HP EliteDesk 800 G4 / G5 Mini with RX560x dGPU (hackintosh)

[ird]

1 hour ago, deeveedee said:

@ird I maintain the change log here.  I think I already have the link that you requested, but if not, then I don't understand your request.  Please let me know.

Yes, you are right about the high level changelog. I just meant an additional link to the post that has a list of detailed changes like the above post.

[ird]

Sorry for the confusion @deeveedee. For whatever reason my mobile browser did not show the URL linked to the word “here” for the detailed changelog in your post.I now see it on the desktop. All good!

[ird]

I've ported over @deeveedee's latest changes and upgraded OC to 1.0.1 for iMac19,2 SMBIOS with headless iGFX in case anyone is interested (don't forget to replace Serial number/MLB/ROM etc. as always). Attached version does not contain Intel Wifi+BT drivers-firmware due to size limitations (but has 1 USB port reserved for internal BT).

 

For EFI with wifi+BT drivers and firmware, you can grab it from my github repo: Github link

 

Both versions come with multiSMBIOS USB kext that works for iMac19,2 as well as MacMini8,1

 

EDIT1: Corrected typos

 

EFI_OC101_iMac19p2_iGFX_Headless.zip

Edited August 7 by ird
Corrected typo

[deeveedee]

Easy upgrade to Sonoma 14.6.1 with SecureBootModel=Disabled.

Spoiler

 

[deeveedee]

In my known issues, I have a warning about the use of "non-Dynamic" Wallpapers/ScreenSavers.  I am now testing "non-Dynamic" WallPapers/ScreenSavers (currently testing "Earth" and "Underwater") without issues.  It may be that the issues I was experiencing were caused by improper patching of the RX 560x in my early attempts.  Now that the RX 560x patching issues are fully resolved and all is working perfectly, I am no longer observing the Wallpaper/ScreenSaver issues.  I'll test more before making any final conclusions and welcome testing feedback from others.

Edited August 8 by deeveedee

[ird]

16 hours ago, deeveedee said:

In my known issues, I have a warning about the use of "non-Dynamic" Wallpapers/ScreenSavers.  I am now testing "non-Dynamic" WallPapers/ScreenSavers (currently testing "Earth" and "Underwater") without issues.  It may be that the issues I was experiencing were caused by improper patching of the RX 560x in my early attempts.  Now that the RX 560x patching issues are fully resolved and all is working perfectly, I am no longer observing the Wallpaper/ScreenSaver issues.  I'll test more before making any final conclusions and welcome testing feedback from others.

I’m not sure if the issue you were facing with non dynamic wallpapers were Sequoia specific and limited to Apple provided or others as well. But in  my case, with Sonoma they’ve always worked fine. I might have missed this note before.

[deeveedee]

@ird Did you have a working RX 560x solution for the EliteDesk 800 G4 / G5 Mini before I posted mine?  If so, you'd be the first (I think), because I didn't find anyone who got this working without a black screen.  My known issues were collected from notes I was taking while I was finding a working solution for the RX 560x, so it's quite possible that I experienced the problem during my initial failed attempts.

 

Kudos to you if you had it working before me! 👏

 

EDIT: My dolphins are swimming!  The "California Dolphin Pod" screen saver is great

 

EDIT2: @ird I re-read your feedback (thank you for that) and realize your were saying that you haven't observed any Wallpaper/ScreenSaver issues since you started testing the EFI that I posted.  That makes sense, since it is likely that the Wallpaper/ScreenSaver issues that I observed were with my experimental EFIs before I posted my first tested/working solution.  Sorry for the confusion.  I'm glad the WallPaper/ScreenSavers are working for you, too!

Edited August 10 by deeveedee

[deeveedee]

@ird My apologies if I'm asking a question that you've already answered.  Have you played with PowerPlay tables in a manner similar to what is described here?  Is that even relevant to our RX 560x and if so, could the method be used to tweak performance (up or down)?  Again, sorry if I'm asking a question that you already mentioned.  I'm still learning about the Radeon cards.  Thanks.

[ird]

3 hours ago, deeveedee said:

@ird My apologies if I'm asking a question that you've already answered.  Have you played with PowerPlay tables in a manner similar to what is described here?  Is that even relevant to our RX 560x and if so, could the method be used to tweak performance (up or down)?  Again, sorry if I'm asking a question that you already mentioned.  I'm still learning about the Radeon cards.  Thanks.

 

That was on my todo list post the unsuccessful attempt to unlock additional cores. Unfortunately, I haven't yet got to it. If you are interested in trying out, you can refer to the raw files I attached to this post. You can skip over the ROM dump part, it's all done for you. You should be able to extract powerplay tables from the binary and the text header for location start/size following that post and then inject stock or modified version into macOS through config.plist device properties.

 

Note that few parameters like fan speed etc. are permanently locked on HP RX560X bios.

[ird]

On 8/7/2024 at 2:52 PM, deeveedee said:

Easy upgrade to Sonoma 14.6.1 with SecureBootModel=Disabled.

