MP All Models AMD Radeon RX 6800 XT / 6900 XT for macOS?

[Stex]

Hi guys,

Just want to ask for my brother as it is not my computer. He just bougth and installed a 6800 XT to his mac pro 2019 and it is detecting fine and being used. Howover when monitoring its usage it seems the % of GPU is much higher for the 580X which came with the machine. I know just becuase % is higher doesn't mean it is faster but is there anyway to ensure that the 6800 XT is priortise?

He did left the old GPU installed for the moment, that shouldn't have any negative impact in performance should it?

He is using mostly Adobe Premier. Thanks in advance.

this is a known issue, suggest him to read this complete thread and call apple support asap:

Thread 'RX6800 slow in OS12.3 Beta'
https://forums.macrumors.com/threads/rx6800-slow-in-os12-3-beta.2333537/

 

[johnnymcc]

Hi guys,

Just want to ask for my brother as it is not my computer. He just bougth and installed a 6800 XT to his mac pro 2019 and it is detecting fine and being used. Howover when monitoring its usage it seems the % of GPU is much higher for the 580X which came with the machine. I know just becuase % is higher doesn't mean it is faster but is there anyway to ensure that the 6800 XT is priortise?

He did left the old GPU installed for the moment, that shouldn't have any negative impact in performance should it?

He is using mostly Adobe Premier. Thanks in advance.

yes, 12.3 broke navi GPUs. I had to revert back to 12.2.1 to get full usage back. It was disgustingly slow in 12.3

Call apple and get a case ID - here's my post in Apple discussions

Monterey 12.3 broke my 6800 XT - Apple Community

discussions.apple.com

 

[erichui84]

yes, 12.3 broke navi GPUs. I had to revert back to 12.2.1 to get full usage back. It was disgustingly slow in 12.3

Call apple and get a case ID - here's my post in Apple discussions

Monterey 12.3 broke my 6800 XT - Apple Community

discussions.apple.com

Thanks johnnymcc and stex! will have a look.

 

[penguinlust]

Monterey 12.3.1 was released this morning. No idea if it fixes the issue, but worth checking.

 

[flowrider]

^^^^^I'm downloading it, but the fixes do not mention this issue.

Lou

 

[penguinlust]

Installed and it fixed the performance issues I was seeing on my 6900XT for me.

 

[flowrider]

Yep - It's fixed. So much for the call I was going to get My Scores:

Before 12.3 Install:

View attachment 1984510

After 12.3 Install:

View attachment 1984509

After 12.3.1 Install:

View attachment 1984511

Lou

 

[SDAVE]

What a big fumble by Apple wow. Especially folks who have auto update on. I wasted a day reverting back to an older version of macOS because of this nasty bug.

 

[flowrider]

^^^^That's the beauty of having more than one startup disk?

Just called Apple to let them know the issue seems to be fixed.

Lou

 

[SDAVE]

^^^^That's the beauty of having more than one startup disk?

Just called Apple to let them know the issue seems to be fixed.

Lou

That's fine but this is a big big issue Apple should've never allowed to happen.

It was a final release not a beta.

 

[flowrider]

^^^^I whole heartily agree with you one that one. But they did, they made a mistake (I think❓) and mistakes happen. I waited until the final release before I verbally reported it to Apple. I had been using the "Feedback Assistant" during the Beta cycle.

Lou

 

[turbineseaplane]

Can you guys get 4k/120hz or 4k/144hz with your 6900xt?

 

[penguinlust]

Can you guys get 4k/120hz or 4k/144hz with your 6900xt?

I have not tried -- I'm driving a 2560 by 1600 at 60 fps.

 

[ZombiePhysicist]

Can you guys get 4k/120hz or 4k/144hz with your 6900xt?

