You can find all scripts, links to external websites and related videos in the video description. If you like this video, please subscribe to my channel: 👉 www.youtube.com/@mylinuxforwork/?sub_confirmation=1
This channel is a real find, it's clean, concise and amazingly edited. Your channel appeared to me not long ago and since then I've learned so much good stuff. Your thumbnails are well done, just by looking at them you can tell what the video is about, in addition to the care you take in placing time markers in the video. Soon you will be among the biggest distro channels on YT.
Your content is awesome!! Would love to see a guide from you that goes through some of the arch Linux General recommendations wiki page. I’m confident with the install instructions you’ve done so far, but maybe you could help. with some tips for setting up and securing an arch based install beyond just the interface customizations. Keep it up! You rock!
Many people use Garuda Linux or Endeavour OS; these don't have mkinintcpio except they have dracut. So please upload an instruction for Dracut-based systems.
Thanks for your feedback. My main platform is currently Arch Linux. That's why it's hard for me to create a video for EndeavourOS GPU passthrough. Have EndeavourOS only installed on aVN without GPU passthrough for testing my dotfiles. I really recommend this great article on Arch Linux Wiki wiki.archlinux.org/title/PCI_passthrough_via_OVMF
man, i am sorry, but i have to say for you: dont , never use simple RDP to access your headless win11 vm. You can use one low latency solution like parsec or looking glass. I see the poor graphic performance in your vm using rdp standard. You can using too moonlight/sunshine .
I'm using looking glass in parallel, if I need high performance garphic. The nice thing on RDP is that RDp renders the passed through GPU as well incl. all desktop effects + the VM window size is fully dynamic and adapts automatically on the fly to the window size. Great for tiling modes and Office Apps. Was not able to get dynamic screen resolutions running with looking glass.
is it possible to have one gpu that is used by default on linux but when the vm is launched the secondary gpu is used. example, no vm open - rtx 4090 used on linux, gtx 1080 idle windows vm open - rtx 4090 used on windows, gtx 1080 used for linux
Hey man.. what a treasure trove of a video! love the effort! Just wondering whether it will work with only integrated graphics as I am rocking a core i7-1065G7 with Intel Iris plus Graphics G7
Should work on nearly every system with two GPUs, also with Intel Iris plus + an additional GPU. But, as mentioned in the video, it's always super individual and you need to try it out.
My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
Hi. This will probably not work unfortunately. Because you need to pass through the complete iommu group to the vm. Maybe there is a chance to regroup. There are some tutorials available in the web.
Great Stuff!! Reaqllyenjoy your site and appreciate all your hard work that you put into this! Keep it going! Question; I would like to be able select a waybar when selecting a wallpaper .. I have made a waybar for each wallpaper that matches with the colors nad waybar config. PLease guide me onhow this can be done. I use my own waybar background but with colors that match my waybar!
This sounds like themes. Preconfigured setup of waybar, background and Hyprland settings would be great. I agree. Maybe something for upcoming versions. But currently it is not possible without restructuring the scripts. But if you study my script you can maybe replicate it with an own custom script.
The only loss is on the guest CPU because the CPU is shared with the host. But this can be adjusted very well by giving the Windows guest more resources. On the GPU that is passed through to the Windows guest is no performance loss because it’s detached from the host and exclusively available for the guest.
@@clithulhu4017 My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
Du brauchst definitiv zwei GPUs. Mit einer integrierten Karte habe ich das noch nicht versucht. Musst die GPU halt vollständig durchreichen, heisst muss in einer IOMMU group liegen. Kannst es mit diesem Script testen: gitlab.com/stephan-raabe/archinstall/-/blob/main/gpu-pt/test-iommu.sh?ref_type=heads
Been thinking of having a similar setup but starting the vm on boot and then saving the VM state to disk after a minute and then resuming it with the connect to rdp script and also save it again after xfreerdp closes, also maybe to use a keybind in hyprland to launch the script. Would probably have to make sure the guest clock syncs correctly, maybe qemu-guest-agent would fix it?
Ohhh this is really cool I'm very new to Arch linux. Was introduced to linux just a couple of months back. Currently using Compton to have transperancy on my terminals while using only i3wm and polybar. My laptop runs amd 5600H and amd Radeon RX5700m as its GPU will this be possible i really want to do some Wayland stuff read the documentation seems really interesting.
