macOS 10.14 Mojave on Unsupported Macs Thread

[AwlsomeAlex]

As far as I remember in previous versions of MacOS Safe mode worked without gpu acceleration. There's no point to change it in 10.14.

So why not boot into that with unsupported hardware in the mean time? (I'm gonna assume quality as good as Classic MacOS)

 

[stormxeron]

Apple should have at least made the Mojave as a paid option for mid 2012 earlier models and free upgrade for later models. That would have been awesome.

 

[AwlsomeAlex]

Apple should have at least made the Mojave as a paid option for mid 2012 earlier models and free upgrade for later models. That would have been awesome.

Sounds like an Apple thing to do, alas Apples thing is also killing off perfectly useable devices in favor of broken keyboards and a touch bar no one uses. Also impossible Because macOS Mojave is trying to move away from OpenGL in favor of Metal, which would be making us pay for something that’s being phased out in every user interface ran on macOS and not even supported to actually boot? Eh.

 

[e_ch]

Apple should have at least made the Mojave as a paid option for mid 2012 earlier models and free upgrade for later models. That would have been awesome.

That would be the easiest way for them to collect all existing insults in one day.

So why not boot into that with unsupported hardware in the mean time? (I'm gonna assume quality as good as Classic MacOS)

You mean using 10.14 in safe mode all the time? It's too sluggish and restricted. High sierra is not so bad.

 

[iphone2g&3gfan]

Just reading through this, what I'm getting out of it sounds like Safe Mode still loads OpenGL natively without Metal even being used, so what would be the flaws of making it so Safe Mode would HAVE to be loaded for debugging purposes on non-native machines? I honestly don't know if what I just said is a complete joke since I'm new in the MacWorld, as I had just acquired and upgraded an old MacBook Pro 2010 and hate to see supported dropped just so suddenly.

You can absolutely do that on any mac, just like setting verbose mode permanent. Just run in terminal:
$ sudo nvram boot-args="-x"

Now, I think the best way to force the OpenGL drivers to load in safe mode (because we haven't yet patched the safe mode OpenGL check out of WindowServer) is the following:

1) Copy all the kexts from High Sierra. If unsure which ones, just ask on this thread, and one of the "specialists" (I guess we'll call them that) will tell you which kexts/bundles/plugins to copy, including myself. This step includes setting the proper file permissions (ownership: root:wheel, mode: 755) + rebuilding PrelinkedKernel to include the new kexts.
2) Boot in safe mode, but ALSO in SINGLE-USER mode. So, sudo nvram boot-args="-x -s"
3) Once you arrive at a SHELL in the text-only boot, kextload the Framebuffer kext FIRST, and then the MAIN/controller kext (not the framebuffer OR accelerator)
Some examples of their filenames:

Sandy Bridge, HD 3000: AppleIntelSNBGraphicsFB.kext (framebuffer), then AppleIntelHD3000Graphics.kext (controller/main kext)

*** Nvidia Tesla Architecture: Geforce.kext/GeforceTesla.kext (basic setup, not either framebuffer or the rest, load both just in case) then NVDAResman.kext/NVDAResmanWeb.kext (framebuffer?) then NVDANV50HalTesla.kext/NVDANV50HalTeslaWeb.kext (if using the web drivers from high sierra + patched out macOS version check: https://github.com/Benjamin-Dobell/nvidia-update, if we get a web driver update for mojave current guesses include Fermi and up, so not including tesla) I think that the kext is actually called NVDANV50HalTesla.kext since Mavericks.

*** Nvidia Fermi: Geforce.kext, then NVDAResman.kext, then NVDAGF100HalWeb.kext (NOT NATIVE, MUST USE WEB DRIVERS FROM HIGH SIERRA + patched out macOS version check)

***= SEE BOTTOM of this post, help needed + explanation

AMD Evergreen/Terascale 2 (5xxx): AMDLegacySupport.kext (support kext for the next one), then AMDLegacyFramebuffer.kext (framebuffer) then AMD5000Controller.kext (controller kext)
Also load after all those, AMDRadeonX3000.kext, just in case.
5000Controller I believe to control the hardware interaction directly, and that AMDRadeonX3000.kext calls it from the software acceleration. It seems like Apple split the hardware controller and the software controller here, but go figure. Load it anyway just to be safe.

