ghost82

Descriptions are nice, pictures are even more, but diagnostics are really needed

I understood now, I missed you wrote no internet connection, so the 'next' button is greyed, see if this helps: https://yyaloo.com/de/index.php/posts/come-installare-windows-11-senza-connettersi-a-internet#:~:text=Wenn Windows 11 Home anbietet,den normalen Installationsprozess nicht unterbrechen. By pressing alt+f4 you should be able to create a local account and skip that window.

ok, no usb, you could run this test: enable the passthrough, enable remote management inside mac os (or any other tool to remote into the vm), boot the vm (black screen), connect inside the vm remotely and run ioregistry explorer, save a copy of that file: you can send it to me in private so we can take a look about how and if the gpu is detected. Can you attach also the diagnostics? Sorry, what do you mean with this?

maybe you explained it well, but I don't understand You want a windows user called 'UserX' able to login in each of the 4 vms? Then, you need to create UserX in each vm...A virtual machine is like a computer, so each os needs a user, at least.

A couple of comments. In your xml change from this: <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x0'/> </source> <rom file='/mnt/disk2/domains/AMD.RX6900XT-REF.rom'/> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x0'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x1'/> </source> <address type='pci' domain='0x0000' bus='0x05' slot='0x00' function='0x0'/> </hostdev> To this: <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x0'/> </source> <rom file='/mnt/disk2/domains/AMD.RX6900XT-REF.rom'/> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x0' multifunction='on'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x1'/> </source> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x1'/> </hostdev> What is the output of lspci?Does this gpu has also an usb controller/serial bus controller?If so you need to pass also that devices, paste following block right under the corrected block above (in the same multifunction), i.e. probably: <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x2'/> </source> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x2'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x04' slot='0x00' function='0x3'/> </source> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x3'/> </hostdev> Note: attach to vfio also usb controller and serial bus if you didn't. Where did you grab AMD.RX6900XT-REF.rom? Dump your own, or if it's not the only gpu in the system do not pass at all the vbios, my 6900xt (sapphire) works without vbios too. If you are passing through a vbios that's not grabbed from your gpu, maybe you are passing through an xtxh vbios, which has different product id and it will not work. Vendor/product ids (from your screenshot) match that inside the apple driver, so it should work. Additional note: if your gpu has a switch dump/use the regular one and not the performance one. If it doesn't work: 1. try other video outputs (if you can) on the gpu 2. xfx gpus may have customized vbioses, that could not play well with mac os; you can try to pass a reference card vbios, I'm seeing that the video output layout should be the same as the reference card (1 hdmi, 2 dp, 1 usb c); you could try this: https://www.techpowerup.com/vgabios/230799/amd-rx6900xt-16384-201104 Memory clock/type is the same so it should be good, at least mine outputs the video signal with the reference vbios, but only on hdmi and on 1 dp out of 3 (take into account that the sapphire layout is different, 1 hdmi, 3 dp). With its vbios or without any vbios the sapphire nitro+ se 6900xt (vendor id 1002, product id 73bf) works good, it only outputs on hdmi and 1 dp port, it outputs to all ports with agdpmod=pikera. Always try all video outputs if you can and always maintain agdpmod=pikera in boot-args.

Hi, only the "another try" seems to be correctly configured: however there's no diagnostics, so we cannot see why it fails. The other xmls, including that inside diagnostics have no trace of the override block, and this cause the error. Please attach diagnostics with the correctly modified override block.

About the new one I think you downloaded the wrong version, that is for arm64 architecture, you need x64.

See if this helps: https://forums.unraid.net/topic/123419-functionnal-vm-wont-boot-anymore-stuck-on-autoboot/

Do you have 3 gpus in one vm? The layout is correct, but again I have no issue in putting a gpu in whatever bus/slot (bus different than 0 in a q35 machine). And yes, I prefer q35.

Strange, never heard of such issue. But on which bus, and on which machine type? for q35 you have bus 0, 1, 2, .....x and each bus have slot 0, 1, 2, ...Y. i440fx is older than q35 chipset, q35 is optimized for pcie passthrough devices, maybe the issue is with your gpu coupled with i440fx.

Sorry, I'm not understanding First, you need to take into consideration the 2 machine types: i440fx and q35. They have different layouts: i440fx has only bus 0 (defined as pci-root) and all the devices attaches to bus 0, so every target address will be bus=0x00, slot=0xX, function=0x0Y. q35 defines pcie-root as its bus 0, and you can attach to it pcie root ports, pcie to pci bridges, other things that you will unlikely (for example pcie-switch-upstream-port and pcie-expander-bus) and devices. So, in q35, devices can be attached to bus 0 (in this case they are "built-in") or to other buses (1, 2, 3,..) defined by the Index value of the pcie root ports, pcie to pci bridges, etc. Both i440fx and q35 support multifunction devices, same bus, same slot, different function. Sometimes it's important to attach devices in a predefined target address, a couple of examples: 1. in a mac os q35 vm, a network device should be attached on bus 0 (built-in), so that it will be seen as builtin in the os, otherwise apple services will not work 2. in a mac os q35 vm, analogic audio should be attached on bus 0 (built-in), so that it will be seen as builtin in the os, otherwise analogic audio will not work 3. in all vms, a multifunction gpu should have target address defined as multifunction, otherwise it may not work because of the drivers.

