Unlike what others say, a bog standard VM might be the wrong choice depending on which features you use due to lack of graphics acceleration in said VM.
You might be able to get GPU passthrough working on a VM, which I have, with both a Windows and macOS VM that can use the GPU (not at the same time); however, this is really complex (took me ages the first time, though I’ve since discovered tricks to make it a bit easier), and you have to have dual GPUs. Single GPU passthrough is technically possible, but then you can’t use your Linux DE while using the VM. I will say, though, that once it’s set up, it’s a better experience than dual-booting; you get to run graphics-intensive Windows apps quite snappily on one monitor (or monitor input) and use your Linux desktop on the other.
I technically have 2 GPUs in my laptop, one integrated into intel CPU and one discrete nvidia (so called “optimus”). Is it possible to pass nvidia while keeping intel for linux DE?
I tried to read some tutorials but tbh they all leave more questions than give answers.
It might be possible, depending on if the screen is connected to the dGPU or iGPU (I’d guess iGPU). I wouldn’t know because I did my setup on a desktop with two dGPUs. I would think it’s possible, but you might need an external monitor (?). I don’t know how Optimus laptops are wired.
Where I started for GPU passthrough, which got me ~90% of the way there, is https://github.com/bryansteiner/gpu-passthrough-tutorial . Gives you the shell scripts, XML, etcetera needed to do it; I had to modify some bits (some of which you can see in issues), but this is my preferred tutorial. Basically, try it, get really frustrated, take a break for a while, get back to it and keep tinkering with it (check permissions, logs, PCIe driver binds, etcetera), and eventually, you’ll figure it out.
I might be able to send over some of my XML to get you started, but I don’t know how helpful that will actually be over the tutorial, as our systems are completely different, and the AMD GPU I use has different bugs/quirks when doing this than Nvidia ones. The truth of the matter on why there’s not really a single-click, easy way to do GPU passthrough is because each system is unique, from the motherboard PCIe implementation to bugs in GPU firmware. That doesn’t mean you shouldn’t try, but it takes a bit of ingenuity.
Unlike what others say, a bog standard VM might be the wrong choice depending on which features you use due to lack of graphics acceleration in said VM.
You might be able to get GPU passthrough working on a VM, which I have, with both a Windows and macOS VM that can use the GPU (not at the same time); however, this is really complex (took me ages the first time, though I’ve since discovered tricks to make it a bit easier), and you have to have dual GPUs. Single GPU passthrough is technically possible, but then you can’t use your Linux DE while using the VM. I will say, though, that once it’s set up, it’s a better experience than dual-booting; you get to run graphics-intensive Windows apps quite snappily on one monitor (or monitor input) and use your Linux desktop on the other.
Could you help me with GPU passthrough?
I technically have 2 GPUs in my laptop, one integrated into intel CPU and one discrete nvidia (so called “optimus”). Is it possible to pass nvidia while keeping intel for linux DE?
I tried to read some tutorials but tbh they all leave more questions than give answers.
It might be possible, depending on if the screen is connected to the dGPU or iGPU (I’d guess iGPU). I wouldn’t know because I did my setup on a desktop with two dGPUs. I would think it’s possible, but you might need an external monitor (?). I don’t know how Optimus laptops are wired.
Where I started for GPU passthrough, which got me ~90% of the way there, is https://github.com/bryansteiner/gpu-passthrough-tutorial . Gives you the shell scripts, XML, etcetera needed to do it; I had to modify some bits (some of which you can see in issues), but this is my preferred tutorial. Basically, try it, get really frustrated, take a break for a while, get back to it and keep tinkering with it (check permissions, logs, PCIe driver binds, etcetera), and eventually, you’ll figure it out.
https://github.com/mysteryx93/GPU-Passthrough-with-Optimus-Manager-Guide is linked in one of the issues and specifically concerns your kind of laptop.
I might be able to send over some of my XML to get you started, but I don’t know how helpful that will actually be over the tutorial, as our systems are completely different, and the AMD GPU I use has different bugs/quirks when doing this than Nvidia ones. The truth of the matter on why there’s not really a single-click, easy way to do GPU passthrough is because each system is unique, from the motherboard PCIe implementation to bugs in GPU firmware. That doesn’t mean you shouldn’t try, but it takes a bit of ingenuity.