When I understand your system correctly, you have an CPU with integrated GPU plus a second GPU RX5700m? If yes, then it’ll work. You need two GPUs tonrun a stable GPU passthrough that is similar to my setup.
Almost feel like taking this route, but not quite. I have a 4080 that I plan to pass-through to the vm for gaming performance and a 1050 I could use for the rest of the environment. Just one thing though. I would want to have my 4080 take over the linux desktop environment after the windows vm shuts down, all automatically of course. Any way this can be done?
On a laptop you will run into bandwidth bottlenecks since Looking Glass works by copying uncompressed frames from the guest GPU into system RAM, and then into the host VRAM. A iGPU doesn't have VRAM and instead uses system RAM, which can create a memory bottleneck. Also, not every laptop has a physical display output on the dedicated GPU, which you need to connect a dummy plug or real monitor to. Without a monitor or dummy plug Looking Glass will not work.
@@clithulhu4017 My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
Thank you for your videos. I am having a problem with looking-glass. Whenever I set in virt-manager the video to 'None' looking-glass doesnt display anything. However if I set the video to VGA it does work but at low performance. How can I make it work.
@@mylinuxforwork Thanks for the reply. Yes, the problem was when running looking-glass-client I wasn't getting anything displayed. I finally figured it out. I had to install IddSampleDriver in my Windows VM for looking-glass-client to display something.
I have an interesting problem with my setup. And any help would be appreciated. My main GPU is an RX 5700XT. My secondary GPU is a Quadro FX 570. When I install the FX 570 into my system, the system refuses to output from the main GPU, and I lack a DVI-to-HDMI cable to see if the system is outputting any video there. Is there a way to force the video to the primary GPU in the BIOS? I have an ASRock X570 Steel Legend.
When you turn of your VM, the second GPU will be detached from the guest but is still binded to the vfio driver for the next start. The GPU from your host is not affected at all.
is it possible to passthrough the integrated gpu in my intel cpu and use my nvidia gpu for linux? Or the nvidia gpu for windows only when windows is running, and for linux otherwise?
@@mylinuxforwork The iGPU in guest and nvidia in host worked, thanks! I had to enable the iGPU in the BIOS. I guess using the nvidia for both would be the same as the more complex single gpu passthrough.
According to the Linux Foundation and Arch Wiki, "Note: the bridge-utils package is DEPRECATED, instead use the bridge command from iproute2 package." But you say to install it. Do you know what's going on with this? I'm confused.
Hi. It’s a very dynamic topic. So changes or new recommendations are possible. What counts is what works at the end. I’m planning a new version of the video in the upcoming months.
I followed all the instructions, managed to start Looking Glass, but I can't move the keyboard or the mouse inside the Looking Glass window. Could someone please help me?
Can you check this article: gitlab.com/risingprismtv/single-gpu-passthrough/-/wikis/3)-IOMMU-Groups Is your 6700xt in one group or in two different groups?
I tried it on my latop(have two GPU) have installed nvidia hyprland. After changing vfio setting and reboot the latop I can't connect to my HDMI monitor anymore, how can I fix that? test vfio config script do well
Hi. That’s difficult to find out without more details. GPU passthrough is always an individual experience and requires trying thinks out. Have you installed the ROM correctly in virt-manager?
Hello im stuck at the part where you test the configuration with the "test-vfio.sh" script, when I launch it the output is "[ 1.083090] VFIO - User Level meta-driver version: 0.3". How can I solve it ? (I havn't found the fix on internet yet...)
Just jump in to this gorgeous guide, in case you have the same problem as mine. I have to bring the microcode comes after the modconf in the hooks HOOKS=(base udev autodetect modconf microcode kms keyboard keymap consolefont block filesystems fsck)
Honestly, I’m using both. Hyprland and Qtile. Switching between both. Android apps should work on Windows 11 KVM in the same way as with real hardware. But I have never test it. Have you ever tested Waydroids? waydro.id
I've heard of waydroid but never tried it. I like the simplicity of your qtile kvm setup. Been following along. I'll try to see if I can get android apps working nativaly on windows 11 kvm. Love tinkering
Hi. Never tested without GRUB. Pretty sure that systemd also has the possibility to define kernel parameters. wiki.archlinux.org/title/kernel_parameters
Hello, do you know if it is possible to add the IOMMU and vfio settings to Limine's CMDLINE section like you did for GRUB at 5:44? Limine's CMDLINE section looks like this: CMDLINE=root=UUID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx rw I'm not entirely sure if this will work but they have yet to publish a way to uninstall Limine so I cannot use GRUB.