Yep, and I even got 8k/60hz

SOLVED: 8k Displays Running on Mac Pro? Any? What Video cards would work that support 8k HDMI 2.1/Displayport 1.4/2.0 displays on Mac Pro? YES you can

UPDATE 2021-12-19: (Some failed tests I tried to get it to work) https://forums.macrumors.com/threads/8k-displays-running-on-mac-pro-any-what-video-cards-would-work-that-support-8k-hdmi-2-1-displayport-1-4-2-0-displays-on-mac-pro.2309750/post-30712355 UPDATE 2021-12-26: A first solution can be...

forums.macrumors.com

 

[turbineseaplane]

Yep, and I even got 8k/60hz

SOLVED: 8k Displays Running on Mac Pro? Any? What Video cards would work that support 8k HDMI 2.1/Displayport 1.4/2.0 displays on Mac Pro? YES you can

UPDATE 2021-12-19: (Some failed tests I tried to get it to work) https://forums.macrumors.com/threads/8k-displays-running-on-mac-pro-any-what-video-cards-would-work-that-support-8k-hdmi-2-1-displayport-1-4-2-0-displays-on-mac-pro.2309750/post-30712355 UPDATE 2021-12-26: A first solution can be...

forums.macrumors.com

Will dig in and read..
Was it a certain cable/adapter?
Custom rez/settings in SwitchResX?

 

[ZombiePhysicist]

Will dig in and read..
Was it a certain cable/adapter?
Custom rez/settings in SwitchResX?

It was a certain cable (the one I used to get 8k) and switchResX with certain settings. After I did that, a few new resolutions showed up in SwitchResX and 4k120 was one of them, and worked well.

 

[turbineseaplane]

Found this post: https://github.com/acidanthera/bugtracker/issues/1812#issuecomment-939978793

No idea if the settings are fully correct for my monitor (Gigabyte M32U), but it worked and I have 120hz with my HiDPI scaled modes on Monterey 12.3.1

Gotta' love SwitchResX!

 

[startergo]

169945
Metal Score

103642
OpenCL Score

 Chipset Model:    AMD Radeon RX 6900 XT
  Type:    GPU
  Bus:    PCIe
  Slot:    Slot- 6
  PCIe Lane Width:    x16
  VRAM (Total):    16 GB
  Vendor:    AMD (0x1002)
  Device ID:    0x73bf
  Revision ID:    0x00c0
  ROM Revision:    113-D4120100-100
  Metal Family:    Supported, Metal GPUFamily macOS 2

System Version: macOS 12.3.1 (21E258)

OpenCL score looks pretty low, but Metal is great.

120585
OpenCL Score

172265
Metal Score

System Version: macOS 11.6 (20G165)
OpenCL is better in BS.

In latest BS:
System Version: macOS 11.6.5 (20G527)

123213
OpenCL Score

177968
Metal Score

 

[startergo]

As a comparison here is the OpenCL GB5 test in Windows:
https://browser.geekbench.com/v5/compute/4634840

162977
OpenCL Score

OpenCL Mac drivers s**k.

 

[exoticSpice]

OpenCL Mac drivers s**k.

Apple's focus is all on Metal. It is a much better API than openCL.

 

[Shawn Moosa]

Are there any particular models of 6900xt you guys will recommend?

 

[ZombiePhysicist]

Are there any particular models of 6900xt you guys will recommend?

I wanted a reference card that also had the USBc connector, so I got the Sonnet version.

Sonnet AMD Radeon RX 6900 XT Graphics Card

Sonnet AMD Radeon RX 6900 XT graphics card for Sonnet DuoModo and Breakaway Box eGPU expansion systems, and 2019 Mac Pro. Also compatible with Windows computers.

www.sonnetstore.com

 

[Syncretic]

An unrelated problem I was looking at led me back to the RX6800 ROM posted by @Petri Krohn some months ago. Further digging shows that the RX6800 ROM assumes UEFI 2.x, while the MP5,1 provides an EFI 1.x+ environment. In particular, the EFI_HII_DATABASE_PROTOCOL (EF9FC172-A1B2-4693-B327-6D32FC416042) is not produced in the Mac Pro 5,1 EFI, but it is consumed by the RX6800 ROM, without error control - meaning the ROM looks for that protocol and stores the returned address (NULL, in this case) without checking the error code, then the ROM later uses that address without checking for NULL. This is almost certainly the cause of the "won't POST" problems with the RX6800; the code is jumping off into nowhere.