  Reveal hidden contents

 

 

Just curious if incremental updates work for you; they don't for me. After 2 tries, macOS downloads the full installer that installs fine.

[deeveedee]

@ird Yes - they work for Sonoma and Sequoia Beta - no issues.  Please post your current EFI if you'd like help.

[ird]

24 minutes ago, deeveedee said:

@ird Yes - they work for Sonoma and Sequoia Beta - no issues.  Please post your current EFI if you'd like help.

 

Interesting. They don't for me. My EFI is the same as posted above in this post.

 

While I use RestrictedEvents.kext, the only explicit boot args I use is to block media process (revblock=media). I wonder if it has to do something with the revpatch=sbvmm key under NVRAM > 4D1FDA02-38C7-4A6A-9CC6-4BCCA8B30102.

[deeveedee]

rev patch=sbvmm is necessary for MM8,1 hack to be notified about macOS updates in Sonoma and Sequoia.  I don't know if you need it for iM19,2 (which does not have T2).  I'm not sure.

 

EDIT: I diff'd our config.plists, but there are too many cosmetic differences for comparison.  IF you don't figure it out, see if you can start with my config.plist and make the minimal changes for easy comparison.

 

EDIT2: I did notice that you use OpenHFsPlus.efi driver.  I think you'll get better HFS performance with HfsPlus.efi.

 

EDIT3: Does changing SecureBootModel->Disabled affect the OTA behavior for you?  I don't think it should, since you're using iMac19,2, but you may want to test that.

Edited August 11 by deeveedee

[deeveedee]

I just ran GeekBench 6 CPU Benchmark after upgrading to Sonoma 14.6.1.  The performance of the i5-8600 65W CPU with 230W power adapter and 32GB RAM is impressive.  Not only does it score reasonably well in synthetic benchmarks, but this hack with RX 560x is extremely responsive.

 

Spoiler

 

[ird]

6 hours ago, deeveedee said:

rev patch=sbvmm is necessary for MM8,1 hack to be notified about macOS updates in Sonoma and Sequoia.  I don't know if you need it for iM19,2 (which does not have T2).  I'm not sure.

 

EDIT: I diff'd our config.plists, but there are too many cosmetic differences for comparison.  IF you don't figure it out, see if you can start with my config.plist and make the minimal changes for easy comparison.

 

EDIT2: I did notice that you use OpenHFsPlus.efi driver.  I think you'll get better HFS performance with HfsPlus.efi.

 

EDIT3: Does changing SecureBootModel->Disabled affect the OTA behavior for you?  I don't think it should, since you're using iMac19,2, but you may want to test that.

 

Yes, I'll try to experiment with a few options to see if it solves the issue. Also, thanks for the tip on HfsPlus.efi. I'll certainly switch. I need to have SecureBootModel set to Disabled to be able to successfully install OTA. That's something I do just before updating the machine but otherwise leave it enabled after installation all the time, so not much I can change to experiment there.

[ird]

1 hour ago, deeveedee said:

I just ran GeekBench 6 CPU Benchmark after upgrading to Sonoma 14.6.1.  The performance of the i5-8600 65W CPU with 230W power adapter and 32GB RAM is impressive.  Not only does it score reasonably well in synthetic benchmarks, but this hack with RX 560x is extremely responsive.

 

  Reveal hidden contents

 

 

I agree 100%.

 

Here's my story: I'd been using an AMD hack for quite some time, but I needed something smaller and with moderate power consumption as I downsized my desk. I've been very happy with my Macbooks since over 2 decades, so a MacMini was an obvious choice for me. While there is no debate on the goodness of performance to power ratio of the M2/M2 Pro MacMini, except the base model I couldn't justify the price of upgrades esp. the storage and RAM both of which were non-negotiable for me. Considering that the current mini is almost 2 generations old with literally no meaningful price cuts, it's a hard pill to swallow.

 

My next natural choice was a miniPC with moderate expansion capabilities that I could hack so I got the HP Elitedesk 800 G5 mini last year and while it offered the RAM and storage expansion options, its extremely weak UHD630 GPU struggled to drive a display anything above 1080p so I gave up on that. This G4 Mini with RX560X has been just the perfect machine for everything I do and Apple's own negligence and non-sensical price pushed me towards hack. I have no plans to upgrade until I can no longer use this current machine for my every tasks.

[deeveedee]

I received notification of Sequoia Beta 6 update availability.

Spoiler

 

Note that the update is incremental (not the full installer).

Spoiler

 

EDIT: Very fast update from Sequoia Beta 5 -> Beta 6.  No issues.

Spoiler

 

Note that after upgrading from Beta 5 -> Beta 6 (as with all Betas), I disable Apple Analytics reporting (System Settings > Privacy & Security > Analytics & Improvements) (move slider to off):

Spoiler

 

Edited August 12 by deeveedee

[deeveedee]

After upgrading to Sequoia Beta 6, I am able to access the EFI with the normal 'sudo diskutil mount' (mkdir EFI and special msdos type not required).  It's possible that Sequoia Beta 6 fixed a problem introduced with Beta 5.  Continuing to monitor this before making a final conclusion.

