Spoofed eGPU Causes Kernel Panic

[Fortitude]

I cannot for the life of me get my external graphics card working properly with macOS Ventura. It’s an AMD Radeon RX 6950 XT connected to an Alienware Graphics Amplifier which is then connected to my Alienware Area 51m R2 laptop. I’ve managed to create a bridge device and spoofing SSDTs and also tried using the device properties to spoof the device-id. I’ve even tried flashing my graphics card with the 6900 XT vBIOS.

 

However, I always get a page fault whenever the external graphics card is plugged in at boot initialization and never reach the login screen. I know that the card is getting recognized because my external monitor brings me to a black screen. In case this is an important point to mention, the amplifier only started working properly in Windows after applying the HackFlags DWORD to the registry.

 

 

 

SSDT-BRG0.aml SSDT-GPU-SPOOF.aml

Edited April 13, 2024 by Fortitude

[niknod]

Have you tried the NootRX.kext way? 

[Fortitude]

I've just tested it out, but unfortunately, it doesn't appear to work either. Come to think about it, I’m not even sure if macOS sees this external GPU as an eGPU because the Alienware Graphics Amplifier connects over PCIe instead of Thunderbolt. It looks like the operating system sees this as a second internal GPU along with my AMD Radeon RX 5700M

.

 

Edited April 13, 2024 by Fortitude

[Fortitude]

I’ve been doing some research and found this video where a user with the original Alienware Area 51m managed to make use of an amplifier with his Hackintosh. This was before Dell allowed you to configure the system with an AMD graphics card, hence, an amplifier was used to get a display output. This makes me wonder if the issues that I’m having may be related to WEG auto-renaming the GPUs? I'm not sure if GFX1 is a valid name for the graphics, and if I try calling it GFX0, the eGPU still gets called GFX1 in the end. (Apparently, eGPUs in macOS are called EGFX.) Does anyone know if it’s possible to prevent WEG from auto-renaming all of the GPUs? I can’t disable WEG because I need agdpmod=pikera to prevent a black screen.

[Fortitude]

Hi everyone,

 

I’m still grappling with this issue and wanted to revisit the topic with a new thought.

To recap this topic, my laptop has an internal 5700M connected over PCIe, and I'm connecting a 6950XT (spoofed as a 6900XT) also over PCIe through an external enclosure. Connecting the 6950XT causes the framebuffer to panic (see the screenshots in the first post).

 

I hope this isn’t a silly question, but does anyone here have a custom Hackintosh desktop with two Navi GPUs running at the same time? Most of the dual-GPU configurations I’ve seen online for Hackintosh systems involve pairing an AMD GPU for macOS with an NVIDIA GPU for Windows/Linux.

Edited October 17, 2024 by Fortitude

[Baio77]

https://github.com/Baio1977/Intel-NUC8i7HN 

I purchased an EGPU dock using an RX 6600XT. I'm having the same problem. The NUC has a Vega GPU and sees my EGPU as GFX1. I also generated SSDT and used GFX0 for the EGPU, but the problems remain the same.

iMac EGPU .ioreg

I attach debug log

LiluLog.logBoot Debug NUC EGPU.txtWhateverGreenLog.log

EDIT:
I also tested the dock on the X1 Tablet 3rd Gen https://github.com/Baio1977/Lenovo-ThinkPad-X1-Tablet-Gen3 

 

X1 tablet + EGPU.ioreg
 

 

As you can see, things are different on Lenovo Tablets.
I think the NUC has a Vega and Navi GPU conflict.
In the case of the X1 Tablet, however, I think it is a different problem, if you notice here the display0 is missing

 

 

I think it is on this Tablet that I have to work to understand what is missing to have video output from EGPU

Edited August 3, 2025 by Baio77

[Fortitude]

Wanted to let everyone know that I finally solved this issue after throwing the kitchen sink at it for months. It turns out that my laptop has a serious firmware issue where it fails to properly initialize the eGPU during POST. As a result, the card never gets enumerated, no MMIO space is reserved, and macOS doesn’t know how to properly map the device.

 

I had to use a device property called PCI-Thunderbolt to get the card working properly. It appears to force macOS to dynamically allocate the required resources instead of relying on firmware. This device property only seems to work on my machine with macOS Sequoia and Tahoe.

 

			<key>PciRoot(0x0)/Pci(0x1,0x1)/Pci(0x0,0x0)/Pci(0x0,0x0)/Pci(0x0,0x0)</key>
			<dict>
				<key>PCI-Thunderbolt</key>
				<true/>
				<key>PP,PP_WorkLoadPolicyMask</key>
				<integer>1</integer>
				<key>ATY,DeviceName</key>
				<string>6950 XT</string>
				<key>ATY,FamilyName</key>
				<string>Radeon RX</string>
				<key>device-id</key>
				<data>v3MAAA==</data>
				<key>model</key>
				<string>AMD Radeon RX 6950 XT</string>
				<key>hda-gfx</key>
				<string>onboard-3</string>
			</dict>

 

Additionally, applying this device property to my Thunderbolt controller (Intel JHL7540) seems to be sufficient to get everything working. No SSDT or custom firmware flashing required.

 

			<key>PciRoot(0x0)/Pci(0x1C,0x4)/Pci(0x0,0x0)</key>
			<dict>
				<key>PCI-Thunderbolt</key>
				<true/>
			</dict>