Contribute
Register

[SUCCESS] Gigabyte Designare Z390 (Thunderbolt 3) + i7-9700K + AMD RX 580

@CaseySJ Thanks for replying.
I said I tried, it's quite extensive guide, so I guess that did got under my radar. Regardless, i used other ports.
As for the link you shared, I don't see an option to start a terminal on that error screen. Menu options are frozen until I click the restart button. After that I have this:

View attachment 421303View attachment 421304
We need to press the following 5 keys when "Installer Resources Could not be Found" occurs:

Shift - Control - Alt/Option - Command - H

This should bring back mouse control and from the top Menu Bar we can select Utilities --> Terminal and then carry out a couple of additional commands. Please see posts 13 and 16 here:
If you need help with those commands, please let me know.
 
Has anyone had issues with Firefox crashing the whole machine? Whilst browsing the mouse pointer becomes slow laggy, like a memory issue, and then freezes. I have to physically press the off button on the machine, and then on restart sometimes doesn't boot, and have restart again.
I've been using the latest Firefox for about an hour across numerous websites and it has been completely stable. If you have a number of Firefox extensions, try disabling them. Is there a particular website that crashes the browser more regularly? Are Safari and Chrome stable?
 
After experimenting with the Asus X99, my current thinking on this is as follows:

Observations from Asus X99 Deluxe II and Falcon Ridge TB2 Controller
  • MacOS initializes the Thunderbolt Bus and Thunderbolt Local Node device trees in IORegistry without the need for an SSDT or EFI drivers.
  • The PCI and Thunderbolt driver families in macOS are sufficient for this purpose.
  • However, certain Thunderbolt NHI device path properties are not injected by the kexts:
    • ThunderboltDROM
    • ThunderboltUUID
    • pathcr
    • ThunderboltConfig
  • These properties, we believe, are injected by Apple's ThunderboltNHI EFI Driver during boot. That driver does more than just inject these properties because simply copying-and-pasting them via a custom SSDT does not seem to have any effect.
  • Hot plug does not work. On Titan Ridge and Alpine Ridge (Thunderbolt 3) it is sufficient to inject 2 properties in the SSDT:
    • PCI-Thunderbolt = 1 --> inject into UPSB
    • PCIHotPlugCapable = 1 or 0 --> inject into DSB0
    • But on Thunderbolt 2, this is not sufficient.
  • Thunderbolt devices do function when connected during boot.
  • For eGPU to work during boot, we must remove PCI-Thunderbolt property from UPSB. On a real Mac, this property is never removed, so there's more going on behind the scene.
  • Connecting this Hack directly to a Mac over a Thunderbolt 2 cable has so far been unsuccessful in establishing a Thunderbolt link between the two.
    • Still testing/experimenting with this.
  • Looking at the SSDT from a MacBook Air 7,1 (with Falcon Ridge chip) we see that a number of Thunderbolt ACPI methods rely on GPIO registers.
    • These GPIO registers and their physical mappings are most likely native to the design layout of the MacBook Air circuit board and cannot simply be copied-and-pasted.
    • Apple might be using these registers to aid in TB2 hot plug capability and in to provide a more complete Thunderbolt implementation.
Implications for Titan Ridge Thunderbolt 3
  • On all currently shipping Macs, Thunderbolt 3 is implemented with a combination of Alpine Ridge and Titan Ridge controllers.
    • Alpine Ridge controls UPSB
    • Titan Ridge controls DSB0 and everything else
  • From the experience on Asus X99, we know that the Thunderbolt Bus and Thunderbolt Local Node device trees can be created by macOS drivers alone, without assistance from EFI drivers.
    • But those kexts seem to be querying the Thunderbolt controller directly to determine their Device and Vendor IDs.
    • My guess is that macOS drivers will not create the Thunderbolt Bus and Local Node devices if they don't specifically see an Alpine Ridge controller at UPSB.
    • So I already tried to spoof the UPSB Device ID in the Thunderbolt SSDT by injecting the Alpine Ridge ID.
      • It didn't work.
    • So another theory is that Apple might be using some special circuit/chip to integrate the two Thunderbolt controllers, and macOS drivers could be specifically looking for this circuit/chip.
    • However, it may be possible to patch the appropriate kext so that it bypasses the check for either (a) Alpine Ridge at UPSB, or (b) a special circuit/chip. To do this, we need to:
      • Determine which kext is looking for this trigger
      • Disassemble the kext and determine the hex patch
    • But then again, experience from the Asus X99 tells us that just obtaining the Thunderbolt Bus and Local Node device trees is insufficient because we still need an EFI Driver to inject:
      • ThunderboltDROM
      • ThunderboltUUID
      • pathcr
      • ThunderboltConfig
Most owners of this system do not really need Thunderbolt Bus and Local Node. But if this is something you really do need, then a real Mac is the only solution at the moment.


Thanks for this. I've suspected such, but didn't have the means to explain it as you've done. This is why the Thunderbolt 2 connected UAD Apollo 8 won't reconnect after powering-off the interface. Powering it back up and the Thunderbolt 2 driver won't connect to the device. UAD sells a user-installable Thunderbolt 3 card to replace the Thunderbolt 2 card: https://www.uaudio.com/audio-interfaces/thunderbolt-3-option-card.html
but I'm not convinced that it would "fix" the issue. But if it did for my Z370 build (see sig) I'd consider it. Onward and upward improvements, etc. Thanks again.
 
@CaseySJ I managed to install it, but I accidentally reboot the system before this:
WARNING: If you're not using an external PCIe graphics card, but are relying entirely on the onboard Intel iGPU with an HDMI connection, then you must apply the framebuffer patch described in the section Using the UHD 630 to Drive HDMI and DP Monitors. Failure to do so will result in a blank screen when you reboot.

Is there a way to recover from the black screen? Also this hole patching the framebuffer is very confusing. Do I need to determine the proper Device ID and Platform ID even using IGPU on 10.14.6? (native support?) How do I simply apply the pre-compiled ones?
 
@CaseySJ I managed to install it, but I accidentally reboot the system before this:


Is there a way to recover from the black screen? Also this hole patching the framebuffer is very confusing. Do I need to determine the proper Device ID and Platform ID even using IGPU on 10.14.6? (native support?) How do I simply apply the pre-compiled ones?
Did you try booting off of your USB drive ?
 
Another experimental question:
Can we have dual dGPUs in iMac ( SMBIOS 19,1 ). If yes, can it be two different GPUs, for example, VII and RX480 ? I saw few people were successful having different memory sizes.
 
Did you try booting off of your USB drive ?
Do you booting into the installer? Or using the USB drive to boot into Mojave? I assume the later, but wouldn't that load that drive's config.plit?

Regardless, I found a forgotten Thunderbolt 3 to DisplayPort. Managed to login. How do I apply the pre-compiled framebuffer patch? I tried reading the guide, but it's confusing.
 
Do you booting into the installer? Or using the USB drive to boot into Mojave? I assume the later, but wouldn't that load that drive's config.plit?

Regardless, I found a forgotten Thunderbolt 3 to DisplayPort. Managed to login. How do I apply the pre-compiled framebuffer patch? I tried reading the guide, but it's confusing.
There is a link in the guide which shows an example using Hackintool. Its in the spoiler where the postboot is described.

PS: sorry using a phone, othewise would have pasted the link out here.
 
Back
Top