@@mylinuxforwork I went ahead and just reinstalled with GRUB this time to make it easier/be safe. Now I am stumped at the VFIO configuration, I am on an AMD GPU and the GPU and audio controller show up in 2 different groups. I followed another link you replied to someone else with, but I can't get 2 IOMMU groups to go through, not sure what I'm doing wrong. Config is follows: options vfio-pci ids=1002:73ff,1002:ab28 The first ID is in group 14 and the second is group 15, using your check script VFIO only shows 1002:73ff as bound but not 1002:ab28. Any ideas? Much appreciated.
@@mylinuxforwork **EDIT I FIGURED IT OUT:** *For AMD GPU users, if the audio device and the GPU show up in two separate IOMMU groups, you'll have to define "vfio-pci.ids=(ID #1),(ID #2)" inside the "GRUB_CMDLINE_LINUX_DEFAULT=" line from **5:58** in order to get them to both bind to VFIO.* *Pre-Edited* It's okay, I know that it is kind of specific per-ID, it's just that in the DOCs it says "If your devices were in 2 different groups like this...: ...you need to pass both the entirety of group 8, AND 9 in your VM" and I can't find any explanation of how to do that, I tried just grabbing the IDs from the separate groups but I can't get them both to bind to VFIO, only one from one group and not both from both groups. I do greatly appreciate your replies though.
Great share, Bro! Thanks a lot, and I success with your step, but when I reboot, and restart the virtual windows 11, failed with "Error starting domain: Requested operation is not valid: network 'default' is not active". And when intstall the nft-iptables, asked the confic the iptables, and I choose remove the iptables, I'm not sure whether this be effictive
Please try sudo virsh net-autostart default and then sudo virsh net-start default should activate the network. Your setup with iptables should be fine.
Hello. I'm a newbie who wants to transition from windows 10 to Linux Mint. I have installed LM in virtual box but I'm having an issue. Virtual box doesn't pass through the GPU. Therefore I can't install the driver for my GPU because it's not being detected. I'm not getting proper hardware acceleration. Is there a VM or something that will allow me to pass through my GPU so that I can test LM properly before switching over to it? I've heard of Hyper V, will that work?
Thanks for your comment. Unfortunately, in your setup I have no experiences. Linux Mint should work probably in Virtualbox. But you’re right that this cannot run in native performance. But should be enough to test your main use cases. Except gaming or extrem video editing.
@@mylinuxforwork It does work in VB but since it's not GPU accelerated, the mouse movement and window window switching is rather sluggish. I checked stats for nerds on a YT video and it showed that it was dropping frames. Oh well, I'll search around. Thanks for your reply and your content is great!
@@corvoattano9303 You can always boot a live USB Image and test it out on real hardware without installing it onto your HDD/SSD. Note that all changes will revert unless you make them permanent, so it's really only for testing it out.
@@DiskyToy yes I did try that out. The proper GPU driver was found by the driver updater. So I'm sure it will be GPU accelerated when I install it on bare metal. I just wanted the driver to be found in the VM so I could test it out properly before moving over. Honestly this is not a big deal. I can test out LM even if it's not properly GPU accelerated.
hey you might be able to help i got everything working upto looking-glass. i'm not too sure why its not working. i've tried release and bleeding edge on both a win10 and win11 vm. either the service doesn't boot or crashes instantly if i have RDP window open and i boot looking glass on the vm and then the client on linux it works and i can see the screen and then i close RDP and windows crashes the looking glass host. you have any idea what might be causing it? just so you have the specs Ryzen 7 5800x 64GB ram - 48gb of it is in the VM Linux GPU: Radeon Pro WX3100 4GB Win/VM GPU: AMD 6900XT
@@mylinuxforworkso manged to get the host not to crash but display still won't show up and checking log doesn't say much says dxgi_init, app_main, lgmpSetup. and on client states the host application seems not to be running
@@mylinuxforwork Also wenn ich das richtig verstanden hab hast du eine GPU für passthrough und deine igpu als host? hast du die GPU mit einen HDMI Connected oder bist du am Mainboard verknüpft?
@@user-lt3rx3qx7z habe eine CPU ohne GPU, eine AMD Karte für den Host und eine Intel für den Guest. Beide Karten hängen an getrennten Monitoreingängen.