As you can see, the pattern is that it is not the framebuffer kext or accelerator bundle. It is the "controller" or "manager" of the capabilities of the card. The Framebuffer kext (i.e AppleIntelSNBGraphicsFB.kext=HD 3000) is responsible for setting up correct resolution + "CoreImage" I believe, and the Accelerator kexts (i.e AppleIntelHD3000GraphicsGLDriver.bundle, AppleIntelHD3000GraphicsVADriver.bundle=HD3000) are responsible for the individual capabilities of the hardware. The "controller" kext handles all other dependencies, and depends on the Framebuffer kext being loaded. So, the tree of GPU drivers is as follows, for my three examples, each file loading the next one in the chain, or at least allocating it into memory for the GLDriver/VADriver so it can be used by WindowServer):

HD 3000: AppleIntelSNBGraphicsFB.kext --> AppleIntelHD3000Graphics.kext --> AppleIntelHD3000GraphicsGLDriver.bundle/AppleIntelHD3000GraphicsVADriver.bundle

AMD Evergreen/Terascale 2/HD 5xxx: AMDLegacyFramebuffer.kext + AMDLegacySupport.kext* --> AMD5000Controller.kext --> AMDRadeonX3000.kext --> AMDRadeonX3000GLDriver.bundle/AMDRadeonX3000VADriver.bundle

NVDA Tesla: Geforce.kext??/GeforceTesla.kext?? --> NVDAResman.kext --> NVDANV50HalTesla.kext/NVDANV50HalTeslaWeb.kext* --> GeForceTesla.kext* --> GeforceTeslaGLDriver.bundle/GeforceTeslaVADriver.bundle


AMD *= Based on data from https://www.insanelymac.com/forum/topic/327712-black-screen-amd-radeon-hd5670/
I believe that AMDLegacyFramebuffer.kext loads resources from AMDLegacySupport.kext, and that LegacySupport is NOT a step in the init of the graphics subsystem. I think it acts like a shared library, not directly used to load the next kext. That's why I put a "+" instead of a -->. I am also not sure where the divide between AMDLegacyFramebuffer.kext/AMDLegacySupport.kext and AMDFramebuffer.kext is in terms of GPU support. Someone could please chime in on this. Also note that there is no Support kext for the AMDFramebuffer. Perhaps this LegacyFramebuffer was foreshadowing the demise of our Radeon GPU's?
Also note that AMDRadeonX3000.kext is the second layer controller kext for BOTH 5xxx and 6xxx, so BOTH Terascale 2 and Terascale 3. AMD uses a FOUR kext solution for some reason, just like Nvidia, presumably due to the need to support more models of Graphics Cards that at least share the same ARCH as stock mac ones. AMDRadeonX3000*/AMDRadeonX3000GLDriver handle acceleration setup for both of those architectures.
In fact, the ONLY difference between 5xxx and 6xxx seems that 6xxx loads AMD6000Controller.kext instead. The rest of the kexts ALL STAY THE SAME. Thus, the difference in porting these two architectures will be minimal, nearly identical processes, thus helping us more.

NVIDIA *= the "Hal" kext is for "hardware abstraction layer". Nvidia's architecture seems to feature FOUR kexts instead of three. The HAL loads first presumably to abstract the individual tesla card's technicalities, so a "one size fits all" controller can load, which then loads the graphics accelerators. For Kepler, it seems to be just Geforce.kext. HELP NEEDED ON THIS, URGENT.

*** = URGENT HELP NEEDED FOR NVIDIA. I DO NOT COMPLETELY UNDERSTAND HOW THE DRIVERS LOAD AND IN WHAT ORDER. THIS IS CRUCIAL...
sorry for the all caps, but we must know the nvidia driver loading stages for each architecture (tesla, fermi, kepler, etc) so we can properly understand how to proceed to patch Mojave.

