Contribute
Register

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

Joined
Apr 1, 2011
Messages
348
Motherboard
Gigabyte z390 Designare
CPU
i9-9900K
Graphics
RX 5700 XT
Mac
  1. iMac
  2. MacBook
  3. MacBook Air
  4. MacBook Pro
  5. Mac mini
  6. Mac Pro
Mobile Phone
  1. iOS
Osy86's Thunderbolt patcher (tbpatch) tool is awesome, but it won't work out of the box. We first need to configure and install the TbtOnPch SSDT, reboot, ensure that Thunderbolt Bus appears, and then run the patcher to modify the Thunderbolt chip. On an Alpine Ridge system, you might wonder why it's even necessary to flash the firmware when an SSDT will do the job! @Elias64Fr has a fully working Thunderbolt Bus on his Asus Maximus X Hero Maximus IX Extreme (with on-board Alpine Ridge). Firmware flashing is not necessarily required for on-board Alpine Ridge controllers.

Correction: Elias has Maximus IX Extreme.
Is this a light at the end of the tunnel for a non-external, software only, patching method to swap out the NVM on our built in Titan Ridge chips?? Potentially??
 

CaseySJ

Moderator
Joined
Nov 11, 2018
Messages
19,809
Motherboard
Asus ProArt Z690-Creator
CPU
i7-12700K
Graphics
RX 6800 XT
Mac
  1. MacBook Air
  2. MacBook Pro
  3. Mac Pro
Classic Mac
  1. Quadra
Mobile Phone
  1. iOS
CaseySJ, interesting question, I have two VEGA 64's one in slot one and one in slot 2. Do you think I should be loading both injection SSDTs in order to provide injections to both cards or does this not work like that??

Any insight would be appreciated, thanks
Yes -- one for each slot.
 

CaseySJ

Moderator
Joined
Nov 11, 2018
Messages
19,809
Motherboard
Asus ProArt Z690-Creator
CPU
i7-12700K
Graphics
RX 6800 XT
Mac
  1. MacBook Air
  2. MacBook Pro
  3. Mac Pro
Classic Mac
  1. Quadra
Mobile Phone
  1. iOS
I can confirm this. My Open CL score was worse with Radeonboost.kext and metal the same on geekbench 5. RX580 nitro+
My 590's metal scores were 2.74% worst with radeon booster kext
Thanks for the feedback. The micro-guide has been modified as follows:

Screen Shot 2020-04-25 at 6.12.02 AM.png
 

CaseySJ

Moderator
Joined
Nov 11, 2018
Messages
19,809
Motherboard
Asus ProArt Z690-Creator
CPU
i7-12700K
Graphics
RX 6800 XT
Mac
  1. MacBook Air
  2. MacBook Pro
  3. Mac Pro
Classic Mac
  1. Quadra
Mobile Phone
  1. iOS
Is this a light at the end of the tunnel for a non-external, software only, patching method to swap out the NVM on our built in Titan Ridge chips?? Potentially??
That applies to Alpine Ridge. For Titan Ridge, this post still applies.
 
Joined
Jul 21, 2016
Messages
334
Motherboard
Gigabyte Z690 Aero G
CPU
i9-12900K
Graphics
RX 6800 XT
Mac
  1. iMac
  2. MacBook Pro
  3. Mac mini
  4. Mac Pro
Mobile Phone
  1. iOS
thanks for doing the testing.
is the Titan Ridge AIC flashed?

currently my issue is that waking my Aorus Pro/AR AIC/TB display from sleep, either by pressing the power button or by a keystroke, spins up the fans and interior lighting, but the display remains black and the computer isn't accessible via screen sharing. I have to shut the power off. Once rebooted using a second monitor connected (TB still connected but black) shows "no thunderbolt devices" in system report. I have to clear CMOS, reset BIOS settings, unlock MSR, and all is well. also of interest is that before the "sleep problem" occurs, using only the TB connected at boot, sometimes the display lights up immediately for part of the boot process then goes dark as expected until the desktop appears, and sometimes it is dark until the desktop appears.

yes, the workaround is to prevent system sleep (screen sleep works OK) and power off at the end of the day. but I am cursed with the engineer's affliction of wanting to know why stuff doesn't work right!

I've spent the last few days on and off trying to come up with a set of steps that reliably reproduces the wake problem so I can describe it here. I'm not quite finished with the testing and documentation.

the testing I've done would suggest my wake problem may be tied to BIOS settings. are you using the settings as recommended by CaseySJ in post #1?

thanks!

@faithie999

What a disappointment to lose your TB bus based on SLEEP failure, then have to go through that extensive restore procedure!

