This page has been translated automatically.
Video Tutorials
Interface
Essentials
Advanced
How To
Basics
Rendering
Professional (SIM)
UnigineEditor
Interface Overview
Assets Workflow
Version Control
Settings and Preferences
Working With Projects
Adjusting Node Parameters
Setting Up Materials
Setting Up Properties
Lighting
Sandworm
Using Editor Tools for Specific Tasks
Extending Editor Functionality
Built-in Node Types
Nodes
Objects
Effects
Decals
Light Sources
Geodetics
World Nodes
Sound Objects
Pathfinding Objects
Players
Programming
Fundamentals
Setting Up Development Environment
Usage Examples
C++
C#
UnigineScript
UUSL (Unified UNIGINE Shader Language)
Plugins
File Formats
Materials and Shaders
Rebuilding the Engine Tools
GUI
Double Precision Coordinates
API
Animations-Related Classes
Containers
Common Functionality
Controls-Related Classes
Engine-Related Classes
Filesystem Functionality
GUI-Related Classes
Math Functionality
Node-Related Classes
Objects-Related Classes
Networking Functionality
Pathfinding-Related Classes
Physics-Related Classes
Plugins-Related Classes
IG Plugin
CIGIConnector Plugin
Rendering-Related Classes
VR-Related Classes
Content Creation
Content Optimization
Materials
Material Nodes Library
Miscellaneous
Input
Math
Matrix
Textures
Art Samples
Tutorials

NVIDIA

The set of samples demonstrates how to interact with an external graphics API via Unigine API (ResourceExternalMemory and ResourceFence). Although the samples show interaction with NVIDIA CUDA Toolkit, you may use them as the basis for integrating the Engine GAPI (Vulkan, DX11, DX12) with other GAPI (OpenGL, DX, CUDA, etc.) that can accept the handle and import a synchronization primitive (such as external semaphore in CUDA).

This approach allows interaction of various APIs using different GAPI, such as interaction of MediaFoundation using DX11 with the Engine using DX12 or Vulkan.

The samples also illustrate how to take the data directly from the video memory without CPU roundtrip and processing it using non-engine graphics API.

The difference in GPU -> CPU texture transfer performance made via CUDA 12.3 and by the Engine are shown in the table below (tested on our working configurations). Average performance gain with CUDA comprised x10 to x24 depending on the viewport resolution.

You can check the results on your PC just by launching the CUDATextureTransfer sample and enabling the Profiler.

CUDAMeshDynamic#


Vertices of a shared dynamic mesh are processed on GPU using CUDA.

To build the sample, use CUDA Toolkit v12.3.

On Linux: after installing the CUDA Toolkit, make sure that the PATH environment variable includes /usr/local/cuda-12.2/bin.


SDK Path: <SDK_INSTALLATION>source\samples\3rdparty\CUDAMeshDynamic

CUDAStructuredBufferWrite#


Particles in a shared structured buffer are updated on GPU using CUDA, and rasterized manually using Unigine Material with Compute Shader.

To build the sample, use CUDA Toolkit v12.3.

On Linux: after installing the CUDA Toolkit, make sure that the PATH environment variable includes /usr/local/cuda-12.2/bin.


SDK Path: <SDK_INSTALLATION>source\samples\3rdparty\StructuredBufferWrite

CUDATextureTransfer#


A shared texture is processed with CUDA and copied from video memory to RAM into Unigine Image.

To build the sample, use CUDA Toolkit v12.3.

On Linux: after installing the CUDA Toolkit, make sure that the PATH environment variable includes /usr/local/cuda-12.2/bin.


SDK Path: <SDK_INSTALLATION>source\samples\3rdparty\CUDATextureTransfer

CUDATextureWrite#


A shared texture is processed with CUDA and displayed on the Unigine Object as a texture on a cube.

To build the sample, use CUDA Toolkit v12.3.

On Linux: after installing the CUDA Toolkit, make sure that the PATH environment variable includes /usr/local/cuda-12.2/bin.


SDK Path: <SDK_INSTALLATION>source\samples\3rdparty\CUDATextureWrite
Last update: 2024-11-25
Build: ()