Big Sur on HP EliteDesk 800 G4/G5 Mini - The Perfect MacMini8,1 Hackintosh - OpenCore

[rafale77]

Adding my i9-9900 Geekbench score (getting the same as on Catalina 1330/6900 and attached my bios boot screen image which are now better centered.

 

[rafale77]

Being a bit nitpicking here but I am wondering how the ACPI settings affect the fan control.

I jacked up the idle fan setting to 80% in the BIOS and can definitely see that thermal throttling is happening during geekbench when I keep it at 20% as I am seeing improved multi-core scores now. I see too much lag between the temperature raising and the fan kicking in and wish I could control the fan curve better.

On a side note, the only thing not working perfectly at the moment is the watch unlock which apparently is a common problem as reported here.

Also I noticed in the BIOS that these puppies should have an onboard microphone which doesn't seem to show up with the ALC layout ID 20... Apparently will have to wait for a new version of AppleALC as seen here

 

[deeveedee]

So i checked our HP mini - it works under clover on Big Sur.
Clover r5126, disabled driver AptioMemoryFix, enabled drivers OcQuirks and OpenRuntime.
Goodbye Catalina! You were a good friend)

Maybe you can post your CLOVER r5126 EFI? Since we had already switched to CLOVER r5122 and had switched from AptioMemoryFix to OpenRuntime here for Catalina, is this as simple as upgrading to r5126 or are there other Quirks that need to be incorporated into the CLOVER config.plist?

I jacked up the idle fan setting to 80% in the BIOS and can definitely see that thermal throttling is happening during geekbench when I keep it at 20% as I am seeing improved multi-core scores now. I see too much lag between the temperature raising and the fan kicking in and wish I could control the fan curve better.

Also I noticed in the BIOS that these puppies should have an onboard microphone which doesn't seem to show up with the ALC layout ID 20... Apparently will have to wait for a new version of AppleALC as seen here

At some point, you're going to reach the thermal dissipation limit of the EliteDesk Mini's heatsink/fan. I had mentioned delidding the CPU here, but it might not make sense given the relatively small heatsink and airflow. Try running without the cover to see if that makes a difference. I think you first need to prove that you can actually remove more heat to improve performance. Maybe even try an external fan blowing on the heatsink with the cover removed.

A user had determined that VoodooHDA enables the Microphone here. It will be good when AppleALC does, too. Nice find!

@rafale77 I just ran GB5 without the cover and observed the attached CPU benchmarks. No difference for me, but my i7-8700 doesn't dissipate as much heat as your I9-9900. My system has an all copper heatsink and I replaced the stock thermal compound with AS5. Note that our systems use the single heatsink to cool the CPU AND the MOSFETs powering the CPU. Not a recipe for peak performance.

 

[rafale77]

@deeveedee, my test with the high idle fan rate was exactly to do what you suggested. My SMC monitor actually reports a notification on screen when the CPU throttles and at 80% fan rate, there was no throttling so the scores I posted are as high as the machine will go. ~1350/~7100 without throttling and ~1300/~6850 when it throttles at 20% fan rate. I found a sweet spot for now at 40% fan rate for noise and vs performance where I can get ~1350/~6950 on geekbench.

I also modified your USBPorts.kext to reflect the actual port types after remapping in an attempt to fix the bluetooth/watch unlock issue. Will be posting it if it works.
If this watch unlock and the onboard microphone issue can get resolved, I will have a full hardware functionality.

 

[deeveedee]

EDIT: I observed the real time clock corruption again when when using OC's VerifyMsrE2.efi after enabling the OC Cfg Lock Quirks, so the OC Cfg Lock Quirks did not resolve the real time clock corruption. I still haven't figured out the conditions that cause RTC issues, since I don't observe this during "normal" operation.
----------------------------------------
I finally got around to testing MSR 0xE2 register (using OC VerifyMsrE2.efi tool described here). The MSR 0xE2 register is locked in the HP EliteDesk 800 G4 Mini BIOS (see attached screenshot). I have attached a revised OC 0.6.3 EFI to Post #1. This revised EFI enables Kernel Quirks AppleCpuPmCfgLock and AppleXcpmCfgLock (which others had already done. I did not since my system seemed to work fine with these Quirks disabled). I'm not sure if locked MSR 0xE2 explains the real time clock error that I observed during the Big Sur install and I'm not sure if both AppleCpuPmCfgLock and AppleXcpmCfgLock need to be enabled. I'd prefer to make the suggested BIOS mod, but that's not a high priority for me. If someone else wants to try the BIOS mod (described here), that would be great!

Note that I have not tested my posted EFI with any other OS (e.g. Windows). I dual boot my system, but I don't use OC to boot Windows. In rare instances when I want to boot Windows, I manually select Windows by pressing F9 at startup.

