3 ms·

> You just have to set it up to fully power down the GPU. Can you please tell us a bit more? I have a NVidia + Intel GPU. I was thinking about using the NVidi

by csdvrx 5y ago

> You just have to set it up to fully power down the GPU. Can you please tell us a bit more? I have a NVidia + Intel GPU. I was thinking about using the NVidia for KVM (the NVidia would be powered down when not using KVM), and the Intel for native linux display with Pop or Ubuntu 22 LTS as another attempt to use Wayland (since I read NVidia GPUs can cause many issues)

sudosysgen 5y ago
On modern Linux systems the way to do it is using PRIME render offload. That way, the Intel (or AMD in my case) integrated GPU will be used normally, with the Nvidia GPU fully powered down (you need a new enough card for this). Then, when you need it, you can use a command or environment variable or graphical launcher to signal to use the dGPU for offloading. The GPU will then power on and render to a virtual display, and the framebuffer will be copied to the framebuffer of the relevant application on-the-fly. Once no more applications are using the NVidia GPU, it will power down. This way, there is no need to use a KVM, or do PCI passthrough, or anything of the sort, it mostly "just works". If the (very small) overhead of passing through the iGPU is too much and your laptop supports it, you can also using MUX graphics and restart the DM with a config that will allow the laptop to rewire the screen into the dGPU, but I have never found this necessary or helpful.
- hedora 5y ago
  > Display is configurable from the base version 15.6” FHD (1920 x 1080) IPS up to 15.6” 4K (3840 x 2160) OLED with Dolby Vision HDR. > I had to choose the 15.6” FHD (1920x1080) Dolby Vision HDR screen, because 4K options are not supported on the integrated GPU. It’s a HDR400 WVA IPS display panel with 16:9 aspect ratio, 800:1 contrast ratio and 72% NTSC color gamut. Sigh. There's no way I would want an NVidia card in any laptop, especially not a linux one. Why must companies do stupid configuration tie ins like this? At least there's a workaround. Related: If you want a Dell laptop with Linux preinstalled, they probably won't let you max out the RAM and/or CPU.

by csdvrx 5y ago

sudosysgen 5y ago

On modern Linux systems the way to do it is using PRIME render offload. That way, the Intel (or AMD in my case) integrated GPU will be used normally, with the Nvidia GPU fully powered down (you need a new enough card for this). Then, when you need it, you can use a command or environment variable or graphical launcher to signal to use the dGPU for offloading. The GPU will then power on and render to a virtual display, and the framebuffer will be copied to the framebuffer of the relevant application on-the-fly. Once no more applications are using the NVidia GPU, it will power down. This way, there is no need to use a KVM, or do PCI passthrough, or anything of the sort, it mostly "just works". If the (very small) overhead of passing through the iGPU is too much and your laptop supports it, you can also using MUX graphics and restart the DM with a config that will allow the laptop to rewire the screen into the dGPU, but I have never found this necessary or helpful.

hedora 5y ago
> Display is configurable from the base version 15.6” FHD (1920 x 1080) IPS up to 15.6” 4K (3840 x 2160) OLED with Dolby Vision HDR. > I had to choose the 15.6” FHD (1920x1080) Dolby Vision HDR screen, because 4K options are not supported on the integrated GPU. It’s a HDR400 WVA IPS display panel with 16:9 aspect ratio, 800:1 contrast ratio and 72% NTSC color gamut. Sigh. There's no way I would want an NVidia card in any laptop, especially not a linux one. Why must companies do stupid configuration tie ins like this? At least there's a workaround. Related: If you want a Dell laptop with Linux preinstalled, they probably won't let you max out the RAM and/or CPU.