MP 1,1-5,1 OpenCore on the Mac Pro

Status
The first post in this thread is a WikiPost, and can be edited by anyone with the appropriate permissions.

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
smcFanControl also work. So, it's just iStat menu can't see the fans
Screenshot 2019-10-31 at 8.17.39 AM.png


And smcFanControl can also display my CPU temperature on the menu bar. So, those sensors are actually working, just depends on if the software can get them correctly.
 
Last edited:
  • Like
Reactions: octoviaa

octoviaa

macrumors regular
Oct 19, 2013
111
39
Uh, you have double the processors.

Edit: And just about double my score with a single x5675...
I'm comapring with the same 2 proc.
Anyway h98 and gig confirm their score is back to normal now. So at least we don't have to worry about OpenCore reducing the performance.
 
  • Like
Reactions: h9826790

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
I'm comapring with the same 2 proc.
Anyway h98 and gig confirm their score is back to normal now. So at least we don't have to worry about OpenCore reducing the performance.
Yeah, I am very happy with the setup now.

I can now have full HWAccel. Most likely the Airplay Display will also work, but I can't test it yet.

CPU performance normal, and I can see the CPU diode temperature.

Fans normal, and I can control them via smcFanControl (MacsFanControl also work).

Even though it's not the best solution (not ident as 5,1), but at least everything work.

Most likely I will keep this as my final config (for Mojave).
Screenshot 2019-10-31 at 9.27.51 AM.png


For Catalina, I prefer use the Mac Pro 7,1 SMBIOS more. So, all Display ports can work without patching. Also, I love to see the computer still ident itself as Mac Pro, and love to see that PCIe Cards page in About This Mac :D
Screenshot 2019-10-30 at 8.15.52 AM.png


P.S. The serial number and UUID are actually there, but I removed them from the screen capture, because I will keep this config.
 

ehsan58

macrumors newbie
Oct 30, 2019
1
0
hi can i use this method with cMP 3.1? High Seira 10.13.6 + windows 7 and boot screen for unflashed ATI 5770
 

MoerBoer

macrumors regular
Jan 27, 2018
121
61
Wow, this is amazing!

I wonder if FileVault will work now, and if you will be able to see the unlock screen.
 

richgoga

macrumors regular
Oct 11, 2013
137
56
The setup described in post #1 does not touch platform information at all. Most would be hard pressed to find differences between such a setup and a natively booted one.

Modifying the SMBIOS in automatic mode is what seems to lead to the wrong frequency. Perhaps manual mode gives more control. This remains to be tested.
So Post 1 setup will give me a Catalina install that I can update using Software Update?
Why are you all working SMBIOS tweaks? To enable stuff not natively supported by a 5.1?
 

gigignopechegno

macrumors newbie
May 11, 2016
11
10
Thanks for all work guys! There is no problem with OpenCore and SMBios ImacPro1,1 tweak with dual cpus Mac Pro, right?

for now work like a solid rock. Davinci Resolve with footage REd Monstro 8k and Arri Alexa raw whit a lot of denoise for the test and conversions ih h265 4 h264 without a hitch. h9826790 will make a tutorial as soon as everything is complete.
 

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
So Post 1 setup will give me a Catalina install that I can update using Software Update?
Why are you all working SMBIOS tweaks? To enable stuff not natively supported by a 5.1?
cdf introduce this method to allow cMP users run vanilla Catalina, and allow to update the OS via the native way.

I am the one that who suggest to mess around with SMBIOS, which may brick the cMP, who knows? The main reason is to unlock HEVC hardware encoding.

When boot as 5,1, no matter using OpenCore or not, the best I can achieve still just H264 HWAccel, and HEVC hardware decode.

For those who only want to run Catalina. My suggestion is to stick with cdf's method in post #1, and keep the cMP as clean as possible. There is no need to touch the SMBIOS.
- - Post merged: - -

I want a CPU expert to give me some help.

As per the OpenCore document.

FSBFrequency

Type: plist integer, 64-bit
Failsafe: Automatic
Description: Sets FSBFrequency in gEfiProcessorSubClassGuid.
Sets CPU FSB frequency. This value equals to CPU nominal frequency divided by CPU maximum bus ratio and is specified in Hz. Refer to MSR_NEHALEM_PLATFORM_INFO (CEh) MSR value to determine maximum bus ratio on modern Intel CPUs.

Note: This value is not used on Skylake and newer but is still provided to follow suit.
Any idea what's this FSBFrequency for a W3690? 133?
 
Last edited:

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
Working with Tsialex right now to determine the safety of OpenCore.

I mainly focus on "if spoofing SMBIOS is safe". This is what I really want to know. However, even we end up determine that's not safe. Simply use OpenCore to boot 5,1 as 5,1 may be still safe.

For advance user only, and only for those know how to recover a bricked logic board (e.g. with Matt card), or don't mind to brick your Mac. You are free to join ;).

Anyway, for the "sensors issue". I am now 100% sure it's just the 3rd party software (e.g. iStat menu) cannot read them properly.

When boot as iMac Pro, I can still access the complete SMC key list with all valid value (and control them).

I dumped the SMC list when boot at 5,1 (natively), and another dump when boot as iMac Pro via OpenCore. The temperature SMC key lists are identical (of course, the values can be different. That's the associated temperature at that moment)
Screenshot 2019-10-31 at 9.53.14 PM.png


Same thing observed for the fans SMC keys. And as you can see, I want the NorthBridge to run cooler, therefore, I set F4Mn (Booster A fan minimum speed) to 1000, which keep TN0D (NorthBridge diode temperature, from the above screen capture) at ~60°C.
Screenshot 2019-10-31 at 9.54.04 PM.png


So, at least we know all the sensors work. No need to worry about that.
 

