Proxmox VE OSX Guide discussion

[iGPU]

There was another entry to adjust DROMs for TB SSDT files! 

 

This one actually works better (and is easier to use), than the one I posted earlier. See this post here. From this post, you can download "HackinDROM".

 

If we can get TB passed, the best fit will probably be the Gigabyte Titan Ridge AIC.

[iGPU]

I've been asked how to make the Enermax Liqtech TR4 II stable, as it has a bad reputation on the internet. Many of these pumps have failed due to internal corrosion thought to be due to poor or non-existent anti-corrosion solution.

 

Before using, I 'repaired' the cooler: first, by flushing out the radiator and pump housing, next by re-filling with a good quality, anti-corrosion solution, and finally, by re-surfacing the plate.

 

I used these two videos as a start for the tear-down (the pump method is cleaner; I did the work in my kitchen sink):

https://www.youtube.com/watch?v=luUwpbdYnaQ

https://www.youtube.com/watch?v=Hx-fhoB5Gyo

 

I used this pump to flush the radiator several times with water before re-filling:

https://www.amazon.com/gp/product/B07GR1HSR7/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

 

I used this solution (it's a concentrate; follow directions on bottle to dilute with water) to re-fill the radiator/pump:

https://www.amazon.com/gp/product/B00CDXQ22M/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

 

I then re-surfaced the plate, as I already posted in this thread here, to make the surface flat to better transfer heat from the CPU.

 

After the above steps, the pump works great with good thermals and it is very quiet. More importantly, after these 'repairs', the functionality and longevity should be much improved. BTW, the fans that come with the pump are actually very good and I did not replace with Noctua fans as originally planned.

 

When setting pump speed in BIOS, you want to set the speed at a minimum of 2500 rpm (less than 2000, the pump won't work and the LEDs won't turn on: the LEDs are a good indicator that the pump is working). I think it's now running closer to 3000 rpm. If running at full speed (>4000 rpm), it is too fast and I think will shorten life-span of the pump. I mainly use fan speeds to vary cooling and keep sound at minimum when thermals are low.

Edited July 8, 2020 by iGPU

[fabiosun]

Big Sur beta 2

More simple than before

Latest open core available in master (download it with OC Efi maker if you want a simple way)

No need to change VM config in Penryn cpu now 

[iGPU]

I finally got kexts working in OC v060 28-June. (I'll update to latest OC in a few minutes.) BT and Ethernet now working just fine.

 

To check kext functionality, run this in Terminal: kextstat | grep -v com.apple

 

 

 

What was preventing kexts from loading was a hold over entry for Kernel/Emulate. Once I deleted the entries for Cpuid1Data and Cpuid1Mask, things were fine.

 

 

 

I also updated to ß2 this morning without any issues.

 

 

 

I rather like Big Sur better than Catalina (and I'd preferred Mojave over Catalina).

 

Edit:

I have seen one glitch with ß2. When trying to access "Accessibility" item (I was looking to increase size of the cursor) within System Preferences section, I get a spinning beachball and must force quit Preferences. I'd not seen that on ß1 (but no size adjustment was visible). <--- I had VirtualSMC.kext enabled. I disabled and enabled WEG. Now Accessibility is working again. In summary: I've left VirtualSMC disabled; WEG, enabled.

Also, when trying to adjust the Date & Time/Menu Bar Clock, items are grayed out in these early betas.

 

Edited July 10, 2020 by iGPU
glitch comment at bottom

[iGPU]

I updated to latest OC: v060, 5 July. Now all boots just fine. It was the kext related Cpuid discussed above that was apparently hampering a boot with latest OC.

 

And with this latest OC, no extra "-lilubetaall" or "booter-fileset-basesystem" or "booter-fileset-kernel" arguments are required in the NVRAM section for kext injection.

 

 

 

Presently, I am not enabling VirtualSMC.  WEG is again enabled; it now seems okay with Radeon VII cards.

 

I'll be re-testing WiFi stability with my build and BigSur, since when running Mojave or Catalina, there were stability issues when WiFi was enabled.

Edited July 10, 2020 by iGPU
WEG comments updated.

[iGPU]

13 hours ago, fabiosun said:

Big Sur beta 2

More simple than before

Latest open core available in master (download it with OC Efi maker if you want a simple way)

No need to change VM config in Penryn cpu now 

 

 

fabiosun,

 

I found that I could not boot if I used 'host' instead of 'Penryn'. The Apple logo appeared, but no progress bar. Strange.

[Driftwood]

Im able to restart Ok inside Mac Prox without Radeon problems. Summit to do with the latest Updates in Proxmox I guess. Also got my front side USBs working without grief to the audio.

 

Just tried a discounted RME AIO DSPe card in the machine. No life in it though. Tried Mac, Windows, BIOS, editing PCIe lane speeds, etc... so this is DOA and Im sending it back and may have to pay for a retail priced one 😞 Determined to get it working though. If the FW card works, this should.

