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

[FreddyA]

Could be slowdown by effects like Filmconvert.
But my personal experience with Final Cut Pro X and custom hardware is mixed.

On my system (8700k with Vega64 and iMacPro1,1) I do a lot of multicam work (4K mixed 25p and 50p) and cant do this without proxies.
When I plug the same SSD with the project into my MacBookPro 2015 (USB3), I can edit realtime without proxies.
Some more weird things: When I remove my Vega64 and use the iGPU, I can edit the multicam without proxies.
But removing Vega64 doubles the export time. Also tried using both iGPU and dGPU and even normal editing was far from smooth.
BTW: BruceX export with Vega64 to h264 is 10 seconds.

Is this general for FCPX and custom hardware?

 

[rj510]

IMPORTANT: ******Please make sure you backup your existing drive/EFI folder before you attempt any changes **********

Adding on

BIOS Version F7
Connected to monitor via HDMI cable

For Machines with IGPU only

• iMac19,1
• AptioMemoryFix
• Slide=0
• EmuVariableUefi
• VirtualSMC
• IGPU enabled. ID 3E9B0007 (set manually). Framebuffer patch = yes
• DGPU in slot 2 disabled (not connected to power)
• Clover 5018
• Clover rc scripts not tested in 19.1. (boot error if installed in 1.1)
• PCIe cards
  • Fenvi FV-T919
  • DGPU (but disabled, not connected to power)

For Machines with DGPU

• iMac19,1
• AptioMemoryFix
• Slide=0
• EmuVariableUefi
• VirtualSMC
• IGPU enabled. ID set automatically. Framebuffer patch = no
• DGPU in slot 2
• Clover 5018
• Clover rc scripts not tested in 19.1. (boot error if installed in 1.1)
• PCIe cards
  • Fenvi FV-T919
  • DGPU RX 590 Saphire Nitro Pulse

Sample EFI available in post #5,939.
There are 2 config.plist - the default one for the IGPU only(so with the Framebuffer patch); the other one is for the DGPU.

I did get AptioMemoryFix.efi to boot using the iMac19,1 SMBIOS. The key to a successful boot seemed to hinge on two parts:

1. removing RC scripts
2. disabling the iGPU.

So while iMacPro1,1 may be a cleaner system without an iGPU, the requirement for AptioMemoryFix.efi is a disabled iGPU, not a SMBIOS setting.

Which still begs the question, why does the presence of the iGPU (enabled, that is) require OsxAptioFix2Drv-free2000.efi on this mobo?

Are some other settings in BIOS needed to allow AptioMemoryFix.efi with iGPU enabled?

 

[madmax559]

Fenvi T919 - USB Splitter per earlier posting from Casey

https://www.amazon.com/gp/product/B076Q8685Y/?tag=tonymacx86com-20

He suggested the above & I ordered it & am using it with no issues with the F_USB header on the z390 motherboard

One cable goes to the bluetooth on the Fenvi card & the other goes to the NZXT x52 cpu cooler

 

[manHands]

Which still begs the question, why does the presence of the iGPU (enabled, that is) require OsxAptioFix2Drv-free2000.efi on this mobo?

With iGPU enabled, there is less memory available to load the kernel. You can see how this changes in the memory map using the EFI shell with iGPU enabled/disabled. That's also why -free2000 is needed as additional PCI-e devices are added. And also why some using VirtualSmc.efi vs FakeSMC.efi are seeing more success with AptioMemoryFix.efi; FakeSMC requires more memory when loaded in the kernel.

 

[rj510]

With iGPU enabled, there is less memory available to load the kernel. You can see how this changes in the memory map using the EFI shell with iGPU enabled/disabled. That's also why -free2000 is needed as additional PCI-e devices are added. And also why some using VirtualSmc.efi vs FakeSMC.efi are seeing more success with AptioMemoryFix.efi; FakeSMC requires more memory when loaded in the kernel.

Are you talking about DRAM (my system has 64GB)?

Will the system boot using AptioMemoryFix, and iGPU enabled, if the DVMT Pre-Alloc (and the DVMT Total Gfx Mem) settings in BIOS are increased?

And do the DVMT Pre-Alloc and DVMT Total Gfx Mem settings use DRAM or memory internal to the CPU?

(I'm at work and cannot test these things, but it sounds like something to look into.)

 

[svandive]

Are you talking about RAM (my system has 64GB)?

Will the system boot using AptioMemoryFix, and iGPU enabled, if the DVMT Pre-Alloc (and the DVMT Total Gfx Mem) settings in BIOS are increased?

And do the DVMT Pre-Alloc and DVMT Total Gfx Mem settings use RAM or memory internal to the CPU?

(I'm at work and cannot test these things, but it sounds like something to look into.)

I'm not sure if it's of any value but I can confirm, my system with the iGPU disabled can boot, and is booted using AptioMemoryFix-64.efi.

