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

[digumo]

Oh, that’s a pity. It works great here. Did you do a NVRAM reset in the Clover boot menu (by pressing F11) after replacing the files?

Hot DAMN!!!! I've never had my machine running with the iGPU in headless and working! I've got HEVC now, Quick View with jpegs works and the thing just booted up as normal as it always does.

@CaseySJ I am running off the NVMe in the suggested location for the Mac drive (I forget which one is which).

I did clear dart=0 and my former shiki values from boot args. So, I'm running with the following.

slide=0
darkwake=0
keepsyms=1
debug=0x100
-v

I keep half expecting this thing to start smoking or just crash hard. I'll have to test this config for the next few days. I'm only 15 minutes after initial boot so if there is anything you want me to test just ask. Also of note for anyone else who reads this, I'm not running any extra hardware except for the dGPU. No BT/WiFi. Just running the graphics card in slot 2 because my CPU cooler is too big.

Uhh, this is pretty cool! Thanks @JimSalabim!!!!

 

[digumo]

Ok, so far everything seems to be working fine with FwRuntimeServices.efi and OcQuirks.efi

Sleep works and it comes out of sleep just fine.

...except one thing I can find so far.

It's Final Cut Pro which is now much slower in the timeline. Since I've never had the iGPU running regularly, it's now possibly prioritizing the iGPU rather than the dGPU. Encoding for upload seems to be a bit faster, likely because of the iGPU, but just working in the app feels slower. Is this a naming issue @CaseySJ? Do you need to see my IOReg output?

Anyone know how to get Final Cut to target the dGPU if the iGPU (headless) is running? In my Devices->Properties I only have the following set PciRoot(0x0)/Pci(0x2,0x0) -> AAPL,ig-platform-id -> 0300983E

Are there more settings I need? I've seen some configs with frame buffer settings.

 

[heythisisdave]

Wow - this Osx free2000 has been sitting in the back of my mind for a few days since I found a few posts talking about why it's not great to be using it.

So cool to try the FwRuntimesServices and OCQuirks and it worked out of the box

I didn't have to do a NVRAM Reset and this is on 10.15.1

The system before was pretty stable (though right before I made the change, I tried removing dart=0 and had a series of hard crashes).

Really cool! Thanks @JimSalabim and @CaseySJ!

Will report if I see any stability issues.

Anyone have ideas why boot is pretty slow? I get to the end of the IOConsoleUsers in verbose mode and it sits there for 10-15 seconds. OTOH, Windows 10 boots for me almost instantaneously.

 

[heythisisdave]

Also would love to understand what the ._ files that are being created automatically in the EFI folder are. It seems like something is watching the folder and creating it (not during boot but right when I put the file there).

Is this just an OSX thing and unrelated to Clover or boot?

 

[digumo]

Is this just an OSX thing and unrelated to Clover or boot?

If you do a cat on one of the files it looks like it's the OS that is doing it. It's tracking things you've downloaded and probably doing some type of comparison somewhere. I have a few which are dated the same as when they were initially downloaded and put into place. Here's the output from the FwRuntimeServices.

me@myMac [21:35:15] [/Volumes/EFI/EFI/CLOVER/drivers/UEFI]
-> % cat ._FwRuntimeServices.efi
Mac OS X            2��ATTR;����$j$9com.apple.quarantine]com.apple.lastuseddate#PSF6-m5)com.apple.metadata:kMDItemDownloadedDate��%com.apple.metadata:kMDItemWhereFroms0083;5dc8d61e;Safari;3FB5AEDB-E1F4-46F6-97D5-7905C2DD6224��]�)bplist00�3A����W��
bplist00�_Dhttps://www.tonymacx86.com/attachments/fwruntimeservices-efi.435372/_thttps://www.tonymacx86.com/threads/success-gigabyte-designare-z390-thunderbolt-3-i7-9700k-amd-rx-580.267551/page-980
                                                        R�This resource fork intentionally left blank   ��%

You can tell it's probably the OS because of kMDItemDownloadedDate and kMDItemWhereFrom, etc.