TL;DR: Just see the top 1/3 of this post. The rest is just some details I recall on the GPU process of loading the kexts. But, if anyone has help/knowledge on NVIDIA and AMD kexts, please share that

 

[Project Alice]

I'm thinking it's the end of the line for those of us on the Mac Pro 3,1 since SSE 4.1 is likely an unpatchable limiting factor. I have two of them, one as a household media server (3.0 Ghz, 8 Core, GT120), the other as my primary desktop (2.8 Ghz, 8 Core with an ancient GTX 8800 card). The desktop machine has lasted me 3 years now as it was a gift from a friend who upgraded to a 5K iMac. The media server 3,1 was also donated from a friend whose a pro photographer and found a 2012 Mac Pro to replace it with. So I cannot complain much about my situation since outside of more hard drives I've bought over the years and a few accessories, it's not like I have that much money in these machines and the efforts of Dosdude, Czo and others on this forum has saved me from buying a new desktop machine for 3 years. I used to keep a Mac Pro 1,1 as my media server but it's now retired (but still working).

But now I think the desktop machine will need to be retired in favor of either a 5K iMac or if my wife goes along, an iMac Pro unless I feel I can wait until the magical unicorn 2019 Mac Pro. For a new media server, I'd kill for a new Mac Mini with TB3 since I have a Drobo 5D3 that would be well mated to it. But I'm growing increasingly despondent that we'll ever see such a machine.

But for all of you complaining about the 2009-2011 class of machines being deprecated, this isn't inconsistent with Apple's past actions. In the last several years, Apple has usually kept OS support for machines made in the last 5-6 years with the exception of Mac Pros which got up to 8 years of compatibility. Last year, High Sierra didn't deprecate anything (much like iOS 12 this year) which gave a lot of these older machines an additional year of life. But it had to end sometime and I'm not surprised that Metal/SSE 4.2 are the cutoffs this year.

Others have pointed out that you can install Windows 10 on much of this ancient hardware. That's true...I have Windows 10 running on a 2008 iMac that used to be in my daughter's room before she got her iPad Pro for Christmas. That iMac is sitting in my living room hooked up to my audiophile rig running Tidal/iTunes for anyone who wants to play music there. As for my other machines, I may see if I can install Windows Server 2016 on the 3,1 Mac Pro and finally retire a Dell 2900 from 2006 that still functions to this day as my household Windows Server. But Bootcamp isn't really setup up for a server install so that may be a limiting factor.

10.14 installer is booting on my 2008 3,1 without issue. When it's finished installing I'll update. But, this is a VM. I haven't tried doing it on the bare metal because the GPU I have in it is a GTX 1060 so I don't have access to a boot screen unless I put the factory card back in, which wouldn't work anyways because it's not metal supported. But parallels doesn't emulate sse4.2, so I don't see why the 3,1 MacPros wouldn't work.

Assuming everything goes right with the VM, I have a GT 740 I can throw in here which works natively being kepler, just no boot screen because its not an EFI card. Makes things interesting I guess.

 

[AwlsomeAlex]

You can absolutely do that on any mac, just like setting verbose mode permanent. Just run in terminal:
$ sudo nvram boot-args="-x"

Now, I think the best way to force the OpenGL drivers to load in safe mode (because we haven't yet patched the safe mode OpenGL check out of WindowServer) is the following:

1) Copy all the kexts from High Sierra. If unsure which ones, just ask on this thread, and one of the "specialists" (I guess we'll call them that) will tell you which kexts/bundles/plugins to copy, including myself. This step includes setting the proper file permissions (ownership: root:wheel, mode: 755) + rebuilding PrelinkedKernel to include the new kexts.
2) Boot in safe mode, but ALSO in SINGLE-USER mode. So, sudo nvram boot-args="-x -s"
3) Once you arrive at a SHELL in the text-only boot, kextload the MAIN/controller kext (not the framebuffer OR accelerator)
Some examples of their filenames:
Sandy Bridge, HD 3000: AppleIntelHD3000Graphics.kext
Nvidia Tesla Architecture: NVDANV50HAL.kext/NVDANV50HALWeb.kext (if using the web drivers from high sierra + patched out macOS version check: https://github.com/Benjamin-Dobell/nvidia-update, if we get a web driver update for mojave current guesses include Fermi and up, so not including tesla) I think that the kext is actually called NVDANV50HalTesla.kext since Mavericks.
Nvidia Fermi: NVDAGF100HalWeb.kext (NOT NATIVE, MUST USE WEB DRIVERS FROM HIGH SIERRA + patched out macOS version check)
AMD Evergreen/Terascale 2 (5xxx): AMD5000Controller.kext

