You are not logged in.
I would like all of my Arch Linux DE to run on my integrated GPU and use my integrated GPU's VRAM instead of my RTX 3070's. I am not even sure if this is possible with this system setup. I have my displays plugged into my RTX 3070. My DE is KDE Plasma 6.2.5 using Wayland.
Last edited by Biod x (2025-01-08 16:34:01)
Offline
Have you tried connecting the display into the motherboard video port (probably try HDMI) and selecting igpu in bios when booting?
Offline
Have you tried connecting the display into the motherboard video port (probably try HDMI) and selecting igpu in bios when booting?
I tried setting it to IGD instead of PEG in my BIOs and plugged in my displays into my motherboard. When I boot I only have a single underscore on the top left of the screen after Arch Boots. I can enter tty when that happens. But it does not work exactly.
Last edited by Biod x (2025-01-07 22:21:39)
Offline
question: WHY do want to this?
do you have a multi-screen setup?
why do you can't/don't want to use your dgpu?
why have you built a system with both an igpu and dgpu in the first place?
sounds to me like an XY problem - anyway - if it all that's a setting for the systems uefi firmware - nothing to do with any OS
Offline
question: WHY do want to this?
do you have a multi-screen setup?
why do you can't/don't want to use your dgpu?
why have you built a system with both an igpu and dgpu in the first place?sounds to me like an XY problem - anyway - if it all that's a setting for the systems uefi firmware - nothing to do with any OS
Answer: More open vram space on my dGPU so that when I fill it which is usually 7gb from my applications. The +1gb of vram being used on my system is not causing my vram to flush to my system ram which then causes lots of lag in my applications and stutters because the system ram is slower then the vram. My solution is to use the rest of my system which is my integrated gpu which I have allocated vram for it via BIOs. Yes my setup is multi screen.
This is not a BIOs setup issue because I have not able to boot my Arch Linux to use my integrated gpu nor have I been able to allocate applications to use my integrated GPU even though I have the mesa drivers installed and I think I installed the right drivers for the integrated GPU which is f86 amd something. My Arch Linux can see my integrated GPU via terminal commands but I have not been able to use it in any way.
Offline
Won't help your use case but to boot on my igpu I can simply connect the motherboard video out, select it on the monitor, and select igpu in bios (after disconnecting power to the dGPU, with PSU off / disconnected obviously) - after that, arch linux boots normally.
If I leave power connected to the dGPU then the bios autoselect apparently chooses dGPU prior to the arch boot even if I select iGPU as primary GPU in bios (though I guess if I turn off autoselect this might avoid the need to disconnect the dGPU).
But to get to the point - looks like arch linux should handle the change without help, your bios might be the trouble after all.
Offline
long reply short: what you want to try here doesn't work for several reasons - if you run out of video ram you either need to adjust your workload to the available system resources or upgrade your hardware
Offline
Won't help your use case but to boot on my igpu I can simply connect the motherboard video out, select it on the monitor, and select igpu in bios (after disconnecting power to the dGPU, with PSU off / disconnected obviously) - after that, arch linux boots normally.
If I leave power connected to the dGPU then the bios autoselect apparently chooses dGPU prior to the arch boot even if I select iGPU as primary GPU in bios (though I guess if I turn off autoselect this might avoid the need to disconnect the dGPU).
But to get to the point - looks like arch linux should handle the change without help, your bios might be the trouble after all.
My BIOs are not the trouble because Arch Boots but the sddm does not display. I can enter and use tty. Also when my dGPU is used when booted from and I login via sddm. When I plug my displays into my motherboard it works and displays Arch Linux. But Arch is still using my 3070 for rendering the OS and programs.
long reply short: what you want to try here doesn't work for several reasons - if you run out of video ram you either need to adjust your workload to the available system resources or upgrade your hardware
No it would work properly because my applications that most of the time hover around +7gb of vram would be using my dGPU exclusivily and my integrated GPU which has 512mb of vram and I allocated 3gb of extra vram for my iGPU via dram. Which would make my dGPU application not have weird lag and stutters when it flushes from dGPU vram to dram because my application is only using +7gb of vram unlike with my OS using another +1gb which takes it over my dGPU's limit causing issues in my application.
Offline
Did you generate a xorg config with nvidia-xconfig? Remove that that will force xorg to load on the nvidia gpu despite you wanting to offload things to the iGPU
I still think you have too much hope for how this will alleviate your VRAM issues, but if you're dead set on trying you can technically do it that way. you will want to install nvidia-prime and use it to force the tools you want on nvidia to the nvidia gpu
Offline
Did you generate a xorg config with nvidia-xconfig? Remove that that will force xorg to load on the nvidia gpu despite you wanting to offload things to the iGPU
I still think you have too much hope for how this will alleviate your VRAM issues, but if you're dead set on trying you can technically do it that way. you will want to install nvidia-prime and use it to force the tools you want on nvidia to the nvidia gpu
Thank you so much it worked exactly as I wanted to and as I expected. Just now tested with my application and there is no lag or stutters now!
Offline