So much bad advise here. I dont know where to start... Fallow the Looking Glass Doc and not this video - do not use AUR Looking Glass Client use the tar.gz provided by LG website... (what if AUR is newer then B6? IT WONT WORK) Virtio Tablet -> NO you have to remove tablet from the config (LG Doc if you read that...) and add virtio Mouse. Where Virtio Nic -> e1000e is bad. VBIOS from NV card? Maybe with your 700 card but newer no just no. No Display on Virt Manager after Pass through how about no (again!) - it works esp with LG B6 amd-iommu=on does not even EXIST (in the source / man pages). Intel Yes but AMD no. etc PP.
Thank you for your feedback. As mentioned in the video... GPU passthrough is very individual work and requires experiments. I have described what worked for me. Your feedback is very welcome and hopefully helpful for more Linux users.
@@mylinuxforwork Yep!! Thanks I saw that , just wanted to make sure if I missed something! By the way..... what part of Germany are you from? I've been to Gremany many times as I have a brother that lives there.. since 1972.... obviouly speakles fluent German. Bad Kreunazch Bad Munster am Stein!
You can find all scripts, links to external websites and related videos in the video description.
If you like this video, please subscribe to my channel: 👉 www.youtube.com/@mylinuxforwork/?sub_confirmation=1
This channel is a real find, it's clean, concise and amazingly edited. Your channel appeared to me not long ago and since then I've learned so much good stuff. Your thumbnails are well done, just by looking at them you can tell what the video is about, in addition to the care you take in placing time markers in the video. Soon you will be among the biggest distro channels on YT.
Thank you so much for your comment. That motivates me a lot. And there is still so much more to discover.
Empeze a ver tus videos por hyprland y realmente esto me sorprendio, gracias por tu contenido, saludos desde chile un abrazo brother
Thank you very much.
Your content is awesome!! Would love to see a guide from you that goes through some of the arch Linux General recommendations wiki page. I’m confident with the install instructions you’ve done so far, but maybe you could help. with some tips for setting up and securing an arch based install beyond just the interface customizations. Keep it up! You rock!
Thanks for your feedback. Any specific topic that you have in mind around settings and security?
very good video, your demonstration is digestible
Thank you.
Thanks Sir. Very helpful 🙂❤️
Thanks for watching.
Many people use Garuda Linux or Endeavour OS; these don't have mkinintcpio except they have dracut. So please upload an instruction for Dracut-based systems.
Thanks for your feedback. My main platform is currently Arch Linux. That's why it's hard for me to create a video for EndeavourOS GPU passthrough. Have EndeavourOS only installed on aVN without GPU passthrough for testing my dotfiles. I really recommend this great article on Arch Linux Wiki wiki.archlinux.org/title/PCI_passthrough_via_OVMF
keep up the AMAZING content!
Thank you. Will do.
Another great video!
Thank you.
man, i am sorry, but i have to say for you: dont , never use simple RDP to access your headless win11 vm. You can use one low latency solution like parsec or looking glass. I see the poor graphic performance in your vm using rdp standard. You can using too moonlight/sunshine .
I'm using looking glass in parallel, if I need high performance garphic. The nice thing on RDP is that RDp renders the passed through GPU as well incl. all desktop effects + the VM window size is fully dynamic and adapts automatically on the fly to the window size. Great for tiling modes and Office Apps. Was not able to get dynamic screen resolutions running with looking glass.
is it possible to have one gpu that is used by default on linux but when the vm is launched the secondary gpu is used.
example,
no vm open - rtx 4090 used on linux, gtx 1080 idle
windows vm open - rtx 4090 used on windows, gtx 1080 used for linux
Don’t think that’s possible. You need to bind the gpu to virtio on startup. But you can try single gpu passthrough.
@@mylinuxforwork alright thanks man i'll try look around and see. i appreciate the tutorial also
Hey man.. what a treasure trove of a video! love the effort! Just wondering whether it will work with only integrated graphics as I am rocking a core i7-1065G7 with Intel Iris plus Graphics G7
Should work on nearly every system with two GPUs, also with Intel Iris plus + an additional GPU. But, as mentioned in the video, it's always super individual and you need to try it out.
My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
Hi. This will probably not work unfortunately. Because you need to pass through the complete iommu group to the vm. Maybe there is a chance to regroup. There are some tutorials available in the web.