 

 

 

[Driftwood]

31 minutes ago, iGPU said:

I found that I could not boot if I used 'host' instead of 'Penryn'. The Apple logo appeared, but no progress bar. Strange.

 

Had same problem @iGPU If you verbose you will get it hanging on screen 2.

 

 

[iGPU]

I've been testing WEG with Radeon VIIs. I now see no problems running under Big Sur. I used to see freezes within just a few minutes, but not now. Good news.

 

I'm also testing WiFi issues. I now seem to have no problems with it under Big Sur (USB audio dropout). But, the previous problems may have been related to 2 kext files. Historically, I've used several kext files for BT and WiFi: AirportBrcmFixup.kext, BrcmBluetoothInjector.kext, BrcmFirmwareData.kext, BrcmPatchRAM3.kext, and BT4LEContinuityFixup.kext.

 

As I began work using Big Sur, no kext files were being loaded, so I turned off most trying to sort things out. Once the kext files were loading, I began turning them on selectively. When I turned on BrcmFirmwareData.kext and BrcmPatchRAM3.kext, I saw more erratic behavior using BT. I've now only have enabled: AirportBrcmFixup.kext, BrcmBluetoothInjector.kext and BT4LEContinuityFixup.kext. These 3 kexts seem to give good BT and WiFi response with the BCM4352 swapped card.

 

So now WiFi also seems stable. Again, good news. And this is with an early beta Big Sur.

Edited July 10, 2020 by iGPU

[Driftwood]

Big Sur: Looks like the MSI is devoid of problems unlike the ASRock @iGPU Pavo and I are trtying to get an emulated boot going on the ASRock like hes done on his MSI. Plus, hes working on a Aquantia fix.

 

Edited July 10, 2020 by Driftwood

[fabiosun]

7 hours ago, iGPU said:

 

fabiosun,

 

I found that I could not boot if I used 'host' instead of 'Penryn'. The Apple logo appeared, but no progress bar. Strange.

Post your config if you want, i have installed from scratch very easily if i compare with beta 1

[iGPU]

1 hour ago, fabiosun said:

Post your config if you want, i have installed from scratch very easily if i compare with beta 1

 

The BigSur VM config:

 

Spoiler

# 23:00 & 4d:00 = Radeon VII
# 1:00 & 2:00 = NVMe drives
# 44:00 = ASMedia USB
# 46:00 = BT/WiFi
# 47:00 = Ethernet
# 48:00 = USB
# 49:00 = SATA
#
args: -device isa-applesmc,osk="our....Inc" -smbios type=2 -cpu Penryn,kvm=on,vendor=GenuineIntel,+kvm_pv_unhalt,+kvm_pv_eoi,+hypervisor,+invtsc,+pcid,+ssse3,+sse4.2,+popcnt,+avx,+avx2,+aes,+fma,+fma4,+bmi1,+bmi2,+xsave,+xsaveopt,check
balloon: 0
bios: ovmf
boot: d
cores: 64
cpu: Penryn
efidisk0: local-lvm:vm-101-disk-1,size=4M
hostpci0: 23:00,pcie=1,x-vga=1
hostpci1: 4d:00,pcie=1
hostpci2: 01:00.0,pcie=1
hostpci3: 02:00.0,pcie=1
hostpci4: 44:00.0,pcie=1
hostpci5: 46:00.0,pcie=1
hostpci6: 47:00.0,pcie=1
hostpci7: 49:00.0,pcie=1
hostpci8: 48:00,pcie=1
machine: q35
memory: 65536
name: BigSur
numa: 1
ostype: other
smbios1: XXXXXXX-XXXX-XXXX-XXXX-XXXXXXXX
sockets: 1
tablet: 0
vga: none
vmgenid: XXXXXXX-XXXX-XXXX-XXXX-XXXXXXXX

 

 

Edited July 10, 2020 by iGPU

[iGPU]

I forgot to mention that I've now ordered 2 more FireWire cards that looked exactly like Driftwood's photo (an older style of power plug is present on card). But what arrived were new models with TI chips, which meant bridges. I had to return them. I now give up on buying more FireWire cards.

 

Unless we can come across some new-old-stock ones, we're out of luck until bridges can be passed-through.

[iGPU]

fabiosun, you interested me in looking at CPU history (part of Activity Monitor utility) when running Cinebench 20.

 

All cores are simultaneously activated, as shown in the green peaks:

Edited July 10, 2020 by iGPU

[fabiosun]

3 hours ago, iGPU said:

 

The BigSur VM config:

 