CMMChris

macrumors member
Oct 28, 2019
51
42
I mainly focus on "if spoofing SMBIOS is safe". This is what I really want to know. However, even we end up determine that's not safe.
Why should it be unsafe to spoof a different SMBIOS? For the reason of firmware updates? It would surprise me if Apple doesn't run a compatibility check before flashing the EFI.
 

startergo

macrumors 68000
Sep 20, 2018
1,739
679
Why should it be unsafe to spoof a different SMBIOS? For the reason of firmware updates? It would surprise me if Apple doesn't run a compatibility check before flashing the EFI.
The real problem is if the OC writes to sections of NVRAM not designed for this like clover does on the cMP5.1. For instance cMP3.1 is safe for clover but it has different structure . According to Vitaly (one of the developers) OC should not write where it is not supposed to. I would recommend back up a snapshot of the firmware before OC usage and after and compare the snapshots.
 
  • Like
Reactions: Macschrauber

w1z

macrumors 6502a
Aug 20, 2013
528
335
The real problem is if the OC writes to sections of NVRAM not designed for this like clover does on the cMP5.1. For instance cMP3.1 is safe for clover but it has different structure . According to Vitaly (one of the developers) OC should not write where it is not supposed to. I would recommend back up a snapshot of the firmware before OC usage and after and compare the snapshots.
I actually did that to see if there were any changes and saw none. Compared using binwalk.

Though, I am only using OC to boot/Software update Catalina so maybe the situation is different for SMBIOS users.
 
  • Like
Reactions: octoviaa and cdf

cdf

macrumors 6502a
Original poster
Jul 27, 2012
539
402
There is recent evidence that macOS installer will actually try to update the firmware for a spoofed model in a hack. In this case, the updater failed and exited.
 
  • Like
Reactions: octoviaa

startergo

macrumors 68000
Sep 20, 2018
1,739
679
@startergo Neither Clover nor OC can overwrite the SMBIOS embedded inside your EFI.
I personally had my firmware corrupted by using Clover. Only because I have used the little Frank connector instead of the original firmware I was able to extract the corrupted bios and boot with the original firmware after removing the Matt card. @tsialex inspected the firmware and confirmed corruption.
 
  • Like
Reactions: octoviaa

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
I actually did that to see if there were any changes and saw none. Compared using binwalk.

Though, I am only using OC to boot/Software update Catalina so maybe the situation is different for SMBIOS users.
Binwalk can’t tell you the whole story. You need to compare the difference in Hex editor bit by bit, then determine if OpenCore is safe. Some entry in the private NVRAM section may not show up in Binwalk unless you teach it how to ident those value.
- - Post merged: - -

@startergo Neither Clover nor OC can overwrite the SMBIOS embedded inside your EFI.
I am not sure if Clover or OC can really corrupt the cMP's SMBIOS. However, Clover is proved can write something into the private NVRAM section (part of the BootROM). And once this NVRAM section fully filled, the cMP can have some abnormal behaviour.

In some case, if that section is filled by something which the cMP cannot understand, it may even refuse to boot (brick the logic board).

That's why we have to do some study in this area.

We introduce this OpenCore workaround for cMP users to more fully utilise their Mac. But we don't want anyone to brick their Mac because of this workaround.
 
Last edited:

h9826790

macrumors G5
Apr 3, 2014
13,477
6,178
Hong Kong
This is what can be done via OpenCore in Catalina.
Screenshot 2019-11-01 at 4.30.13 AM.png


Ident as Mac Pro in About This Mac (but if further go to System Report. It will shows as 7,1).

All display ports work natively (I didn't even install Lilu and WhateverGreen).

BootROM keep 144.0.0.0.0

SMC keep 1.39f5 for 2009 cMP

System Report still suggest there is a processor tray.

CPU performance normal (but Clock speed report as 3.5GHz for my W3690, obviously this is incorrect)

Temperature sensors work. Fans work. Can build my own fan profile via TG-Pro.

Full HWAccel.

Sidecar shows up natively.

MacOS can be updated via system preferences natively.

I am not sure if I still miss anything, but if I want to run Catalina on my cMP. Most likely this is what I want to see.

Anyway, there is a warning in the Memory tab.
Screenshot 2019-11-01 at 2.50.11 AM.png


But I am quite sure that's just cosmetic error.
Screenshot 2019-11-01 at 3.36.26 AM.png
 
Last edited:

startergo

macrumors 68000
Sep 20, 2018
1,739
679
I wonder if this will bring the efi boot screen on RX-580. From the OC manual:

"On HiDPI screensAPPLE_VENDOR_VARIABLE_GUID UIScaleNVRAM variable may need to be set to02to enableHiDPI scaling in FileVault 2 UEFI password interface and boot screen logo. Refer to Recommended Variablessection for more details.Note: This will fail when console handle has no GOP protocol. When the firmware does not provide it, it can beadded withProvideConsoleGopUEFI quirk set totrue"

" 6.ProvideConsoleGopType plist booleanFailsafe:falseDescription: macOS bootloader requires GOP (Graphics Output Protocol) to be present on console handle.This option will install it if missing "
 
Last edited:
  • Like
Reactions: octoviaa
Status
The first post in this thread is a WikiPost, and can be edited by anyone with the appropriate permissions.