Great Stuff!! Reaqllyenjoy your site and appreciate all your hard work that you put into this! Keep it going! Question; I would like to be able select a waybar when selecting a wallpaper .. I have made a waybar for each wallpaper that matches with the colors nad waybar config. PLease guide me onhow this can be done. I use my own waybar background but with colors that match my waybar!
This sounds like themes. Preconfigured setup of waybar, background and Hyprland settings would be great. I agree. Maybe something for upcoming versions. But currently it is not possible without restructuring the scripts. But if you study my script you can maybe replicate it with an own custom script.
Thanks for the video, that's amazing. I have one question: Do I have any loss of performance in games when I do this?
The only loss is on the guest CPU because the CPU is shared with the host. But this can be adjusted very well by giving the Windows guest more resources. On the GPU that is passed through to the Windows guest is no performance loss because it’s detached from the host and exclusively available for the guest.
This is incorrect. The guest GPU will have to do the capture, which has significant overhead@@mylinuxforwork
@@clithulhu4017 My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
wie geil ist das denn. sag mal. kann man auch die integrierte grafikkarte des motherboards benutzen oder muss es eine seperate graphikkarte sein?
Du brauchst definitiv zwei GPUs. Mit einer integrierten Karte habe ich das noch nicht versucht. Musst die GPU halt vollständig durchreichen, heisst muss in einer IOMMU group liegen. Kannst es mit diesem Script testen: gitlab.com/stephan-raabe/archinstall/-/blob/main/gpu-pt/test-iommu.sh?ref_type=heads
Awesome!
Thanks.
Been thinking of having a similar setup but starting the vm on boot and then saving the VM state to disk after a minute and then resuming it with the connect to rdp script and also save it again after xfreerdp closes, also maybe to use a keybind in hyprland to launch the script.
Would probably have to make sure the guest clock syncs correctly, maybe qemu-guest-agent would fix it?
Yes. This should work.
Thanks for the guide
How did you managed to install Looking-Glass client on Arch ? Seems wayland-protocols dependencies aren't available anymore.
The installation with yay -S looking-glass works for me.
@@mylinuxforwork For some reason, was installing the -git version which can't get all dependencies... the "looking-glass" is just fine. Thanks !
@@MichelStumpf Perfect! Thanks for the good news.
Ohhh this is really cool I'm very new to Arch linux. Was introduced to linux just a couple of months back. Currently using Compton to have transperancy on my terminals while using only i3wm and polybar. My laptop runs amd 5600H and amd Radeon RX5700m as its GPU will this be possible i really want to do some Wayland stuff read the documentation seems really interesting.
When I understand your system correctly, you have an CPU with integrated GPU plus a second GPU RX5700m? If yes, then it’ll work. You need two GPUs tonrun a stable GPU passthrough that is similar to my setup.
Almost feel like taking this route, but not quite. I have a 4080 that I plan to pass-through to the vm for gaming performance and a 1050 I could use for the rest of the environment. Just one thing though. I would want to have my 4080 take over the linux desktop environment after the windows vm shuts down, all automatically of course. Any way this can be done?
I think that's not possible. You need to detach the GPU from your system to enable passthrough. In your case, you can try single GPU passthrough.
Where's the rdp script? You don't include it anywhere it seems.
Hi. This is the latest version and will be shipped with 2.9.1 gitlab.com/stephan-raabe/dotfiles/-/blob/dev/scripts/launchvm.sh?ref_type=heads
i love your videos. can you make a video attempting this on a dual gpu laptop ? thanks
Thank you. Unfortunately I’m not owning a laptop with two GPUs. But it should work similar to setup GPU passthrough.
On a laptop you will run into bandwidth bottlenecks since Looking Glass works by copying uncompressed frames from the guest GPU into system RAM, and then into the host VRAM. A iGPU doesn't have VRAM and instead uses system RAM, which can create a memory bottleneck.
Also, not every laptop has a physical display output on the dedicated GPU, which you need to connect a dummy plug or real monitor to.
Without a monitor or dummy plug Looking Glass will not work.
@@clithulhu4017 My laptop has iGPU with the CPU, that means I have 2 GPU, when I checked iommu list, the iGPU was in same iommu group with the NVIDIA dedicated GPU. Will this still work? If not, how do I separate it out into there own groups?
Thank you for your videos. I am having a problem with looking-glass. Whenever I set in virt-manager the video to 'None' looking-glass doesnt display anything. However if I set the video to VGA it does work but at low performance. How can I make it work.
Are you sure that you’re using the looking class viewer? It is a different app than virt-manager.
