Intel UHD Graphics 630 Coffee Lake R on macOS

How to set the integrated graphics card Intel UHD Graphics 630 Coffee Lake R (9th gen. i7 and i9) in headless mode (no cable to monitor) to be used by macOS (Catalina, Big Sur, Monterey or Ventura) in computing and video encoding tasks or setting it as the main card

Note: based on

• [GUIDE] General Framebuffer Patching Guide (HDMI Black Screen Problem) by CaseySJ in tonymacx86 forum
• Framebuffer patch feature of headkaze's Hackintool app
• Desktop Coffee Lake chapter of the OpenCore Dortania's guide.

Preface

Macs with an integrated graphics card (iGPU) and a dedicated one (dGPU) use the iGPU for video encoding and decoding tasks. When building a Hackintosh with both types of GPU we can find that, if the iGPU is not properly installed, video encoding can fail. When this happens, we must configure the iGPU as headless mode (it is so called when it is enabled but no cable to monitor) so that the dGPU acts as main card but the iGPU is available for encoding / decoding video.

iGPU setting depends on 2 factors:

Motherboard because manufacturer can put 1, 2 or 3 HDMI connectors for the iGPU on the back pannel

Intel processor generation, different Intel generations have different iGPU

My PC has a Z390 Aorus Elite board with an Intel 9th gen. CPU (Coffee Lake Refresh, it is configured as Coffee Lake) with Intel UHD Graphics 630:

PCI path is PciRoot(0x0)/Pci(0x2,0x0)

Plattorm ID is 3E9B0007

On this board there is only one HDMI v1.4 connector for the iGPU, it corresponds to index 3 in the theoretical list of 3 internal connectors this iGPU can have:

Index 1, BusID 0x00, Type HDMI (type does not matter on this port)

Index 2, BusID 0x00, Type HDMI (type does not matter on this port)

Index 3, BusID 0x04, Type HDMI (this is the active port, the only physical one, type is HDMI)

This is important when using the iGPU as a main or single card but not when using it in headless mode.

Headless mode

• iGPU and dGPU must be enabled in BIOS with dGPU as primary
• There should be no cable between iGPU HDMI port and any type of display
• Lilu and WhatEverGreen properly installed
• SMBIOS iMac19.1.

You have to add in DeviceProperties >> Add:

<key>PciRoot(0x0)/Pci(0x2,0x0)</key>
<dict>
	<key>AAPL,ig-platform-id</key>
	<data>AwCRPg==</data>
	<key>device-id</key>
	<data>mz4AAA==</data>
	<key>enable-metal</key>
	<data>AQAAAA==</data>
</dict>

This code has data values in Base64, in plist editors they can be seen as hexadecimal, e.g. AwCRPg== in Base64 (AAPL,ig-platform-id) = 0300913E in hexadecimal.

With these changes you can boot from a dGPU with the iGPU well installed. To check if the VDA Decoder function is activated you can get Hackintool app (Fully Supported or Failed in the first System tab).

Notes:

• AAPL,ig-platform-id=0300913E is mandatory, to detect the iGPU as Coffee Lake
• device-id=9B3E000 to be displayed as Intel UHD Graphics 630 instead of Kabylake Unknown (optional)
• enable-metal=01 to enable Metal 3 in Ventura (recommendable on macOS 13).

Image: iGPU as secondary card

iGPU as main card

This card can also be configured to be the main or single one, so that it outputs a signal to the monitor and also encodes video. Here's what to do.

