GPU Virtualization Now Available on NVIDIA Consumer Graphics Cards (3 Photos)

NVIDIA graphics cards have undergone anotheramateur modification. After breaking the mining limitation, enthusiasts have now been able to unlock the GPU virtualization (vGPU) capability. Previously, vGPU was only supported on select data centers and professional boards (Grid, Tesla, and Quadro devices).

VGPU function on standard consumerNvidia GeForce gaming graphics cards are limited by software, but supported by hardware. Therefore, unlocking vGPU was only a matter of time. It was reported on Reddit that unlocking GPU virtualization could save thousands of dollars for users looking for this feature. These users will no longer be able to buy expensive Quadro or Tesla GPUs. The technology is known from the GeForce NOW platform, on which NVIDIA used GRID processors.

VGPU function allows multiple usersuse the GPU at the same time. This is one of the main differences between data center GPUs and GPUs designed for consumer computers. The "vgpu_unlock" modification will help activate vGPU for consumer GPUs.

The vGPU unlock code is available on Github.To implement GPU virtualization, a simple principle is used that replaces the device ID of a video card with the device ID of an officially supported GPU with the same feature set. Graphics cards with GPUs based on Pascal, Turing and Ampere architectures are referred to as Quadro or Tesla graphics cards. Currently supported GPUs are GP102, GP104, TU102, TU104 and GA102 and this feature works on Linux and with KVM virtual machine software. A detailed list of video cards supporting this feature is presented on Github.

Linux users can virtualizehigh-performance video cards, and use them for games, video encoding and cryptocurrency mining simultaneously on different virtual machines. However, this feature does not work with Windows and Vmware, making it useless for most consumers.

Source: videocardz