@@mylinuxforwork Thanks for the reply. Yes, the problem was when running looking-glass-client I wasn't getting anything displayed. I finally figured it out. I had to install IddSampleDriver in my Windows VM for looking-glass-client to display something.
I have an interesting problem with my setup. And any help would be appreciated.
My main GPU is an RX 5700XT.
My secondary GPU is a Quadro FX 570.
When I install the FX 570 into my system, the system refuses to output from the main GPU, and I lack a DVI-to-HDMI cable to see if the system is outputting any video there.
Is there a way to force the video to the primary GPU in the BIOS? I have an ASRock X570 Steel Legend.
As far as I know you can select the primary GPU in a Bios. On my BIos it is possible to. Please check the manual of your motherboard.
I have a question, when I turn off the VM is my host machine still use the GPU or it will not work only in the VM, and thank you
When you turn of your VM, the second GPU will be detached from the guest but is still binded to the vfio driver for the next start. The GPU from your host is not affected at all.
is it possible to passthrough the integrated gpu in my intel cpu and use my nvidia gpu for linux? Or the nvidia gpu for windows only when windows is running, and for linux otherwise?
You need to change the primary GPU in the BIOS and make sure that the integrated GPU is in an own IOMMU Group. If this is the case then it could work.
@@mylinuxforwork The iGPU in guest and nvidia in host worked, thanks! I had to enable the iGPU in the BIOS. I guess using the nvidia for both would be the same as the more complex single gpu passthrough.
According to the Linux Foundation and Arch Wiki, "Note: the bridge-utils package is DEPRECATED, instead use the bridge command from iproute2 package." But you say to install it. Do you know what's going on with this? I'm confused.
Hi. It’s a very dynamic topic. So changes or new recommendations are possible. What counts is what works at the end. I’m planning a new version of the video in the upcoming months.
hey man, rom file and dump flash for using with qemu is not mandatory. I never need this and i using one rtx4060 only for guests.
In some situation a rom dump is required. But this depends on the system setup. In the most of the cases it's not required.
I followed all the instructions, managed to start Looking Glass, but I can't move the keyboard or the mouse inside the Looking Glass window. Could someone please help me?
Can you check this article looking-glass.io/docs/B6/install/#keyboard-mouse-display-audio Hope it works.
@@mylinuxforwork I follow that part and i still have the same problem. Any other suggestions? And thank you for answering me.
In my case the Nvidia driver is just not installing :(
Hi. Difficult to find out the reason without seeing your configuration. Try again, restart and install the drivers manually.
Hi, I want to passthrough my 6700xt, but the VGA compatible controller and Audio is seperated. What's wrong?
Can you check this article: gitlab.com/risingprismtv/single-gpu-passthrough/-/wikis/3)-IOMMU-Groups Is your 6700xt in one group or in two different groups?
@@mylinuxforwork My 6700xt is in two different groups.
Hello in this particular video can you tell us which version of edk2-ovmf are you using thank you
Hi. Do you know how to get the version information?
@@mylinuxforwork Hi doing this command pacman -Qi edk2-ovmf
I tried it on my latop(have two GPU) have installed nvidia hyprland.
After changing vfio setting and reboot the latop I can't connect to my HDMI monitor anymore, how can I fix that?
test vfio config script do well
Hi. That’s difficult to find out without more details. GPU passthrough is always an individual experience and requires trying thinks out. Have you installed the ROM correctly in virt-manager?
Hello im stuck at the part where you test the configuration with the "test-vfio.sh" script, when I launch it the output is "[ 1.083090] VFIO - User Level meta-driver version: 0.3". How can I solve it ? (I havn't found the fix on internet yet...)
Have you found a solution?
Just jump in to this gorgeous guide, in case you have the same problem as mine. I have to bring the microcode comes after the modconf in the hooks HOOKS=(base udev autodetect modconf microcode kms keyboard keymap consolefont block filesystems fsck)
What is your main window manager? qtile or hyperland. can you install android apps on windows 11 kvm?
Honestly, I’m using both. Hyprland and Qtile. Switching between both. Android apps should work on Windows 11 KVM in the same way as with real hardware. But I have never test it. Have you ever tested Waydroids? waydro.id
I've heard of waydroid but never tried it. I like the simplicity of your qtile kvm setup. Been following along. I'll try to see if I can get android apps working nativaly on windows 11 kvm. Love tinkering
Hi Bro, I installed the arch with systemd-boot, and there is no grub file, so must I change to grub bootloader ?