As you can see, the pattern is that it is not the framebuffer kext or accelerator bundle. It is the "controller" or "manager" of the capabilities of the card. The Framebuffer kext (i.e AppleIntelSNBGraphicsFB.kext=HD 3000) is responsible for setting up correct resolution + "CoreImage" I believe, and the Accelerator kexts (i.e AppleIntelHD3000GraphicsGLDriver.bundle, AppleIntelHD3000GraphicsVADriver.bundle=HD3000) are responsible for the individual capabilities of the hardware. The "controller" kext handles all other dependencies, and depends on the Framebuffer kext being loaded. So, the tree of GPU drivers is as follows, for my three examples, each file loading the next one in the chain, or at least allocating it into memory for the GLDriver/VADriver so it can be used by WindowServer):

HD 3000: AppleIntelSNBGraphicsFB.kext --> AppleIntelHD3000Graphics.kext --> AppleIntelHD3000GraphicsGLDriver.bundle/AppleIntelHD3000GraphicsVADriver.bundle

AMD Evergreen/Terascale 2/HD 5xxx: AMDLegacyFramebuffer.kext + AMDLegacySupport.kext* --> AMD5000Controller.kext --> AMDRadeonX3000.kext --> AMDRadeonX3000GLDriver.bundle/AMDRadeonX3000VADriver.bundle

NVDA Tesla: NVDAResman.kext --> NVDANV50HalTesla.kext/NVDANV50HalTeslaWeb.kext* --> GeForceTesla.kext* --> GeforceTeslaGLDriver.bundle/GeforceTeslaVADriver.bundle


AMD *= Based on data from https://www.insanelymac.com/forum/topic/327712-black-screen-amd-radeon-hd5670/
I believe that AMDLegacyFramebuffer.kext loads resources from AMDLegacySupport.kext, and that LegacySupport is NOT a step in the init of the graphics subsystem. I think it acts like a shared library, not directly used to load the next kext. That's why I put a "+" instead of a -->. I am also not sure where the divide between AMDLegacyFramebuffer.kext/AMDLegacySupport.kext and AMDFramebuffer.kext is in terms of GPU support. Someone could please chime in on this. Also note that there is no Support kext for the AMDFramebuffer. Perhaps this LegacyFramebuffer was foreshadowing the demise of our Radeon GPU's?
Also note that AMDRadeonX3000.kext is the second layer controller kext for BOTH 5xxx and 6xxx, so BOTH Terascale 2 and Terascale 3. AMD uses a FOUR kext solution for some reason, just like Nvidia, presumably due to the need to support more models of Graphics Cards that at least share the same ARCH as stock mac ones. AMDRadeonX3000*/AMDRadeonX3000GLDriver handle acceleration setup for both of those architectures.
In fact, the ONLY difference between 5xxx and 6xxx seems that 6xxx loads AMD6000Controller.kext instead. The rest of the kexts ALL STAY THE SAME. Thus, the difference in porting these two architectures will be minimal, nearly identical processes, thus helping us more.

NVIDIA *= the "Hal" kext is for "hardware abstraction layer". Nvidia's architecture seems to feature FOUR kexts instead of three. The HAL loads first presumably to abstract the individual tesla card's technicalities, so a "one size fits all" controller can load, which then loads the graphics accelerators. For Kepler, it seems to be just Geforce.kext. HELP NEEDED ON THIS, URGENT.

Look that’s a lot to take in at this hour, I’m happy to help with my MacBook Pro 2010, in the name of Science. Just tell me what to do and I’ll be on my way. But like step by step Bc I have no clue what like kextloading is and stuff, I’m natively a Linux Power User, so it won’t be too hard to adapt to BSD-ish-notreally. And like you said, y’all are the experts.