My SMBIOS is set to iMacPro1,1

 

[Shiitaki]

I'm getting barely above 150% (of a theoretical 800%?) CPU utilisation in FCPX and it doesn't seem to use hyper-threading at all. Render speeds are about 30-40% slower than what comparable systems get in BruceX. Usually you find 12-13s for a Vega 56 with i9-9900k and I'm getting around 16-20s. It's still blazing fast, which is why it doesn't really bother me too much, but do you have any ideas what that could be?

Geekbench results are perfectly in line with what to expect, so it's a FCPX-specific issue.

Uh, isn't that what you would expect if the GPU was doing the work and not the CPU? The question is which GPU is doing the work? I would expect you have integrated CPU graphics enabled since final cut pro can use it.

 

[Shiitaki]

I'm not sure if it's of any value but I can confirm, my system with the iGPU disabled can boot, and is booted using AptioMemoryFix-64.efi.

My SMBIOS is set to iMacPro1,1

Hello guys, I had a problem where I could not boot my Z390 Ultra board with the iGPU enabled, but disabled it would boot fine. The solution for me was change out AptioMemoryFix for OsxAptioFix2Drv-free2000.efi

The symptoms would be I set IGPU to angled and then I would get a stop sign.

Apologies for interrupting your guys chat but it seemed relevant.

oh, yes the integrated GPU uses system ram. I have the understanding that OS X will override those settings once it's booted anyway, so it's not like you are wasting ram. It definitely uses system ram. The error is in allocating memory, all devices and all system memory is mapped to a single address space. You should also make sure Above 4G decoding is enabled.

 

[rj510]

Hello guys, I had a problem where I could not boot my Z390 Ultra board with the iGPU enabled, but disabled it would boot fine. The solution for me was change out AptioMemoryFix for OsxAptioFix2Drv-free2000.efi

The symptoms would be I set IGPU to angled and then I would get a stop sign.

Apologies for interrupting your guys chat but it seemed relevant.

oh, yes the integrated GPU uses system ram. I have the understanding that OS X will override those settings once it's booted anyway, so it's not like you are wasting ram. It definitely uses system ram. The error is in allocating memory, all devices and all system memory is mapped to a single address space. You should also make sure Above 4G decoding is enabled.

Yes, this is exactly the point. Why does an enabled iGPU require OsxAptioFix2Drv-free2000.efi?

If it's a memory issue (due to an enabled iGPU and how many PCIe slots are being used), can we modify the system to keep the iGPU enabled, have a functionally useful computer (without limiting our usage of PCIe slots) and yet use AptioMemoryFix?

 

[FriFlo]

@Fabrizzioa04 @FriFlo

You can both benefit from the same instructions so here goes:

How to replace FakeSMC with VirtualSMC:

1. Make a full bootable backup of the current system. This is your insurance.
2. First we need to remove FakeSMC from 2 places:
   • Place #1: /Library/Extensions folder
     • Use Finder to move FakeSMC.kext and all of the FakeSMC_Sensor*.kext files (should be 5 of them) to your Desktop or Documents or Downloads folder. Just get them out of /Library/Extensions, but save a copy so you can restore them later if needed.
     • The run Kext Utility to rebuild kernel cache.
   • Place #2: CLOVER/kexts/Other folder
     • Mount EFI Partition of the Mojave SSD using "Mount EFI" option in Clover Configurator (on the left side panel).
     • An "EFI" disk icon will appear. Go to the EFI/CLOVER/kexts/Other folder.
     • Move FakeSMC.kext and any FakeSMC_Sensor*.kext files to some location on your Mojave SSD (Desktop, Documents, Downloads).
     • Then open the EFi/CLOVER/drivers64UEFI folder (or EFI/CLOVER/drivers/UEFI for Clover v4988 or newer) and move SMCHelper.efi to the same location on your Mojave SSD.
3. Now we need to download VirtualSMC.
   • Download the latest 1.0.6 release by clicking here. Specifically, get the file called 1.0.6.RELEASE.zip
   • MacOS should automatically unzip the files into a folder called "1" or "1-1" or something like that.
4. Install VirtualSMC as follows:
   • Make sure the EFI Partition of Mojave SSD is still mounted (see Step 2, Place #2).
   • Now take a look at the folders and files in the VirtualSMC folder that you just downloaded. It should look like this:
     View attachment 417385
   • Copy the green box item (VirtualSmc.efi) to the CLOVER/drivers64UEFI folder (or CLOVER/drivers/UEFI for Clover v4988 and newer).
   • Copy the red box items to CLOVER/kexts/Other.
5. Reboot.

In my case, it unfortunately did not help. I did the modifications as written. With the 64gb kit it still worked, but with the 128gb kit installed I got "Couldn't allocate runtime area" as a result. Any other things to try? by the way: this was a basic installation without the post installation performed. It is my understanding that you usually do the installation from scratch with all hardware as the intended system shall have. Or should I rather have installed everything with the working ram kit and done the modifications after that?