A work-in-progress project dedicated to making GPU Paravirtualization on Windows Hyper-V easier!
GPU-PV allows you to partition your systems dedicated or integrated GPU and assign it to several Hyper-V VMs. It's the same technology that is used in WSL2, and Windows Sandbox.
Simple-GPU-P aims to make this easier by automating the steps required to get a GPU-PV VM up and running. Simple-GPU-P does the following...
- Allows managing GPUs in your VM
- Adds GPU to existing VM
- Updates Drivers in VM
- Removes GPUs from VM
- Changes allocation of your GPU
- Windows 10 20H1+ Pro, Enterprise or Education OR Windows 11 Pro, Enterprise or Education. Windows 11 on host and VM is preferred due to better compatibility.
- Latest GPU driver from Intel.com or NVIDIA.com, don't rely on Device manager or Windows update.
- Virtualisation enabled in the motherboard and Hyper-V fully enabled on the Windows 10/ 11 OS (requires reboot).
- Allow Powershell scripts to run on your system - typically by running "Set-ExecutionPolicy unrestricted" in Powershell running as Administrator.
- Make sure your system meets the prerequisites.
- Download files from this repository
- Use Generation 2 of VM
- Install Windows on your VM
- DISABLE dynamic memory
- DISABLE automatic snapshots
- Run
RUN-ME.BAT
as Administrator - Follow the steps in command prompt
- In Hyper-V settings enable enhanced session to add sound support
- Hyper-ConvertImage for creating an updated version of Convert-WindowsImage that is compatible with Windows 10 and 11.
- gawainXX for help testing and pointing out bugs and feature improvements.
- Your GPU on the host will have a Microsoft driver in device manager, rather than an nvidia/intel/amd driver. As long as it doesn't have a yellow triangle over top of the device in device manager, it's working correctly.
- If you do not have administrator permissions on the machine it means you set the username and vmname to the same thing, these needs to be different.
- AMD Polaris GPUS like the RX 580 do not support hardware video encoding via GPU Paravirtualization at this time.
- Linux Virtual Guests are not supported yet
- Original Project is available here.