 

[techguy15]

I normally abstain from commenting on Forums, as a Mac user from the days of SI's, constantly having to purchase newer mac's to keep up with newer OS's, there's a point when you just say pass.
At the current price of Macs, even entry-level (PB's & Imacs) puts it out of reach (Maxed out iMac Pro could cost over $17K) for many, and to top it off, we can't upgrade to 10.14 if you fall into some of the unsupported categories. I understand there needs to be a cutoff, however, if you're running High Sierra, why not support those Mac's?
Thanks Apple!

watch out saying that. I started a thread on it and got attacked for "whining" over lack of support for my 2011 MBP
[doublepost=1528257015][/doublepost]

I am currently typing this on a mid-2009 MacBook Pro thanks to the work that @dosdude1 has done but I am still pissed off that people are complaing about not being able to run Mojave on a almost 10 year old system...c'mon and get real! If you are so impressed that Windows 10 runs on your old system...switch! If not, be happy with everything you got out of your system and be thankful for dudes like @dosdude1 who have done all the work they have...when he didn't need to. Sure I'm disappointed that I can't run Mojave, but I'm not going to bitch about Apple...my computer is almost 10 years old!!!

 

[iphone2g&3gfan]

Is something like this possible to use ATI 5870 Accelerators with ATI 5750?

All the architectures of a GPU maker share kexts. 5xxx series should already work out of the box. As a rule, if Apple uses at least one card from a GPU architecture, then the kexts supporting that default card will support all the other graphics cards of the same architecture. 5770/5870 are Evergreen/Terascale 2, so they share the same code, and they used those 2 stock in macs. I have an HD 5570, which works OOB, since it uses the same architecture. "It only takes one to support all" - for macOS gpus.
[doublepost=1528257240][/doublepost]

Look that’s a lot to take in at this hour, I’m happy to help with my MacBook Pro 2010, in the name of Science. Just tell me what to do and I’ll be on my way. But like step by step Bc I have no clue what like kextloading is and stuff, I’m natively a Linux Power User, so it won’t be too hard to adapt to BSD-ish-notreally. And like you said, y’all are the experts.

I know both systems really well (Linux + Darwin/XNU/macOS) so I would be glad to help you with anything. Thank you for offering your support, its people like you that really make stuff like this work. ATM we're just setting up the "methodology" for testing (I guess) so just hang tight for a few more days until I get down a "reporting" procedure to this thread that is standardized so we can compare results easily. I'll do my best, just post in this thread along with everyone else so all knowledge we gain can be used to improve the situation
P.S my peak of macos knowledge was about 3 months ago, when I switched to linux on my 2 macs. I'm just getting acquainted again, so excuse me for any slight errors...
[doublepost=1528257466][/doublepost]

10.14 installer is booting on my 2008 3,1 without issue. When it's finished installing I'll update. But, this is a VM. I haven't tried doing it on the bare metal because the GPU I have in it is a GTX 1060 so I don't have access to a boot screen unless I put the factory card back in, which wouldn't work anyways because it's not metal supported. But parallels doesn't emulate sse4.2, so I don't see why the 3,1 MacPros wouldn't work.

Assuming everything goes right with the VM, I have a GT 740 I can throw in here which works natively being kepler, just no boot screen because its not an EFI card. Makes things interesting I guess.

You can still throw in the GT 120 just for the bootscreens. Sure, it won't have acceleration, but it won't do any harm either...

 

[Luba]

Now to find out what Metal cards are supported on the Mac Pros. 4,1 to 5,1 conversions should have no problem installing with a proper card. I'm assuming most AMD RX cards are fine, but what about nVidia?

So the Radeon HD 4870 that came with the '09 Mac Pro is NOT a GPU that supports Metal?

 

[iphone2g&3gfan]

10.14 installer is booting on my 2008 3,1 without issue. When it's finished installing I'll update. But, this is a VM. I haven't tried doing it on the bare metal because the GPU I have in it is a GTX 1060 so I don't have access to a boot screen unless I put the factory card back in, which wouldn't work anyways because it's not metal supported. But parallels doesn't emulate sse4.2, so I don't see why the 3,1 MacPros wouldn't work.

Assuming everything goes right with the VM, I have a GT 740 I can throw in here which works natively being kepler, just no boot screen because its not an EFI card. Makes things interesting I guess.

Theoretically, if you boot it on real hardware, it should kernel panic (due to the lack of SSE4.2 instructions in pre-Core "i" architectures...) I am surprised that it works on a VM, what is the VM software you are using?
[doublepost=1528257970][/doublepost]

So the Radeon HD 4870 that came with the '09 Mac Pro is NOT a GPU that supports Metal?