 

[digumo]

Ok, so here's my reporting back about running my iGPU and Final Cut Pro.

It may be that I have some perception bias or something with the iGPU enabled. Final Cut does seem to opt for the iGPU for certain tasks. It will flatline when FCP is sitting still, but will then ramp up when I play the timeline.

I watched the Intel Power Gadget while doing stuff in the timeline and also when starting an export. The power seems to ramp up on export, but not as much during timeline scrubbing and editing. Although frequency does pop up and down while editing.

But I'm all for hearing any tips when it comes to Final Cut and making things faster or making it opt for the dGPU. When watching iStat Menu, it seems like FCP never really takes full advantage of the dGPU. Even when the iGPU was disabled.

 

[MVoort]

Thank you very much. In my BIOS I turned off Chipset --> Internal Graphics (I cannot find IGFX but I presume you mean the Internal Graphics). And I am using the new config.plist. Running a test now and when I have a result I will report back about the results.

As promised I would give feedback on the tests. For some reason it seems that my system freezes more frequently after the changes. I added back the -disablegfxfirmware in the boot option. Will see how the system works.

 

[digumo]

WOOOOT!!!! I think I'm in love with this Hackintosh all over again. After doing a bit of research I found this post which clued me into trying something.

So, check this out... Running FwRuntimeServices.efi and OcQuirks.efi I hopped into BIOS and decided to bump the DVMT Pre-allocated from the default 64MB to 128MB with the total still set to 256.

I hop back into Final Cut after a successful boot and watching Intel Power Gadget as I maintain a GFX flatline while scrubbing within FCP which is faster and what seems like back to normal.

I also tried some exporting. So this means one of two things. Either the iGPU is not being used - after increasing the DVMT Pre, or it's just not registering in Intel Power Gadget. Looking at iStat Menu, while exporting, my Radeon VII is now showing it's doing the processing.

If the iGPU "IS" still being used... and... given that I'm only working with 1080p footage I imagine you can bump these values if you're working with 4K+ footage. But, does this seem right? It's as if FCP is determining what it should use based on the graphics memory value?

I'll have to try with some other combinations of memory settings and see what results.

[BIG TIME EDIT - 5 minutes later] This is the hazard of accidentally tricking yourself because your testing if flawed. The above results are bogus because I was working with a config_testing.plist file which I needed to switch to within clover. My default config.plist still had dart=0 set in boot args so the GPU, while showing was obviously flatlined because it was disabled! Argggg!!! Even still, I am noticing improvements by bumping the iGPU memory, but not as much as if the dGPU is doing the whole job.

 

[digumo]

[BIG TIME EDIT - 5 minutes later] This is the hazard of accidentally tricking yourself because your testing if flawed. The above results are bogus because I was working with a config_testing.plist file which I needed to switch to within clover. My default config.plist still had dart=0 set in boot args so the GPU, while showing was obviously flatlined because it was disabled! Argggg!!! Even still, I am noticing improvements by bumping the iGPU memory, but not as much as if the dGPU is doing the whole job.

Clearly I have more learning to do about all this stuff as dart=0 doesn't have anything to do with the iGPU but rather "Disables VT-x or VT-d" - straight from the clover docs.

As of recent, I've been able to boot with DVMT Pre-allocated at 256Mb and total allocation set to MAX for the iGPU. FCP is still not as fast as when the iGPU is disabled and I'm hopeful that someone has the info which allows FCP to use the dGPU but still keep the iGPU enabled for encoding/decoding for H264/H265.

 

[chabeus57]

@CaseySJ
Hello Casey. After multiple tests, I disconnected all usb connections to test them one by one. And I ended up finding a single connection that accepts the Super Drive (USB 3.1 red). I attach a screenshot for those who encounter this problem.

See you later.