How to build your own iMac Pro [Successful Build/Extended Guide]

[Bagi1992]

#762

I have the same issue... I used the attached EFI folder, I checked the guide many times... I did the same AGDPfix.app method and I get the same reboot every time. I am still trying now!

P.S. How did you fix the boot issue? I managed to boot with the same bootflags as the attached config.plist, but I had to add UseKernelCache=No npci=0x2000 -f -v

 

[dcheng]

#762

I have the same issue... I used the attached EFI folder, I checked the guide many times... I did the same AGDPfix.app method and I get the same reboot every time. I am still trying now!

P.S. How did you fix the boot issue? I managed to boot with the same bootflags as the attached config.plist, but I had to add UseKernelCache=No npci=0x2000 -f -v

I have a USB that also boots (if you read my other posts earlier). I booted on the USB and removed everything I did. So basically I reverted back to preinstall.

 

[Bagi1992]

I have a USB that also boots (if you read my other posts earlier). I booted on the USB and removed everything I did. So basically I reverted back to preinstall.

Mate, just do the other method, I have GPU acceleration now!

b.) Only in case that the AGDPfix for what reason ever would not apply, install NvidiaGraphicsFixup.kext v1.2.0 in
/EFI/CLOVER/kexts/Other of your EFI partition instead. If you choose this black screen prevention, you also need Lilu.kext v1.2.0 in /EFI/CLOVER/kexts/Other of your EFI partition, because NvidiaGraphicsFixup.kext is implemented as a Lilu-plugin.

After that, just install the WebDriver and reboot. Should work for you too!

Now I can do the rest post install...

 

[send2moran]

@kgp
any plans to support asus x299 delux sensor information ?
how can i help ?

 

[Bagi1992]

@kgp

I still noticed some GUI lag in 10.13, is that an un optimised OS GUI issue? Or something is still not working perfectly for me?

For exemple in Launchpad, if I open the "other folder" I feel some fps drops in GUI. I remember, with OS X El Capitan, there was some know GUI lagg until 10.11.4 .... do you notice the same?

Thanks for your work man!

 

[kgp]

another observation,
after booting to windows ssd restarting and booting macos on other ssd
i'm getting a kernel panic, the only option to boot again is to remove the nvidia drivers and restart..
idk why that happens ..

Uffff!!! Something is definitely wrong with your system installation/configuration!

I can confirm full functionality of dual boot systems! I would not find any reason why windows should affect the OSX web driver performance! This is totally unlogical, sorry!

 

[kgp]

@kgp
any plans to support asus x299 delux sensor information ?
how can i help ?

No own plans so far.. you can help in finding working alternatives to HWMonitor and HWSensors..

 

[kgp]

@kgp

I still noticed some GUI lag in 10.13, is that an un optimised OS GUI issue? Or something is still not working perfectly for me?

For exemple in Launchpad, if I open the "other folder" I feel some fps drops in GUI. I remember, with OS X El Capitan, there was some know GUI lagg until 10.11.4 .... do you notice the same?

Thanks for your work man!

Good to see that you finally achieved a fully working Skylake-X/X299 System in line with reasonable benchmarks!

I do not notice any workspace performance drop on my system..

All the best,

KGP

 

[TheOfficialGypsy]

@TheOfficialGypsy , @ALL ,

the PC0 -> PCI replaces can indeed only be performed for the populated PCI slots of the ASUS Prime X299 Deluxe! If your perform PC0 -> PCI replacements of unpopulated PCI slots your rig won't boot!

In my case PCI0, PCI2, and PCI3 are populated.

View attachment 285912


Thus, in my case I had had to perform DSDT replacement patches PC00 -> PCI0, PC02 -> PCI2, and PC03 ->PCI3, in contrast yo your initial instructions instruction to implement DSDT replacement patches PC00 -> PCI0, PC01 -> PCI1, PC02 -> PCI2, and PC03 ->PCI3!

However, yet I still ask myself how the proper PC0 -> PCI replacement relates with AppleALC.kext. Could you please explain the latter relation, if there is any? My suspicion is that there is no relation at all and that the PC0 -> PCI replacement is not required for a successful AppleALC audio implementations in contrast to your earlier instructions.

In any case, now the APCI Tables are correctly implemented, the system boots flawless and Apple's System Report correctly reveals:

View attachment 285913

View attachment 285914

View attachment 285917

View attachment 285919

View attachment 285924

View attachment 285923

When using Lilu.kext v1.1.7 and AppleALC.kext v1.1.4, even analogue audio is properly implemented:

View attachment 285925

While Lilu.kext v1.2.0 works just fine with all other related plugins, AppleALC.kext v1.2.0 yields on my system a boot kernel panic system under 10.13 High Sierra. Surprisingly, the combination of Lilu.kext v1.2.0 and AppleALC.kext v1.2.0 works fine under 10.12.6, while neither the ACPI Table nor the SSDT-X299.aml seems to apply!