Nope, sadly. AMD 7xxx or later (excluding older architecture rebrands like the 7570) is required, A.K.A "Southern Islands"/GCN 1.0 GPUs.

 

[e_ch]

All the architectures of a GPU maker share kexts. 5xxx series should already work out of the box. As a rule, if Apple uses at least one card from a GPU architecture, then the kexts supporting that default card will support all the other graphics cards of the same architecture. 5770/5870 are Evergreen/Terascale 2, so they share the same code, and they used those 2 stock in macs. I have an HD 5570, which works OOB, since it uses the same architecture. "It only takes one to support all" - for macOS gpus.

So I don't need kext manipulation with 5750? If i've got right, to make 10.14 work on iMac 11,3 I need just to full OS with Mac model check or it should KP/turn off/freeze because actual gpu can't work with metal?

 

[iphone2g&3gfan]

So I don't need kext manipulation with 5750? If i've got right, to make 10.14 work on iMac 11,3 I need just to full OS with Mac model check?

You can make 10.14 boot (CPU has required instructions) but you are gonna be out of Graphics Acceleration, unless you upgrade the MXM GPU to something supported in macOS 10.14: AMD "Southern Islands"/GCN 1.0/7xxx or later, and Nvidia "Kepler"/6xx/some quadros. Nvidia Fermi is still a possibility, but those were never native to macOS; hence the wait for the Web Driver update that will make or break compatiiblity. ATM, you can't even use Maxwell or Pascal (9xx and later) due to a lack of Web Drivers, unless you strip out the macOS version check with nvidia-update. Only Kepler for nvidia works out of the box.

 

[e_ch]

You can make 10.14 boot (CPU has required instructions) but you are gonna be out of Graphics Acceleration, unless you upgrade the MXM GPU to something supported in macOS 10.14: AMD "Southern Islands"/GCN 1.0/7xxx or later, and Nvidia "Kepler"/6xx/some quadros. Nvidia Fermi is still a possibility, but those were never native to macOS; hence the wait for the Web Driver update that will make or break compatiiblity. ATM, you can't even use Maxwell or Pascal (9xx and later) due to a lack of Web Drivers, unless you strip out the macOS version check with nvidia-update. Only Kepler for nvidia works out of the box.

But that is impossible due to most recent GPU I can install into iMac 11,3 is HD 6xxx which is unsupported too (((

 

[Project Alice]

Theoretically, if you boot it on real hardware, it should kernel panic (due to the lack of SSE4.2 instructions in pre-Core "i" architectures...) I am surprised that it works on a VM, what is the VM software you are using?

Parallels 13. I have VMware 8 as well, but I have found parallels 13 as its way faster, which is odd. Anyways, It's almost finished. Since this VM isn't really emulation, it should run anything the host is capable of running and vice versa.

 

[toru173]

Hi all, I'm working on a side project - avoiding the forced conversion to APFS. It looks like the installer sets a flag in
minstallconfig.xml when staging the install at /macOS\ Installation\ Data. I have successfully patched BaseSystem.dmg with the startosinstall file from High Sierra, but it then fails the integrity check against BaseSystem.chunklist.

How do I re-generate the chunklist? I've not been able to find anything in my research, though it's not been terribly thorough

 

[iphone2g&3gfan]

Parallels 13. I have VMware 8 as well, but I have found parallels 13 as its way faster, which is odd. Anyways, It's almost finished. Since this VM isn't really emulation, it should run anything the host is capable of running and vice versa.

That's probably due to Parallels leveraging Hypervisor.framework, allowing it to run like KVM on a linux machine. Well, I think given that it seems to work, go ahead and copy yourself a USB and try booting directly. Apparently SSE4.2 works for you... this is worth investigating, as current understanding is that a Core "i" architecture CPU is required (Nehalem or later). I've gotta rest, but I'll be back tomorrow. Let us know how it goes!
[doublepost=1528259437][/doublepost]

Hi all, I'm working on a side project - avoiding the forced conversion to APFS. It looks like the installer sets a flag in
minstallconfig.xml when staging the install at /macOS\ Installation\ Data. I have successfully patched BaseSystem.dmg with the startosinstall file from High Sierra, but it then fails the integrity check against BaseSystem.chunklist.