EDIT: I attached GB5 CPU benchmarks (collected after enabled Kernel CfgLock Quirks). No change in GB5 CPU benchmarks, so enabling the Quirks does not appear to have affected CPU Power Management.

 

[deeveedee]

@deeveedee, my test with the high idle fan rate was exactly to do what you suggested.

Ok. If you think the fan in our EliteDesk Mini's is sufficiently able to dissipate the heat. I have my doubts about that.

 

[rafale77]

Ok. If you think the fan in our EliteDesk Mini's is sufficiently able to dissipate the heat. I have my doubts about that.

Yeah I am fairly sure, Through the entire test, the CPU cores all remain below 80C so if it is throttling, it should be due to power limit from the BIOS and not because of temperature. I can see the power consumption of the CPU stay at 67W (65W TDP) under stress while it can spike up to 90W for short period of times. The single core scores are now a little higher than what I was getting on my 9900K (90W) but the multi-core are still 1200pts lower which may be expected...

PS: There are other factors like the RAM speed and as I am browsing the results on geekbench, it seems like the scores I am getting are pretty normal for a vanilla i9 9900 with 2666MHz DDR4. My previous build had 3466MHz DDR4 and it seems to be adding a bit in the multi-core score. Knowing that it is powered by a 90W power brick, it is not bad at all. Without the board specific power limitation (allowing 130+W spikes, it looks like I should be getting 7800-8000.) Given the fact that temperatures are not really a problem and the gain from delidding the 9th gen CPUs is much smaller than for the 8th gen, it's not something I am considering.

 

[rafale77]

I finally got around to testing MSR 0xE2 register (using OC VerifyMsrE2.efi tool described here). The MSR 0xE2 register is locked in the HP EliteDesk 800 G4 Mini BIOS (see attached screenshot). I have attached a revised OC 0.6.3 EFI to Post #1. This revised EFI enables Kernel Quirks AppleCpuPmCfgLock and AppleXcpmCfgLock (which others had already done. I did not since my system seemed to work fine with these Quirks disabled). I'm not sure if locked MSR 0xE2 explains the real time clock error that I observed during the Big Sur install and I'm not sure if both AppleCpuPmCfgLock and AppleXcpmCfgLock need to be enabled. I'd prefer to make the suggested BIOS mod, but that's not a high priority for me. If someone else wants to try the BIOS mod (described here), that would be great!

Note that I have not tested my posted EFI with any other OS (e.g. Windows). I dual boot my system, but I don't use OC to boot Windows. In rare instances when I want to boot Windows, I manually select Windows by pressing F9 at startup.

EDIT: I attached GB5 CPU benchmarks (collected after enabled Kernel CfgLock Quirks). No change in GB5 CPU benchmarks, so enabling the Quirks does not appear to have affected CPU Power Management.

View attachment 495877

• AppleCpuPmCfgLock: NO
  • Only needed when CFG-Lock can't be disabled in BIOS
  • Only applicable for Ivy Bridge and older
    • Note: Broadwell and older require this when running 10.10 or older
• AppleXcpmCfgLock: YES
  • Only needed when CFG-Lock can't be disabled in BIOS
  • Only applicable for Haswell and newer
    • Note: Ivy Bridge-E is also included as it's XCPM capable

Only one of the two is required.

 

[rafale77]

@deeveedee

I extracted the BIOS options from the powershell tool of my backup 600G5. I am not seeing any cfglock option but it may be called something else.

 

[deeveedee]

• AppleCpuPmCfgLock: NO
• AppleXcpmCfgLock: YES

Only one of the two is required.

I'm finding conflicting reports on the proper use of these. So far, I haven't found any combination of the two to make any difference in the behavior of my EliteDesk 800 G4 Mini. I do need to enable CLOVER's KernelPm patch in order to boot macOS with CLOVER, so I would expect to require OC's 'Kernel > Quirks > AppleXcpmCfgLock = YES'
to boot macOS with OC, but I don't. No idea why.

I extracted the BIOS options from the powershell tool of my backup 600G5. I am not seeing any cfglock option but it may be called something else.

I'm not sure how to interpret those 'BIOS options' or how we would use them. I downloaded the HP installer for the latest HP EliteDesk 800 G4 Mini BiOS and extracted the BIN firmware component (see attached archive). I can view the .bin with UEFITool, but I don't see 'Setup' in the tool, so I'm unable to find CFG Lock (let alone its offset).

EDIT: A note about the Dortania docs: The Dortania docs are incredibly well written, but not without flaws. I haven't been taking anything from the docs as gospel without trying it myself. For example, in the "converting from CLOVER" section here, the Dortania docs indicate that "Clover will auto-apply this patch [KernelPm] without setting it if the MSR E2 was locked, so you may actually need AppleXcpmCfgLock even if Clover didn't." I never found CLOVER to automatically set KernelPm.