New Multiuser GPU Platform Announced By AMD
The CPU/GPU makers AMD has announced a new multi-user GPU platform. AMD’s multi-user GPU platform will compete with Nvidia’s GRID virtualization.
AMD is putting focus on hardware virtualization, where 15 users per card are allowed. That means a single GPU is shared between 15 different users. On hardware level, users will be able to access features like OpenCL, OpenGL and DirectX acceleration.
On the other hand, Nvidia is upgrading GRID to 2.0. According to a post by AMD, software virtualization has been limited for users wanting full GPU acceleration for Open CL.
Software virtualization has traditionally been a limiting factor for those who want to fully utilize GPU hardware acceleration for compute tasks under Open CL. With AMD’s implementation of the new Multiuser GPU, users are no longer as limited to what they can or can’t do in a virtualized environment. Users will have access to native AMD display drivers for OpenGL, DirectX®, and OpenCL acceleration, enabling them to work with few if any restrictions. Whatever users were able to do on a local machine they can now do using our new virtualization technology.
Here’s how Multi-user GPU fares against Nvidia GRID:
- Virtualization method – Hardware
- OpenCL™ 2.0 acceleration support – Yes
- Stable, predictable performance – Yes
- Dedicated share of local memory for increased security – Yes
- Maximum users per physical GPU – 15
- Virtualization method – Software
- OpenCL™ 2.0 acceleration support – No
- Stable, predictable performance – No
- Dedicated share of local memory for increased security – No
- Maximum users per physical GPU – 8
You can read more about AMD Multi-user GPU, here.