It will probably ask for reactivation, from unraid website: Changing from i440fx to q35 and the other hand is the same as changing the motherboard on a real pc.

From the description it seems that you are using a 3.x port on your mobo and a 2.0 cable, so your devices works in 2.0 mode; when attached directly to the port it's 3.x (?).

Fastest way is to create a new vm and point it to the actual vdisk and add the same devices. This is because i440fx and q35 have different layouts, i440fx attaches all on bus 0, q35 has bus 0, defined by the pcie-root, with different pcie-root-ports attached to it for other buses. Changing the machine type in the gui will not change the layout so you will receive error, most probably the first one that a pcie-root is not defined. One can modify the xml manually, but it will be the same as creating a new vm, in this last case in less time.

you need: agdpmod=pikera in your boot-args, in the opencore config.plist.

No, I don't think so, the log is that of libvirt/qemu, unrelated to what's doing the vm inside. Have a read at this, it seems you are not alone: https://docs.microsoft.com/en-us/answers/questions/578585/pc-meets-requirments-for-windows-11-but-windows-up.html Some people wrote some solutions. https://www.windowslatest.com/2021/10/07/microsoft-confirms-false-this-pc-cant-run-windows-11-error/ As an alternative you can try to upgrade in place, by downloading the windows 11 iso and run the local installer, upgrading your windows 10, instead of using windows update. https://answers.microsoft.com/en-us/windows/forum/all/upgrade-from-windows-10-to-windows-11-using-iso/cc730f52-cd85-45d6-8b52-c10d3ff2d67e

mmm that should not be because of the tpm/secure boot capable requirements, because ovmf is correctly set and so the tpm block in the xml (vm: Windows 10 Pro (2)). Does the checker program tells more about what's incompatible?There should be an app that you can run that should give more detailed outputs to what's wrong. Maybe the app tells you that you don't meet minimum requirements but you can install it anyway? Is your vdisk larger than 64 Gb? Other requirements, that however should not be mandatory, include: - gpu compatible with direct x 12, attached to a display - compatible cpu But as far as I know these should not be mandatory.... I never upgraded from windows 10, but I installed a fresh windows 11, with tpm, secure boot compatibility, enough ram and disk space and incompatible cpu and all was fine.

Nice I built in the past one of these myself with an old keyboard board. So you are emulating the usb controller; which emulated controller type did you use?Did you try with other types? I'm afraid but if it's still missing the only option could be passing a usb controller and attach the joystick.. What is the device that is not showing in your image?

If you don't pass any gpu the system couldn't use hardware video acceleration and it will be quite limited, since you will need to use qxl video or equivalent; if this is the case any remote desktop solution is near the same, even rdp.

If you google enable gpu acceleration over rdp you will find solutions for discrete gpu by tweaking the windows registry if I remember well..this could work also for the igpu...or not... I know you are now thinking, if you don't know why you reply? I would not consider RDP at all for your use, if you passthrough a gpu/igpu you are enabling hardware acceleration, so your apps/games should use hardware acceleration but rdp will render the screen output and it will send it over the network; this is how rdp works, vnc for example works in a different way. I would consider parsec or any other protocol studied for game streaming and avoid rdp, and obviously with a passed through gpu/igpu.

That's because windows has no virtio driver installed and this results in an unbootable disk. Have a read also at this: https://forums.unraid.net/topic/112725-vm-performance-issues-disk-speed-with-nvme-and-ssd-drives-solved/?do=findComment&comment=1026429

Nice that you found the issue, bad news that it probably hasn't an easy solution. And probably you wont get help from the hackintosh community in general because they don't need any vendor reset since they're running bare metal. I never had any amd gpu that needed vendor reset patch so unfortunately I cannot help too. If I were you I would open an issue directly on github: https://github.com/gnif/vendor-reset/issues gnif was always nice and maybe he could have a look and maybe issue a patch, then it could be probably merged/updated into new unraid versions (if this happens I imagine it could take a lot of time). If and when you open the issue, describe in details what's happening, provide logs, the syslog in unraid diagnostics, you will see some lines referencing the vendor reset patch.

mmmm...the 4 layout partitions is that of an uefi installation: System - MSR - Windows - Recovery The 3 layout partitions is that of a legacy installation: System - Windows - Recovery However you wrote that nvme had 4 partitions and was installed with uefi (and this is accordingly to the 4 partitions layout), but the xml inside your diagnostics (Windows 10 vm) has ovmf and this is not in accordance to the 3 partitions layout you got: did you change from ovmf to seabios in the new installation?

Obviously you can run the os you want but take into account that it wont be easier. Both unraid and proxmox are based on qemu+kvm and libvirt, proxmox and unraid are linux oses. You can obtain the same with any linux distribution as far as qemu and libvirt are installed, so why it should be easier?

I edited my post above, in particular: and for the new vm: