Meta GPU is a virtual GPU technology, which can improve the rendering power of Ethanim in metaverse applications.
The workflow of a typical GPU device is:
- The application layer calls one of the APIs supported by the GPU, such as OpenGL or CUDA
- The OpenGL or CUDA library submits a workload to the KMD (Kernel Mode Driver) via the UMD (User Mode Driver)
- KMD writes CSR MMIO and submits it to the GPU hardware
- GPU hardware starts working... When done, DMA to memory, sends interrupt to CPU
- CPU locates the interrupt handler -- which KMD previously registered with OS Kernel -- and calls it
- The interrupt handler finds which workload has been executed and... Finally, the driver wakes up related applications
Meta GPU implements GPU virtualization by virtualizing the OpenGL API, and is currently the most widely used GPU virtualization solution in the industry. Its benefits contain:
- Flexible. 1:N of N. The software can decide how much it wants to set; which VM has higher priority and which VM has lower priority, same.
- It is not dependent on the GPU hardware vendor. Microsoft, VMWare, Citrix, Huawei ... can all implement. These APIs are always public.
- Not limited to system virtualization environments. Containers or regular physical machines can be forwarded to the remote side by the APIs.
Meta GPU (vGPU) shares GPUs between virtual machines (VMs) and virtual applications, providing a superior experience for every remote user, every application, and any device. Meta GPU software products include GRID Virtual PC (GRID vPC), GRID Virtual Application (GRID vApp), and Quadro Virtual Data Center Workstation (Quadro vDWS). All products are designed to deliver a higher level of performance, flexibility, manageability and security to bring the appropriate user experience to every virtual workflow.