Contribute
Register

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

I have an interesting one. If I boot from my USB Clover install, everything works. If I boot from my internal NVMe, which as worked the entire time, I get a memory allocation error. In fact, it seems the random seed doesn't even attempt to work, as I don't see any of the +++++++ symbols that I see when booting from USB.

However, if I disable the internal graphics in the BIOS, the NVMe based boot works. I have slide=0 enabled, and I am using AptioMemoryFix. The only thing that changed yesterday was that I was doing a LOT of experimenting with headless, non-headless, iGPU on/off, etc. All of this was using the USB to boot, then when I went to go back to my "normal" internal Clover/Boot, it failed.
 
I have an interesting one. If I boot from my USB Clover install, everything works. If I boot from my internal NVMe, which as worked the entire time, I get a memory allocation error. In fact, it seems the random seed doesn't even attempt to work, as I don't see any of the +++++++ symbols that I see when booting from USB.

However, if I disable the internal graphics in the BIOS, the NVMe based boot works. I have slide=0 enabled, and I am using AptioMemoryFix. The only thing that changed yesterday was that I was doing a LOT of experimenting with headless, non-headless, iGPU on/off, etc. All of this was using the USB to boot, then when I went to go back to my "normal" internal Clover/Boot, it failed.
Please check whether CLOVER folder in the NVMe SSD has more than one memory fix driver. If so, remove all others and keep only one. Also try replacing AptioMemoryFix with OsxAptioFix2Drv-free2000.
 
Both of those platform IDs are correct for your system. One is headless and the other is not. Because you have the RX560, we recommend switching the iGPU to headless mode with platform ID 0x3E980003.

In BIOS, Initial Display Output should be set to PCIe Slot 1 where the AMD GPU is installed.

If the system does not boot, please post your config.plist (you may redact serial numbers from SMBIOS).
Temporarily I don't have an RX560. I sent to the service. Then I must use 0x3E9B0007 for boot and full acceleration, right?
 
Please check whether CLOVER folder in the NVMe SSD has more than one memory fix driver. If so, remove all others and keep only one. Also try replacing AptioMemoryFix with OsxAptioFix2Drv-free2000.

Hi Casey,
Thanks for the quick reply! Ive verified no extra memory fix files. I was messing around in Hackintool also, so I wonder if I broke the EFI partition somehow. There sent anything in Hackintool that could have done any register/HW changes to the iGPU when I was messing around, I assume?

As for the free2000 version, I've seen a lot of people shying away from it since the author said it should have never been released to the public. He goes into detail about how it actually can damage the memory footprint, etc. He was very vocal about only using the four different versions of the memory fix file "OSX1,2,3" and the current AptioMemoryFix. Ill tinker with an EFI/ESP re-do, Clover re-install and see how that goes.

Finally, given how this works on the memmap, I wonder if there is a way with these MoBo's to clear the ROM/NVRAM? Is it simply a CMOS reset?
 
Temporarily I don't have an RX560. I sent to the service. Then I must use 0x3E9B0007 for boot and full acceleration, right?
Yes, that's correct. However, simply switching to this platform ID will not enable the HDMI port on the rear IO panel. The two Thunderbolt/DisplayPort ports will work by default so it's a good idea to use a USB-C to DisplayPort cable. However, if HDMI is the only option, then it's necessary to apply a frame buffer connector patch to enable HDMI. A Mini-Guide for that can be found here:
 
Hi Casey,
Thanks for the quick reply! Ive verified no extra memory fix files. I was messing around in Hackintool also, so I wonder if I broke the EFI partition somehow. There sent anything in Hackintool that could have done any register/HW changes to the iGPU when I was messing around, I assume?

As for the free2000 version, I've seen a lot of people shying away from it since the author said it should have never been released to the public. He goes into detail about how it actually can damage the memory footprint, etc. He was very vocal about only using the four different versions of the memory fix file "OSX1,2,3" and the current AptioMemoryFix. Ill tinker with an EFI/ESP re-do, Clover re-install and see how that goes.

Finally, given how this works on the memmap, I wonder if there is a way with these MoBo's to clear the ROM/NVRAM? Is it simply a CMOS reset?
If you like you may compress and upload the CLOVER folder from EFI partition of the Mojave SSD (NVMe). But as always, redact serial numbers from SMBIOS section of config.plist. Re-installing or re-doing the EFI might be a little overkill if the problem is relatively easy to fix.
 
@CaseySJ I've finally build my Hack Pro 5K! I'm incredible grateful for your awesome guide and advice!.

As you know in order to get full 5K I needed to go with another motherboard. So I picked up the Gigabyte Z390 Aorus Pro. So far, everything seems to be working, I'm receiving the Titan Ridge on Monday. So I can't test for now the LG 5K and TB3 connectivity. But there's already 2 users that got that working, so I'm not so worried about it.

Going to sleep seems to work fine, BUT It takes around 20 seconds to shutdown everything (lights, fans, etc). The display goes off immediately after I press Sleep. The first problem I found is that the fans ramp up for a few seconds before they stop. Is this normal or can be fixed?

And my second problem is, when waking up, the monitor is not showing up image up until around 20-25 seconds, which makes it not that unusable.

I'm running a Vega 56 connected through DP port to my 27'' LED Cinema Display on Gigabyte Z390 Aorus Pro. If I try with darkawake=0, the problem improves quite a bit (takes around 5-6 seconds to show image), but still seems it takes quite a few seconds to wake and show image on the display. Anything I can do to improve this?
 
It’s not necessary to change SMBIOS even for the latest 10.14.6 update. Instead, we should delete any RX580 SSDT from CLOVER/ACPI/patched folder and set platform ID to 0x3E980003.

In a multi-GPU system Adobe applications tend to choose the GPU based on alphabetical order of the name. Changing “Vega 64” to “AMD Vega 64” will ensure that it gets selected over the “Intel UHD630”. Please see:

Give these suggestions a try and let us know if you need help with any part.

Thanks! Removing the RX580 SSDT works. Regards to setting platform ID to 0x3E980003, I can just select it in config.plist/graphics/ig-platform-id? I have attached a file for confirmation.

May I know what is this Mini-Guide: Using Hackintool to Set Headless Platform ID 0x3E980003 for? Is this the same as above?
 

Attachments

  • Screenshot 2019-08-31 at 12.14.07 AM.png
    Screenshot 2019-08-31 at 12.14.07 AM.png
    211.3 KB · Views: 70
Last edited:
These scores look fine. You can always experiment with over clocking if higher performance in real world apps is really necessary. I think the multicore scores you’re getting now are nothing to worry about. Also keep in mind that for top-notch performance, the i9-9900K is available now at a discounted price.

The iMac (21.5-inch Retina Early 2019) from Apple, when shipped with i8700k : Single-Core Score 5837 // Multi-Core Score 26572. And I dont think it is overclocked from apple. This got me thinking that I should hit higher numbers than single 5937 - multi 24546 . What do you guys think?
 
In case anyone's missed it, there's a second supplemental install of 10.14.6 uploaded by Apple (18G95)
Maybe it will solve some of those sleep problems, as that's the reason Apple have uploaded it
It updates without any problem
 
Back
Top