It should be fairly straightforward to write an auxiliary module that produces EFI_HII_DATABASE_PROTOCOL and the handful of other protocols the RX6800 expects; however, since this is running at PEI/DXE time, such a module would probably need to be flashed into either the RX6800 ROM or the Mac Pro BootROM in order to be available when it's needed. I doubt it's worth going down that path.

Someone asked me in PM about getting the RX6800 working on a classic (5,1) Mac Pro, so I spent a few more minutes looking at the RX6800 ROM posted by @Petri Krohn. Closer analysis makes this poorly-written code look even worse: first, they try to locate five of the HII protocols (which don't exist in the MP5,1 EFI), and save the results without checking for errors (which results in NULL pointers being stored). Later, they blindly try to call functions using those NULL pointers (which is the apparent cause of the crashes/failures to POST). Then the kicker - it looks like they don't even use the values that would have come back from a successful function call. In other words, the problem code appears to be entirely unnecessary.

It looks to me like a 2-6 byte patch of the uncompressed ROM should fix this problem (I have no idea if there might be other unrelated problems that may come to light once the card is able to POST). I don't have an RX6800 to test with, so I'd be flying blind. If someone with a flashable RX6800 is willing to experiment, I'd be willing to try creating this patch. Before you volunteer, there are a few things you should be aware of:

• Obviously, the patch may or may not be successful. This is an experiment, positive results are not guaranteed. The patch I have in mind is based on the RX6800 ROM that was posted earlier; your particular ROM may be different, and require different patches (or it may not be workable at all).
• You'll need to be prepared to recover from any mishaps - notably, if your flashing fails, or one of the patched ROMs fails to work properly, you'll need a way to restore the RX6800's original ROM. You should first make a backup copy of your current ROM, then flash it back to the card so you're comfortable with the procedure; you should also investigate (in advance) what methods are available to recover from a failed flash. I'm far away from you, and I don't have an RX6800, so I'll be of little help if you have flashing problems.
• I can reconstruct the PCIr records, including compression, but there may be internal checksums/CRCs that won't work after patching. Notably, there's a Microsoft certificate embedded in the EFI ROM image; if that's associated with some hash of the ROM, any modifications may cause security issues (either warnings or errors that crop up at runtime, failure to recognize WHQL certification, or the ROM may refuse to POST because it thinks it's been compromised (which will be true in this case)). Therefore, even if the patched code would work flawlessly, it's possible that the card will refuse to operate due to the modifications. (This shouldn't affect your ability to re-flash the ROM back to the original, but I don't know that for certain.)
• Flashing a graphics card (or, really, any device) carries an inherent risk of failure or damage. It's usually fairly safe and easy, but there's a non-zero chance that your card could end up being inoperable (aka "a brick"). If your RX6800 is precious to you, it's probably better not to experiment on it.
• My availability is sporadic, so it's possible that there may be some delays during our testing process (assuming it takes more than one attempt to get this right). I'll try to be as responsive as I can, but I make no guarantees.

So, given all those caveats, if any brave RX6800 owner would like to try some experiments, drop me a PM and let me know.

 

[Spacedust]

Don't even waste your time. MacVidCards Europe already released RX6800 and RX6600XT cards with EFI boot screen working on Mac Pro 5,1.

 

[OVERKILL338LM]

Don't even waste your time. MacVidCards Europe already released RX6800 and RX6600XT cards with EFI boot screen working on Mac Pro 5,1.

I'm seeing the 16GB cards on their ebay store, so definitely possible to make the RX 6600 and 6800 work, which is great news! The 6600 in particular is very well priced right now, and available! Just need the proper modifications to the ROM as @Syncretic has noted, and clearly the folks over at MacVidCards have already figured out.