GPU 硬件虚拟化Hardware Virtualization
GPU 硬件虚拟化Hardware Virtualization
1. Principles
一个物理GPU可以虚拟化为多个vGPUs。VMs可以绑定到vGPUs以直接访问一些物理GPU资源。
2. Features
- 基于NVIDIA网格卡提供GPU虚拟化,提高图形应用体验。GPU virtualization is provided based on NVIDIA GRID cards to improve the graphics application experience.
- vGPU资源管理和调度支持GPU负载平衡。vGPU resource management and scheduling enable GPU load balancing.
- 支持OpenGL和DirectX多媒体编程接口。Multimedia programming interfaces OpenGL and DirectX are supported.
- 支持Aero特效、多监视器和DirectX视频加速(DXVA)。Aero special effect, multi-monitor, and DirectX Video Acceleration(DXVA) are supported.
3. 应用场景Application Scenarios
- Engineering drawings(such as ProE, Catia, and AutoCAD), media production, 3D games, and GIS applications running in the virtualization environment.
在虚拟化环境中运行的工程图(如ProE、Catia和AutoCAD)、媒体制作、3D游戏和GIS应用程序。
- Industrial design, multimedia editing, energy, financial services and trade, medical imaging system, and educational best practices. 工业设计、多媒体编辑、能源、金融服务和贸易、医疗成像系统和教育最佳实践。
4. GPU硬件虚拟化解决方案的技术原理。Technical Principles of the GPU Hardware Virtualization Solution.
- GPU hardware virtualization adopts a single graphics card to provide graphics capabilities for multiple graphics desktops. GPU硬件虚拟化采用单个图形卡为多个图形桌面提供图形功能。
人工智能芯片与自动驾驶