macOS 12.3 Update Causes Problems for 5700/6800/6900 Graphics Cards

[maxxx]

@maxxx @tedyun I just what you asked and used the tool ...does this look correct to you?

View attachment 544237

yes, looks fine, can confirm with geekbench to verify the speed.

 

[xtreem.p]

I can concur that with these settings, im able to use my 5700xt with 12.3 to its fullest

Thanks for the solution community!

 

[steff89]

@xtreem.p One maybe stupid question, but why do you have a GPU tab in your Activity Monitor? I do not see that?

 

[xtreem.p]

@xtreem.p One maybe stupid question, but why do you have a GPU tab in your Activity Monitor? I do not see that?

I have no idea why i have that.. its just there

 

[raulneo]

is there any conflict to apply the framebuffer patch with the 6900 XTXH ssdt patch?

 

[athei]

several things wrong with this. 6900 should use Belknap too, Carswell is the framebuffer used with W6800X Duo mpx modules that have two navi cores. Deepbay isn't for 6600, but - I think - for navi 21 gpus with no connectors (used only for compute, unsure if such a device even exists for apple platform right now tho)

Can you elaborate on this? I am curious how to find out which are the correct framebuffers except owning original hardware and dumping the EFI.

 

[MysticalOS]

I am seeing many users add more than they need to for device properties, for example inserting card rom#, but chances are that was already working. for the performance fix you ONLY need the ATY,Type lines for at symbol 0-3 and the device_type line

most of other stuff is read correctly from card so the replacements are not needed unless you are spoofing device ID like XTXH then you need just the new Id, or you are inserting a custom name, usually again for XTXH cards.

EDIT: More testing, I found that these are best to add to fix performance and not have sleep issues

@0,name
@1,name
@2,name
@3,name
AAPL00,DualLink
device_type

before I only added @N values and that fixed performance but sleep had a tendency to be a little off and performance wasn't 100% with dual link set and evice type also set to correct ATY,typeforcard I can say performance is 100%, actually more than 100% since i'm betting setting duallink and proper device type is actually what fixed performance issues in general even pre 12.3. 180,000 geek bench scores now and even more consistent fps in WoW than 12.2.x without values, so 12.3 was a blessing in disguise in setting up peak performance of 6900XT

all the other values i still found are there without altering, so replacing them seems pointless. You might be able to edge a tiny bit more performance by the power management values but i found those also could mess with computer staying asleep.

 

[MysticalOS]

is there any conflict to apply the framebuffer patch with the 6900 XTXH ssdt patch?

it's better to just use opencore or clover device properties injection for both frame buffer AND XTXH because whatevergreen supports doing the patching with a boot arg+device ID replcement in properties. two birds one stone IMO

 

[erbrango]

Bad update. Animations are stuttering now on a RX 5700 XT.

---

 

[buddyjack2]

I am seeing many users add more than they need to for device properties, for example inserting card rom#, but chances are that was already working. for the performance fix you ONLY need the ATY,Type lines for at symbol 0-3.

most of other stuff is read correctly from card so the replacements are not needed unless you are spoofing device ID like XTXH then you need just the new Id, or you are inserting a custom name, usually again for XTXH cards.

Thanks for clarifying that!

I'm testing a fresh install (not an update) of Monterey 12.3 on a separate partition of an internal SATA drive. The only 3rd-party software I've added is Geekbench, to be able to check results; everything else on that 12.3 installation is "virgin" from the Apple installer. I boot from a USB thumbdrive containing OpenCore 0.7.9 (with the config.plist edits suggested in this thread, i.e., ATY,Belknap), then select the Monterey 12.3 partition at the OpenCanopy screen. Config.plist is attached here.

Am glad to report my Geekbench Metal score in 12.3 returned to its levels I get when booting 12.2.1 from my internal SSD. However, almost as soon as I finished running Geekbench the system reported a WindowServer crash (which I've never seen before) and reverted to the login screen. After entering my user password and logging back in, I then got notices that my "pixart device" was not properly recognized, as well as my (PC-style) keyboard. Didn't know what "pixart" was so I launched Safari to try Googling the term, but Safari immediately crashed.

I decided to reboot, but now Monterey 12.3 seems totally fried. Bootup nearly finishes, then loops back to my Hackintosh UEFI logo and OpenCanopy screen.

I've since discovered that "pixart device" is apparently my gaming mouse. I can try the procedure again with a fresh 12.3 install and using an Apple Magic mouse/keyboard, but am hoping someone here could check if any other culprits exist in my config.plist. It's the same I use successfully with 12.2.1, except for the device properties additions for Belknap.