to any one of claims 1 to 6, characterized in that: Let Tf be the data acquisition time, Tp be the data preprocessing time, Ts be the data transmission time to the GPU, and Tr be the GPU rendering time; assume that there are m available processors, and the best performance can be obtained when Tf+Tp=Ts; when Ts≥Tr, a balance can be achieved between data loading and rendering processes by increasing the number of processors.

8. A terrain loading and rendering optimization method for a large-scale multi-GPU city simulation system according to any one of claims 1 to 6, characterized in that: A quadtree refinement level LOD processing algorithm is used in conjunction with DirectX11 API surface subdivision to optimize terrain rendering; DirectX11 texture array is used to load four-level terrain LOD data; a graphics element calculated from a full quadrilateral mesh is received through a DirectX11 shell shader, and various transformations of control points and subdivision configurations of each edge of the input graphics element are calculated, thereby performing surface subdivision; wherein the calculated control points are used to define the graphics parameters of the desired graphics.

9. A parallel rendering optimization method for a multi-GPU city simulation system for a large scene according to any one of claims 1 to 6, characterized in that: By utilizing the multi-threaded processing features of DirectX11, combined with CPU resource scheduling, screen subdivision and deferred rendering are used to optimize the engine rendering strategy.

Citation Information

Patent Citations

  • Multi-GPU large-resolution multi-screen graphics block parallel rendering method

    CN107958437A

  • Block-based fragment filtration with feasible multi-GPU acceleration for real-time volume rendering on standard PC

    WO2005104042A1