• Enable it on the motherboard as main card: Initial Display Output IGFX instead of PCIe 1 Slot (this would be the final step)
• Lilu and WhatEverGreen properly installed
• SMBIOS iMac19.1
• Add in config.plist >> DeviceProperties >> Add the code below (note: BwCbPg== is 07009B3E in hexadecimal):

<key>PciRoot(0x0)/Pci(0x2,0x0)</key>
<dict>
	<key>AAPL,ig-platform-id</key>
	<data>BwCbPg==</data>
	<key>device-id</key>
	<data>mz4AAA==</data>
	<key>device_type</key>
	<string>VGA compatible controller</string>
	<key>enable-hdmi20</key>
	<data>AQAAAA==</data>
	<key>enable-metal</key>
	<data>AQAAAA==</data>
	<key>framebuffer-con0-busid</key>
	<data>AAAAAA==</data>
	<key>framebuffer-con0-enable</key>
	<data>AQAAAA==</data>
	<key>framebuffer-con0-pipe</key>
	<data>EgAAAA==</data>
	<key>framebuffer-con1-busid</key>
	<data>AAAAAA==</data>
	<key>framebuffer-con1-enable</key>
	<data>AQAAAA==</data>
	<key>framebuffer-con1-pipe</key>
	<data>EgAAAA==</data>
	<key>framebuffer-con2-busid</key>
	<data>BAAAAA==</data>
	<key>framebuffer-con2-enable</key>
	<data>AQAAAA==</data>
	<key>framebuffer-con2-pipe</key>
	<data>EgAAAA==</data>
	<key>framebuffer-con2-type</key>
	<data>AAgAAA==</data>
	<key>framebuffer-patch-enable</key>
	<data>AQAAAA==</data>
	<key>framebuffer-stolenmem</key>
	<data>AAAwAQ==</data>
	<key>hda-gfx</key>
	<string>onboard-1</string>
	<key>force-online</key>
	<data>AQAAAA==</data>
</dict>

Note:

• force-online=01 is mandatory, to force online status on all displays.

In this way, Intel UHD 630 is well installed and works fine on macOS.

Image: iGPU as main card

Note about igfxfw and rps-control (thanks 5T33Z0)

Many Coffee Lake iGPUs work fine without loading Apple GUC firmware or enabling RPS Control. But those who have graphical issues (low max frequency, fixed frequency that does not vary, screen glitches, etc.) can play with 2 available properties:

• Apple GUC (disabled by default): it can be enabled by igfxfw property (02 as Data) or igfxfw=2 boot arg. It forces loading of Apple Graphics Unit Control (GUC) firmware (graphics accelerator uses Apple firmware scheduler). It improves Intel Quick Sync Video and iGPU performance. Ir requires chipset supporting Intel Management Engine v12 or newer (H310, C246, B360, H370, Q370, Z390, Z490, etc.). It's buggy and not advisable to use. Should be tested on a case by case basis.
• RPS Control (disabled by default): it can be enabled by rps-control property (01 as Data) or igfxrpsc=2 boot arg. It enables RPS control patch (graphics accelerator uses host preemptive scheduler). Improves iGPU performance on Kaby Lake and newer using older chipsets without Intel Management Engine v12 support (Z370 and others). rps-controlproperty was enabled in WhateverGreen until macOS 10.15.6, in this version it had major issues and, since then, WhateverGreen has disabled it by default.

Recommendations:

• If the iGPU has normal frequencies and power consumption for that model and the screen works fine >> it is not necessary to use any of the 2 settings
• Both should not be used at the same time because they are based on different modes of operation
• igfxfw takes precedence over igfxrpsc when both are set
• Boards supporting Intel ME 12 and newer can try igfxfw=2
• For boards with older chipsets, RPS-Control is an option
• There are comments of OC developers saying that rps-control is better than igfxfw (Big Sur and newer) because of the bug that seems to be in Apple GUC.

Note about KP waking from sleep

If you have KP or black screen when macOS wakes from sleep, you have to replace hda-gfx property with No-hda-gfx, this usually fixes those KPs but audio through HDMI is lost. Replace:

<key>hda-gfx</key>
<string>onboard-1</string>

with:

<key>No-hda-gfx</key>
<data>AAAAAAAAAAA=</data>

Note about QEMU

If macOS is being installed in a QEMU virtual machine, the primary card in BIOS must be set to dGPU. Otherwise vesafb grabs the iGPU which, due to a memory allocation conflict, results in a long delay on startup/shutdown and multi-gigabyte logs.

If vga=off is added to grub, then the card is not claimed, however it also does not work (blank screen).