Hi. Never tested without GRUB. Pretty sure that systemd also has the possibility to define kernel parameters. wiki.archlinux.org/title/kernel_parameters
Hello, do you know if it is possible to add the IOMMU and vfio settings to Limine's CMDLINE section like you did for GRUB at 5:44?
Limine's CMDLINE section looks like this: CMDLINE=root=UUID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx rw
I'm not entirely sure if this will work but they have yet to publish a way to uninstall Limine so I cannot use GRUB.
Never did this before. Sorry. But give it a try.
@@mylinuxforwork I went ahead and just reinstalled with GRUB this time to make it easier/be safe.
Now I am stumped at the VFIO configuration, I am on an AMD GPU and the GPU and audio controller show up in 2 different groups. I followed another link you replied to someone else with, but I can't get 2 IOMMU groups to go through, not sure what I'm doing wrong. Config is follows: options vfio-pci ids=1002:73ff,1002:ab28
The first ID is in group 14 and the second is group 15, using your check script VFIO only shows 1002:73ff as bound but not 1002:ab28. Any ideas? Much appreciated.
@@user-bm2vs1ee1g I am so sorry but I cannot help you in this case. The passthrough setup is bery individual. Mybe you need to switch the cards?
@@mylinuxforwork
**EDIT I FIGURED IT OUT:**
*For AMD GPU users, if the audio device and the GPU show up in two separate IOMMU groups, you'll have to define "vfio-pci.ids=(ID #1),(ID #2)" inside the "GRUB_CMDLINE_LINUX_DEFAULT=" line from **5:58** in order to get them to both bind to VFIO.*
*Pre-Edited* It's okay, I know that it is kind of specific per-ID, it's just that in the DOCs it says "If your devices were in 2 different groups like this...: ...you need to pass both the entirety of group 8, AND 9 in your VM" and I can't find any explanation of how to do that, I tried just grabbing the IDs from the separate groups but I can't get them both to bind to VFIO, only one from one group and not both from both groups. I do greatly appreciate your replies though.
@@user-bm2vs1ee1g 👍
Great share, Bro! Thanks a lot, and I success with your step, but when I reboot, and restart the virtual windows 11, failed with "Error starting domain: Requested operation is not valid: network 'default' is not active". And when intstall the nft-iptables, asked the confic the iptables, and I choose remove the iptables, I'm not sure whether this be effictive
Please try sudo virsh net-autostart default and then sudo virsh net-start default should activate the network. Your setup with iptables should be fine.
Thanks a lot!!!@@mylinuxforwork
Hello. I'm a newbie who wants to transition from windows 10 to Linux Mint. I have installed LM in virtual box but I'm having an issue. Virtual box doesn't pass through the GPU. Therefore I can't install the driver for my GPU because it's not being detected. I'm not getting proper hardware acceleration. Is there a VM or something that will allow me to pass through my GPU so that I can test LM properly before switching over to it? I've heard of Hyper V, will that work?
Thanks for your comment. Unfortunately, in your setup I have no experiences. Linux Mint should work probably in Virtualbox. But you’re right that this cannot run in native performance. But should be enough to test your main use cases. Except gaming or extrem video editing.
@@mylinuxforwork It does work in VB but since it's not GPU accelerated, the mouse movement and window window switching is rather sluggish. I checked stats for nerds on a YT video and it showed that it was dropping frames. Oh well, I'll search around. Thanks for your reply and your content is great!
@@corvoattano9303 You can always boot a live USB Image and test it out on real hardware without installing it onto your HDD/SSD. Note that all changes will revert unless you make them permanent, so it's really only for testing it out.
@@DiskyToy yes I did try that out. The proper GPU driver was found by the driver updater. So I'm sure it will be GPU accelerated when I install it on bare metal. I just wanted the driver to be found in the VM so I could test it out properly before moving over. Honestly this is not a big deal. I can test out LM even if it's not properly GPU accelerated.
Bro I changed my grub files and my entire qtile config went bye bye
Hi. Sorry for that. Not sure how grub is connected with Qtile. Keep trying.
@@mylinuxforwork I think it may be one of the dependencies tbh because I had conflicts with iptables-nft
How do you remove Windoze bloat and spying/telemetry?
I am using Windows in standard installation. So, not removing the bloat etc because I try to do all of my jobs on Arch directly.