[ird]

7 hours ago, deeveedee said:

I received notification of Sequoia Beta 6 update availability.

  Reveal hidden contents

 

Note that the update is incremental (not the full installer).

  Reveal hidden contents

 

EDIT: Very fast update from Sequoia Beta 5 -> Beta 6.  No issues.

  Reveal hidden contents

 

Note that after upgrading from Beta 5 -> Beta 6 (as with all Betas), I disable Apple Analytics reporting (System Settings > Privacy & Security > Analytics & Improvements) (move slider to off):

  Reveal hidden contents

 

 

All the reading I've done so far seems to point to BlueToolFixup.kext as the culprit for incremental updates failing and forcing a full update (after of course, disabling SecureBootModel). Seems to be orthogonal to what SMBIOS one uses (so MacMini8,1 users should also see the same issue if they use BT).

 

I will have to wait until Apple issues the next incremental update to confirm this theory.

[deeveedee]

@ird That makes sense.  The next time you have an issue and need help, please post your current EFI that is causing the problem. The EFI that you previously posted does not have BluetoolFixup.kext and SecureBootModel=Disabled.

[ird]

18 hours ago, deeveedee said:

@ird That makes sense.  The next time you have an issue and need help, please post your current EFI that is causing the problem. The EFI that you previously posted does not have BluetoolFixup.kext and SecureBootModel=Disabled.

 

Yes, sorry about that. It's unfortunately not possible to attach EFI with BT firmware kexts as it blows up the size to >30MB (even after compression). Single file is >10MB which exceeds the file size attach limit, the only way is to use github. 

[deeveedee]

5 hours ago, ird said:

 

Yes, sorry about that. It's unfortunately not possible to attach EFI with BT firmware kexts as it blows up the size to >30MB (even after compression). Single file is >10MB which exceeds the file size attach limit, the only way is to use github. 

Ok.  Without an accurate representation of your actual EFI, we're less likely to be able to help.  Post at least your current config.plist and/or reference your EFI stored in an external repo.

[ird]

3 hours ago, deeveedee said:

Ok.  Without an accurate representation of your actual EFI, we're less likely to be able to help.  Post at least your current config.plist and/or reference your EFI stored in an external repo.

Agreed about the need for full EFI. The GitHub link with full EFI is the post above though that I referenced, specially this part:

 

“For EFI with wifi+BT drivers and firmware, you can grab it from my github repo: Github link”

 

[deeveedee]

I disassembled my RX560x card to capture the photos below (and to clean-off the dust and apply new thermal paste).  The variant of the AMD chipset on this card (with AMD 216-0909018) is the "embedded" version that is also on the Lenovo Yoga All-In-One motherboard.  Based on brief searching, it appears that our RX560x chipset is different from what would normally be on a discrete RX560 graphics card.  I haven't searched extensively for chipset specs and am hoping that based on these photos, others might determine more about our RX560x.

 

Spoiler

 

Spoiler

 

Spoiler

 

Spoiler

 

[deeveedee]

EDIT3: Leaving the post below, but I now believe that I never tested Unigine Valley benchmarks in Full-Screen mode on an iGPU-connected display.  The results below are no longer surprising to me, because Unigine Valley is not a Metal benchmark (it was released in 2013 - 11 years ago as of this post).  To summarize, Unigine Vallue benchmarks demonstrate full dGPU graphics acceleration (not Metal) on dGPU and iGPU-connected displays in Windowed (not Full-Screen) mode and full dGPU acceleration (not Metal) on a dGPU-connected display in Full-Screen mode.  GFXBench Metal (a Metal Benchmark) is fully accelerated in Full-Screen mode on both dGPU and iGPU-connected displays.

 

=====================================

 

I'm performing new dGPU performance testing and I am finding that if I run Unigine Valley benchmark in Full Screen mode on an iGPU-connected display, the benchmark runs at iGPU frame-rates (pathetic).  If I run Unigine Valley benchmark in "Windowed" mode on an iGPU-connected display, it runs at RX560x-accelerated frame-rates (excellent).

 

I could have sworn that I tested Unigine Valley full-screen mode before, so at this time I feel that these results are different from what I had observed when testing here.  I'll investigate to figure out what I changed to cause this.  I am currently running with the EFI that I posted in Post #1.  I have tested CFG,CFG_FB_LIMIT = 0 and 2 and the results are the same, so the new observed full-screen Unigine Valley benchmarks are not caused by CFG,CFG_FB_LIMIT.  I'm currently testing in Sonoma 14.6.1.

 

EDIT: I duplicated these "new" full-screen Unigine Valley benchmarks in Sequoia Beta 6.

 

EDIT2: With the same EFI used to test Unigine Valley benchmark above, GFXBench Metal benchmark runs fully accelerated by RX560x dGPU in Full-Screen mode on dGpu AND iGPU-connected displays... so, at this time, the full-screen issue on a iGPU-connected display is only with Unigine Valley.

Edited August 15 by deeveedee