The following information is based on my Z390 Aorus Pro with a flashed GC-Titan Ridge AIC. Flashing was done per @CaseySJ mini-guides. I'm running fully updated Mojave 10.14.6, BIOS F11, Clover 5107, i9-9900K, RX-580 (headless), jumper cable between the GC-TR AIC and DP on the RX-580. Apple TB Display with the TB3->TB converter plugged into a GC-TR TB3 port, and the other LG display plugged into an HDMI port on the RX-580.

The attached Aorus Pro BIOS - TB.zip file contains three jpg files (BIOS-TB-x) with screen shots of my BIOS setting for Thunderbolt.

Also a file - AP_04_25_2020 - saved from my BIOS with my current BIOS settings. You COULD SAVE your current BIOS settings to a file (in the EFI partition of a thumb drive), and load my BIOS setting to see if there are any other settings that might make a difference.

Finally a screenshot of my Clover Configurator Acpi screen with a "Disable GPE.TINI - TB3 Slow Boot" patch that significantly increases boot time with the GC-TR AIC installed. Credited to @Elias64Fr . (Since you are doing a lot of rebooting!)

Are you confident that the flashing of the GC-TR AIC was successful? Would it make sense to re-flash the card if none of this information I've offered helps?

It seems that typically two seemingly very similar hackintoshes can act very differently.

Good luck!
 

Attachments

  • Aorus Pro BIOS - TB.zip
    533.7 KB · Views: 70
Joined
Feb 18, 2019
Messages
356
Motherboard
ASUS Z690 Maximus Extreme
CPU
i9-12900KF
Graphics
6900XT
Mac
  1. MacBook Air
Mobile Phone
  1. iOS
My 590's metal scores were 2.74% worst with radeon booster kext

I don't think CMMChris made things worse. Keep in mind that if you repeatedly run GeekBench, you'll get different readings, so under 5% bewteen 2 runs is more likely statistical variation. Run it 20x each way and watch the trend.

Also, I've noticed better readings once the computer has been running for a while rather than immediately running tests as soon as the computer has booted. So to compare Apple to Apple (pun), test comparisons should have the boot time be about the same between test groups.

The APGM section is still useful, so I'm using it, as is, on one machine with an RX580, where I noticed no difference.

If you're still convinced it is worse, you could delete the RX480-590 section from Info.plist of RadeonBoost (see the highlighted section in the Spoiler below; where you can also see where I added the Vega56-64 section to v1.4 at the bottom of the list). This way you could continue to use the APGM section as opposed to entirely removing RadeonBoost.

EditRadeonBoost.jpg

 
Joined
Jul 7, 2010
Messages
129
Motherboard
ASUS ROG MAXIMUS HERO X Z370
CPU
i9-9900KS
Graphics
RX 580
Mac
  1. MacBook Air
Classic Mac
  1. PowerBook
Mobile Phone
  1. iOS
I don't think CMMChris made things worse. Keep in mind that if you repeatedly run GeekBench, you'll get different readings, so under 5% bewteen 2 runs is more likely statistical variation. Run it 20x each way and watch the trend.

Also, I've noticed better readings once the computer has been running for a while rather than immediately running tests as soon as the computer has booted. So to compare Apple to Apple (pun), test comparisons should have the boot time be about the same between test groups.

The APGM section is still useful, so I'm using it, as is, on one machine with an RX580, where I noticed no difference.

If you're still convinced it is worse, you could delete the RX480-590 section from Info.plist of RadeonBoost (see the highlighted section in the Spoiler below; where you can also see where I added the Vega56-64 section to v1.4 at the bottom of the list). This way you could continue to use the APGM section as opposed to entirely removing RadeonBoost.

What would be the additional gains using the AGPM injection from the kext over what is native in macOS especially running a machine on iMac 19,1 with an RX580 and i9-9900 which is surely as close to a real Mac combination as you can get.
 
Joined
Sep 8, 2011
Messages
78
Motherboard
Asus WS X299 SAGE 10G
CPU
i9-7960X
Graphics
WX4100
Mac
  1. Mac Pro
Mobile Phone
  1. iOS
My results were pretty good too for both OpenCL and Metal.

I'd already removed WEG from my EFI due to adverse affects on the Radeon VII, so I didn't have any issues with the SSDT (in fact, the PowerPlayTable gave me another 10% or so because I boosted frequency besides adjusting voltages). Also, I'm running Mojave (Catalina is too finicky for me).

The Radeon VII is water-cooled on the GB Z390 Aorus Xtreme Waterforce. Spoiler shows only GeekBench v4 results.

Pre-RadeonBooster:

After RadeonBooster:.
WOW... can you Attach your PowerPlayTable or file for your Radeon VII? I have one water-cooled too. Thanks
 
Top