@@mylinuxforwork makes sense
Hey! I am having problems connecting via RDP i tried looking at your other video and other solutions online 😭
Difficult to bugfix without knowing more details. Is RDP activated in Windows 11. Have you created a local user?
@@mylinuxforwork its a local user the VM is pulling from my windows 10 machine its a dual boot system but i am using the windows boot as a vm
I have RDP enabled and disabled NLA
Yet somehow it just fails to connect i tried diffrent parameters too
im assuming its a problem on the windows side however I am not sure i made sure RDP was allowed passthrough firewall too
hey you might be able to help i got everything working upto looking-glass. i'm not too sure why its not working. i've tried release and bleeding edge on both a win10 and win11 vm. either the service doesn't boot or crashes instantly if i have RDP window open and i boot looking glass on the vm and then the client on linux it works and i can see the screen and then i close RDP and windows crashes the looking glass host. you have any idea what might be causing it?
just so you have the specs
Ryzen 7 5800x
64GB ram - 48gb of it is in the VM
Linux GPU: Radeon Pro WX3100 4GB
Win/VM GPU: AMD 6900XT
Have you added the additional information into your VM configuration in virtmanager? looking-glass.io/docs/B6/install/
@@mylinuxforwork yes thats what strange I'll try doing a fresh install again and will confirm. As it could be something in the xml
@@konrad50023 sounds like a plan. Thank you.
@@mylinuxforworkso manged to get the host not to crash but display still won't show up and checking log doesn't say much says dxgi_init, app_main, lgmpSetup. and on client states the host application seems not to be running
@@konrad50023 do you have an output on the GPU output when using a second monitor or input? This must work properly.
does this work on Rocky Linux 9
Hi. I have no experience with Rocky Linux. But give it try. There are many more tutorials available if this one will not work with this distro.
Hast du in deinem System überhaupt 2x GPU verbaut?
Ich habe eine AMD als primäre GPU und eine Intel zum durchreichen.
@@mylinuxforwork Also wenn ich das richtig verstanden hab hast du eine GPU für passthrough und deine igpu als host? hast du die GPU mit einen HDMI Connected oder bist du am Mainboard verknüpft?
@@user-lt3rx3qx7z habe eine CPU ohne GPU, eine AMD Karte für den Host und eine Intel für den Guest. Beide Karten hängen an getrennten Monitoreingängen.
will it work if I dont have a igpu? I have a i5-9400f and rx570
You need two independent GPU. If your i5 has an integrated GPU you can try to passthrough the rx570.
I don't think so, CPU should have internal GPU
@@abdelmoneimelshafei6570 If your CPU has an internal GPU then you can passthrough your second GPU. Please the the Arch Linux Wiki as well.
@@mylinuxforwork I tried those single GPU pass through before but never worked
@@rikxdragneel5290 That’s why I suggest GPU passthrough with a dedicated second GPU for the virtual machine.
dude, you gotta back the camera up from your face at least 10 feet
👍
So much bad advise here. I dont know where to start...
Fallow the Looking Glass Doc and not this video - do not use AUR Looking Glass Client use the tar.gz provided by LG website... (what if AUR is newer then B6? IT WONT WORK)
Virtio Tablet -> NO you have to remove tablet from the config (LG Doc if you read that...) and add virtio Mouse.
Where Virtio Nic -> e1000e is bad.
VBIOS from NV card? Maybe with your 700 card but newer no just no.
No Display on Virt Manager after Pass through how about no (again!) - it works esp with LG B6
amd-iommu=on does not even EXIST (in the source / man pages). Intel Yes but AMD no. etc PP.
Thank you for your feedback. As mentioned in the video... GPU passthrough is very individual work and requires experiments. I have described what worked for me. Your feedback is very welcome and hopefully helpful for more Linux users.
I'm trying th run the ./iommu.sh script....but see it in the dir "archinstall/gpu-pt/ ther's no ./iommu.sh file?? PLease let me know..
sorry..... but DON"T see it in the dir!
Please check gitlab.com/stephan-raabe/archinstall/-/tree/main/gpu-pt?ref_type=heads it’s now test-iommu.sh
@@mylinuxforwork Yep!! Thanks I saw that , just wanted to make sure if I missed something! By the way..... what part of Germany are you from? I've been to Gremany many times as I have a brother that lives there.. since 1972.... obviouly speakles fluent German. Bad Kreunazch Bad Munster am Stein!