  Reveal hidden contents

# 23:00 & 4d:00 = Radeon VII
# 1:00 & 2:00 = NVMe drives
# 44:00 = ASMedia USB
# 46:00 = BT/WiFi
# 47:00 = Ethernet
# 48:00 = USB
# 49:00 = SATA
#
args: -device isa-applesmc,osk="our....Inc" -smbios type=2 -cpu Penryn,kvm=on,vendor=GenuineIntel,+kvm_pv_unhalt,+kvm_pv_eoi,+hypervisor,+invtsc,+pcid,+ssse3,+sse4.2,+popcnt,+avx,+avx2,+aes,+fma,+fma4,+bmi1,+bmi2,+xsave,+xsaveopt,check
balloon: 0
bios: ovmf
boot: d
cores: 64
cpu: Penryn
efidisk0: local-lvm:vm-101-disk-1,size=4M
hostpci0: 23:00,pcie=1,x-vga=1
hostpci1: 4d:00,pcie=1
hostpci2: 01:00.0,pcie=1
hostpci3: 02:00.0,pcie=1
hostpci4: 44:00.0,pcie=1
hostpci5: 46:00.0,pcie=1
hostpci6: 47:00.0,pcie=1
hostpci7: 49:00.0,pcie=1
hostpci8: 48:00,pcie=1
machine: q35
memory: 65536
name: BigSur
numa: 1
ostype: other
smbios1: XXXXXXX-XXXX-XXXX-XXXX-XXXXXXXX
sockets: 1
tablet: 0
vga: none
vmgenid: XXXXXXX-XXXX-XXXX-XXXX-XXXXXXXX

 

 

sorry I mean opencore config

 

[fabiosun]

3 hours ago, iGPU said:

fabiosun, you interested me in looking at CPU history (part of Activity Monitor utility) when running Cinebench 20.

 

All cores are simultaneously activated, as shown in the green peaks:

 

So it is correct , before you don't have this behaviour?

would be also interesting for you (MSI user) art with PBO in auto mode and then in enable mode..

the second option seems to be more aggressive for frequency and it produces during OCCT test a OC about 4K for all cores...

I have not verified in that condition TDP..but I think it is about 400 W

in auto mode more check are there (I think) and it is more safe for a daily use

 

[fabiosun]

11 hours ago, Driftwood said:

Big Sur: Looks like the MSI is devoid of problems unlike the ASRock @iGPU Pavo and I are trtying to get an emulated boot going on the ASRock like hes done on his MSI. Plus, hes working on a Aquantia fix.

 

Explain please what you are trying to do with Pavo (emulated boot??)

By the way have you and Pavo published in some place a guide as you said some weeks ago?

 

edit:

which fix for acquantia?

or better which problem do you have with it?

 

[Driftwood]

On 7/10/2020 at 11:16 AM, fabiosun said:

which fix for acquantia?

or better which problem do you have with it

Its to do with Big Sur getting Aquantia working on boot for beta 2. ASRock wont install from USB beta 2 Flash, it appears.

Ive done quite a bit on the write up, but Im still fine tuning so once Im happy Ill finish it. Then theres work...

Pavo did a write up some time ago on his github btw

Edited July 11, 2020 by Driftwood

[fabiosun]

@Driftwood I was curious about acquantia problem with Big Sur because I have not with it.

problems for first beta 1 was related to the bootloader and vm config (kernel and cpu type).

with latest oc I have installed from scratch without modify my vm config which is the same for all OS X I have installed

 

when you have done with the guide I will very happy to read and maybe with it I could improve also my old one👍

[iGPU]

5 hours ago, fabiosun said:

sorry I mean opencore config

 

 

Attached is a current config.plist file.

 

There are references to SSDTs I use specific to my mobo (but SSDTs not included) as well as typical kexts being used or disabled (like VirtualSMC). Also, I removed details inside PlatformInfo section (it is not functional as written; so if using, replace this section).

 

And finally, I did re-place the "booter" comments inside NVRAM for some more testing. As is, this config file boots fine into ß2 with latest release of OC (5 July, v060).

config.plist.zip

Edited July 10, 2020 by iGPU

[Driftwood]

Is everyone using 6.2.6 Proxmox? And use Backup Server? Its been updated today.

Edited July 10, 2020 by Driftwood

[fabiosun]

 No, tomorrow i will check and i will try to update if it is also kernel related

thank you for the info

 

[Driftwood]

@fabiosun@Rox67erI am currently running the 'lower voltage' O/C settings in BIOS to achieve constant 4.0 Ghz clock speed

 running 51% avg on idle to 75% under regular GFX work stress.  95% still switches off the system if overstressed (think Davinci Noise reduction Temporal/Spatial noise reduction tab!)

 

Are you saying if you run AMD Master once on Windows and the PBO stuff that you can achieve regulated clock speeds accordingly where needed inside the Mac OS VM ? or do you have to boot windows, do AMD Master, then soft reboot and go into Proxmox to keep the AMD Master settings going?

[fabiosun]

@Driftwood I do not not what AMD master is 🙂

I am saying I have very different behaviour if I put in my PBO bios option enabled instead auto mode

with enabled CPU is pushed during test as rendering or in OCCT stress test @ about 4000 Mhz

not graphic card related. CPU I mean

[fabiosun]

@iGPUwhen you have time can you post a screen capture of your BIOS settings for Cpu fan and Pump?

 

Temperature with Enermax liquid cooler are very good, but I have a weird whistle I can hear  (it seems a sort of coil wine)and also Cpu fan have too much ramping also in idle

Thank you