Contribute
Register

AMD and Nvidia graphic cards at same time

Status
Not open for further replies.
So, I've been thinking about trying to add my RX 480 into my Desktop for macOS and then hiding my 1080ti via the Nvidia bogus class-code trick which is also used to hide Nvidia cards on laptops with Optimus. This method uses the code from this link along with Inject Nvidia set to true to force the VESA driver to ignore the Nvidia card, bypassing the gioscreenlockstate bug in High Sierra when a headless dGPU (as in Optimus) is present:

https://github.com/RehabMan/OS-X-Cl...mmit/9b8610f32098ce224af044c4e482d2b3da2abae0

In case this helps anyone attempting this. I'll probably wait to do it until 10.13.4 in hopes that it might fix the gioscreenlockstate bug and make the hiding unnecessary . . .
 
@Gigamaxx I am trying to install a RX580 in MAC for daily usage, gaming. also a 1080TI in MAC for cuda computing. Do you know if there're any fixes that could let this setup work in 10.13.5? ( AMD is primary card and nvidia is just for cuda), checked with Scoot1204 before, it seems that AMD card won't have any signal if Nvidia web driver is installed.

Thanks in advance!
 
Not sure about the web drivers issue. In early days we used Nvidia card in slot 1 and RX card in slot 2. But we didn’t use the web drivers so the Nvidia acted as an IGPU.
Maybe you could partition your drive and install OS on each partition one with no Nvidia drivers and the other with them? So you would boot into the web version for cuda and the other for AMD.
 
Hmm, Thanks @Gaigamaxx, that is one option but it's a bit troublesome as I need to reboot everytime and also need to sync files between 2 OSes..

I'd hope to find a way to work around the web driver issues with AMD cards.
 
I just purchased an AMD Radeon Vega Frontier Edition to go along with my EVGA Nvidia Geforce GTX 1070 on MacOS 10.13.5.

I installed the Vega FE in the first slot alone and everything worked fine over DisplayPort.

I then tried to add the Nvidia 1070 in the second slot and it booted with the DisplayPort cable still connected to the AMD. Could see the video up until the progress bar got to the end, probably when loading the Nvidia drivers, then the screen went black.

I was able to VNC in and see that everything loaded and was showing a desktop but no video coming out of the DP of the AMD.

I then pulled the DP cable out of the AMD and put it in the Nvidia and was able to see the desktop.

So at this point, I have both cards installed and able to run Geekbench on both but am only able to see the video when plugged into the Nvidia.

Not sure what to do next to get the AMD pushing FreeSync to my Monitor and the ability to also use CUDA.

2018-07-07_12-02-32.png 2018-07-07_12-02-42.png

2018-07-07_12-07-14.png
 
I just purchased an AMD Radeon Vega Frontier Edition to go along with my EVGA Nvidia Geforce GTX 1070 on MacOS 10.13.5.

I installed the Vega FE in the first slot alone and everything worked fine over DisplayPort.

I then tried to add the Nvidia 1070 in the second slot and it booted with the DisplayPort cable still connected to the AMD. Could see the video up until the progress bar got to the end, probably when loading the Nvidia drivers, then the screen went black.

I was able to VNC in and see that everything loaded and was showing a desktop but no video coming out of the DP of the AMD.

I then pulled the DP cable out of the AMD and put it in the Nvidia and was able to see the desktop.

So at this point, I have both cards installed and able to run Geekbench on both but am only able to see the video when plugged into the Nvidia.

Not sure what to do next to get the AMD pushing FreeSync to my Monitor and the ability to also use CUDA.

View attachment 339698 View attachment 339699

View attachment 339704

Have you tried it with the card slots reversed?
 
Swapping the PCI slots didn't help either.

I tried updating to 10.13.6 hoping it would have some improvements but had trouble getting it to boot.

Had to wait for Nvidia to release the new driver and install with webdriver.sh, old drivers would not work, although totally uninstalling the web drivers would allow the Vega to stay on all the time and use it as primary but Nvidia was useless cause CUDA couldn't detect the card.
 
@hugheba Appreciate your updates! I have the same use case as you, do you have any luck in 10.13.6?
 
Status
Not open for further replies.
Back
Top