What could be the reason for the AppleALC.kext v1.2.0 Kernel Panic under 10.13? Any helpful comments would be extremely appreciated! We both have the same board, this it is enological that AppleALC.kext v1.2.0 works fine on your case and fails in my case..

This in conclusion 3 main questions:

a.) How does the proper PC0 -> PCI replacement relates with AppleALC.kext? Could you please explain the latter relation, if there is any? My suspicion is that there is no relation at all and that the PC0 -> PCI replacement is not required for a successful AppleALC audio implementations in contrast to your earlier instructions.

I would need your immediate reply, as currently the AppleALC audio approach is still implemented in my guide with the PC00 -> PCI0, PC01 -> PCI1, PC02 -> PCI2, and PC03 ->PCI3 replacement dependency, which is simply wrong, as first the PC0 -> PCI replacement can very like just successfully applied for ASUS Prime X299 Deluxe, and secondly the PC0 -> PCI replacement cannot be generalised to PC00 -> PCI0, PC01 -> PCI1, PC02 -> PCI2, and PC03 ->PCI3, as the PC0 -> PCI replacement apparently depends on the actual PCIe slot population!

I have to urgently update my guide concerning this open issue! The currently implemented description of the AppleALC audio approach is strongly misleading and wrong! It has to potential to make nearly all systems unbootable!

b.) Do you see any possibility for solving the AppleALC.kext v1.2.0 issue on my system?

c.) Do you see any possibility for extending your ACPI Table nor the SSDT-X299.aml to 10.12.6?

Cheers,

KGP

One thing I cannot understand is you mention your build cannot boot when you rename PC00 - PC05 slots when unpopulated..

In my build I have in PC00 - 1080Ti, PC01 nothing, PC02 nothing, PC03 Thunderbolt card, PC04 BCM94360CD Naitive wifi bluetooth, PC05 nothing

I have all the renames PC00 - PC05 and can boot without an issue..

From my belief AppleALC and CloverALC both rely on the HDEF coming under PCI'X' opposed to PC0'X', as this is the scenario on real Macs, hence why I renamed devices to match those names of real Macs. Even with the CAVS to HDEF rename I believe that both AppleALC and CloverALC require HDEF to show under PCI'X', but this is just my understanding that I gathered from extensive testing.

I would for now remove the AppleALC audio method from the guide and revert to VoodooHDA until further testing has been done and the issue has been rectified.

Due to the nature of this platform, I would strongly recommend anyone looking at experimenting with X299 to have two EFI partitions (1 on boot HD and 1 that is being experimented with on USB) and a clone of your MacOS HD on standby.
This is what I am doing, in the event that my system is unable to build, I simply insert my USB with stable EFI folder and boot from there and revert any changes that were done to the EFI partition of the system HD.

 

[TheOfficialGypsy]

@kgp

I still noticed some GUI lag in 10.13, is that an un optimised OS GUI issue? Or something is still not working perfectly for me?

For exemple in Launchpad, if I open the "other folder" I feel some fps drops in GUI. I remember, with OS X El Capitan, there was some know GUI lagg until 10.11.4 .... do you notice the same?

Thanks for your work man!

I can confirm I experience the same lag / choppiness.

When switching MacOS 'Spaces' the UI is far from smooth and stutters tremendously, also when moving mouse along dock the system stutters and is far from smooth.

I am trying to figure what is causing it but for me it seems to resolve itself once I wake my machine from sleep.

EDIT: After some preliminary testing I have determined the issue must be related to CPU power management
If I click on Safari window and move it around the screen while having Intel Power Gadget running, I notice no increase in frequency Stable 1.2GHz - 1.5Ghz and lagging, very odd. I also notice the frequency stays at around 1.2Ghz for most tasks, opening and browsing finder, using iTunes. Of course if I run benchmark then system runs at 4.4Ghz
Running Geekbench yields some fluctuations however no where near as much as after sleep.

I do the same on my Z170-UD5-TH the frequency jumps around (2 - 3Ghz) and everything is smooth.

I then put my X299 system to sleep, after wakeup I move the Safari window around and observe CPU frequency.. fluctuations between (2-3Ghz) just like the i7 6700k build, also using apps causes frequency to jump around. Of course doing nothing causes frequency to return to 1.2Ghz, but the minute I move window around the frequency responds. No lag whatsoever

Also after running Geekbench after wakeup from sleep, machine has frequency fluctuating a lot and increased result by ~1000 points. 385XX before sleep, 395XX after sleep

This suggests there indeed is some correlation between CPU power management and the laggy UI

EDIT 2: There is no more lag! I decided to review bios settings and noticed Intel Speed Shift was enabled...
Disabling Intel Speed Shift fixed the issue of laggy MacOS

For now until we can use our native CpuID and MacOS gains support for Speed Shift (which is expected really soon) we have to remember to disable this technology