giganode

I went for the macinabox solution as I have no experience with opencore. Have always used clover. After the installation I just added AAPL,ig-platform-id .... and agdpmod=pikera. Edited xml for passthrough and booted into macOS.

Did a quick fresh install with opencore and big sur. I can confirm that a 5700xt runs with opencore aswell.

As I only have an 5700xt I sadly cannot try to help you...

Do you use a 5700xt? Which Bootloader do you use? Are you using a custom unraid build? I have a working macOS vm with a 5700xt passed through, but atm I am using the clover bootloader. I will setup a macOS vm with opencore soon for testing.

I just updated to Big Sur 11.0.1 Works like a charm.. with slightly improved performance

Ich nutze tatsächlich Unraid mit einem UltraFit.. Der Weg über den konventionellen Flash Creator eignet sich hierfür nicht. Mach es manuell. Das du nur einen Stick zur Zeit haben kannst ist sinnvoll. Verhindert Piraterie, Punkt.

I have no problems with testing or helping but first off all let the two 5700xt users use the patched build. Output as follows: Display 0 vendor id 4c2d product id f9c EDID: 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00, 0x4c, 0x2d, 0x9c, 0x0f, 0x00, 0x00, 0x00, 0x00, 0x2b, 0x1c, 0x01, 0x04, 0x90, 0x77, 0x22, 0x78, 0x3b, 0xa2, 0xa1, 0xad, 0x4f, 0x46, 0xa7, 0x24, 0x0e, 0x50, 0x54, 0xbf, 0xef, 0x80, 0x71, 0x4f, 0x81, 0x00, 0x81, 0xc0, 0x81, 0x80, 0xa9, 0xc0, 0xb3, 0x00, 0x95, 0x00, 0xd1, 0xc0, 0x74, 0xd6, 0x00, 0xa0, 0xf0, 0x38, 0x40, 0x40, 0x30, 0x20, 0x3a, 0x00, 0xa9, 0x50, 0x41, 0x00, 0x00, 0x1a, 0x00, 0x00, 0x00, 0xfd, 0x00, 0x30, 0x78, 0xbe, 0xbe, 0x61, 0x01, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x00, 0x00, 0xfc, 0x00, 0x43, 0x34, 0x39, 0x52, 0x47, 0x39, 0x78, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x00, 0x00, 0xff, 0x00, 0x48, 0x31, 0x41, 0x4b, 0x35, 0x30, 0x30, 0x30, 0x30, 0x30, 0x0a, 0x20, 0x20, 0x02, 0xf3

Clover with WeG version 1.4.3 running as a iMacPro1,1. But first of all, they have to use a patch.

Now things get really interesting. First off all, everything works now. Even the immidiate restart or shutdown via button, as long as the refresh rate is at 100Hz and I do not change it back up to 120 Hz.

Works!

That's exactly what I was thinking, too!! But anyways guys... I inspected it a bit further and I think I found something out. I am now able to restart the macOS vm directly. Furthermore I am able to start the macOS vm directly after a shutdown now!! I tried it multiple times to verify. What have I done? On restart or shutdown you are prompted with a window with the selection to start open apps on next boot and to directly shut the system down with that specific button. I did not hit that button, instead I waited for 1 minute. Whats the difference? On immediate restart/shutdown macOS closes everything and just goes for it. If the 1 minute expires and the restart/shutdown gets initiated macOS closes everything, but instead after that the screen turns black and back on so I can see the background for around 1-2 seconds again and the restart just simply works flawlessly!!! I would be very thankful, if people with 5700xt's pn @ich777, ask for the link to download the patched build and to verify my described behaviour.

ok... But if you haven't got one... then don't deny my statement...

I'm currently running 6.9.0-beta35 with the old Navi Reset Patch. You can find it here: https://forum.level1techs.com/t/navi-reset-kernel-patch/147547/47 It's no longer supported, because gnif works on something new. But that new project is in the early stages and cannot supersede the old patch as Navi and Vega cards are not working correctly, atm. I tested it. As soon as the new vendor-reset - https://github.com/gnif/vendor-reset - does work, we should give it a try. And maybe - or better hopefully - the new fix can deal with macOS vms better than the old patch.

I use a patched kernel. So no, its just related to my macOS vm.. everything else works fine.

I am familiar with this procedure. I think thats a specific problem on macOS vm as the problem does not occur on linux or windows vms. I am thankful for your idea, but that's just another workaround - which by the way puts the whole system to sleep and requires physical interaction with the server.

First boot with my 5700xt works fine, but after a restart or shutdown the card does not work. I have to start a different vm with the 5700xt assigned first, before I can use the catalina vm with gpu again.

That's interesting. Do you have any new information, because Wendell from Level1 reported that the 3000 series does not output a 43 error anymore?!

I may have good news for some folks here. I actually tested to passthrough my 5700XT to my macOS vm and I can tell you it works. I use macOS Catalina 10.15.7 with clover bootloader. EDIT: It works half way atm. Restart or a start after a shutdown does not work correctly. I need to start a different vm with an other os started and shut down before the gpu works correctly again in macOS. I need to investigate that further. I will keep you updated.

That's good news! 😊

That would be nice! What happens? We will try it ✌️😏 If everything works there could be a possibility to implement it, I think. But only if it does not create new problems for other users.

Up until now nVidia never had a reset bug, afaik. But the RTX 3000 series does not have the Code 43 as f.e. the 2000 series or the gtx series before. As reported by gnif and Level1Techs the new RX 6000 series does not have a reset bug anymore. This is very nice!! Let's all sell our old stuff to bare metal user as soon as we have enough availability of the new generation and let's never have gpu passthrough problems anymore...........

The vendor reset does not work with Navi cards right now. The sound card gets a Code 10 error after 2nd Boot of the vm. The quoted issue also shows that Vega cards don't work atm, too. edit: we could only test with an 5700xt in reference design. If you update to 6.9.0-beta35 you do need the patched build @ich777 provides. Otherwise you cannot use the gpu in vms. Keep in mind, this is a workaround until gnif releases a new version of the vendor-reset. As soon as it works, we should move to the vendor-reset, because the vendor-reset, As I tested hours and hours together with @ich777 last week, does indeed handle the gpu in a better way imo.

Works good so far. Thank you very much!

Just to keep everyone updated. Recently, gnif answered to this issue and stated that their first focus has been on Polaris and Vega cards. So I think this should get fixed relatively quickly 😏

Thank you for sharing that link. I added a question to that thread.. There maybe could be a specific problem with non reference cards or something like that. At least I hope that 🙈 "Doable" sounds pretty good to me 🤪 Have a nice weekend!!!