How do I re-generate the chunklist? I've not been able to find anything in my research, though it's not been terribly thorough

Try to run the chunklist through 'file' in terminal.
$ file basesystem.chunklist

If you have homebrew, install binwalk with "brew install binwalk" and run that on the chunklist. If you could post a copy of the chunklist somewhere I could try and reverse it, because I don't have a macOS install anymore - I'm 100% linux now.

 

[PowerMac G4 MDD]

No graphics acceleration = Not usable. If that's the case, it's not worth it.

Do you think that, once 5,1 support comes along, I can at least use my GT 120 to just install Mojave? For troubleshooting purposes (and firmware updates / installations), I keep such a card on the side. My main card is not flashed for Mac.

In other words, does no acceleration mean physically unusable—or does it merely mean miserable to use?

 

[MP39]

Parallels 13. I have VMware 8 as well, but I have found parallels 13 as its way faster, which is odd. Anyways, It's almost finished. Since this VM isn't really emulation, it should run anything the host is capable of running and vice versa.

I'm trying the same thing now too. So far so good. It is installing. Is there a specific way to clone the drive to a real drive so then I can natively boot now that the installer check has been bypassed via Parallels?

 

[iphone2g&3gfan]

Do you think that, once 5,1 support comes along, I can at least use my GT 120 to just install Mojave? For troubleshooting purposes (and firmware updates / installations), I keep such a card on the side. My main card is not flashed for Mac.

In other words, does no acceleration mean physically unusable—or does it merely mean miserable to use?

Just miserable to use. You will always - no matter if the GPU is at all supported with any drivers - get a 1024x768 display with absolutely any graphics card in macOS. On a real mac, EFI handles the selection of the video mode, so you will get full resolution in the macOS ui, but you will not have metal, which means the software renderer processes everything. Apps like Final Cut Pro will crash, and apps like Chrome/Chromium will severely glitch, but it will still boot (those apps require Metal/OpenGL).

EFI boot screens are important, and the GT 120 will still work as always. Just the no acceleration in macOS part, but it will still output a full resolution display.

 

[Zwhaler]

Unlike the last several releases of MacOS... I would actually like Mojave on my 5,1. My R9 280X took a crap recently (either that or the external PSU is dying; I haven't taken the time to thoroughly investigate) so I'm back on the HD5870. If my GPU needs replacing, I would definitely be interested in getting Mojave and keeping this system for years to come.

 

[iphone2g&3gfan]

I'm trying the same thing now too. So far so good. It is installing. Is there a specific way to clone the drive to a real drive so then I can natively boot now that the installer check has been bypassed via Parallels?

If you use a VMDK disk, or can, you can use https://github.com/libyal/libvmdk/wiki/Mounting. This will set up a /dev/something device, which means you can then use 'dd' to copy it to a real disc.

1) Turn off VM that uses VMDK.
2) Mount VMDK.
3) Figure out what /dev/diskX VMDK is
4) Figure out what target /dev/diskX real disk is (and diskXs1, diskXs2 for partitions 1 and 2, and so on)
5) dd if=/dev/diskVMDK of=/dev/diskReal bs=1m

I know, complicated stuff but doable

 

[PowerMac G4 MDD]

Just miserable to use. You will always - no matter if the GPU is at all supported with any drivers - get a 1024x768 display with absolutely any graphics card in macOS. On a real mac, EFI handles the selection of the video mode, so you will get full resolution in the macOS ui, but you will not have metal, which means the software renderer processes everything. Apps like Final Cut Pro will crash, and apps like Chrome/Chromium will severely glitch, but it will still boot (those apps require Metal/OpenGL).

EFI boot screens are important, and the GT 120 will still work as always. Just the no acceleration in macOS part, but it will still output a full resolution display.

Thanks for your response—that's what I was thinking. I just need my GT 120 for the Mojave installation, so all is good. Initially, I was worried that I'd have to do something along the lines of a TDM installation or a clone-over.

 

[iphone2g&3gfan]

I gotta go now, expect me back tomorrow evening like the past few days...

 

[toru173]

If you use a VMDK disk, or can, you can use https://github.com/libyal/libvmdk/wiki/Mounting. This will set up a /dev/something device, which means you can then use 'dd' to copy it to a real disc.

1) Turn off VM that uses VMDK.
2) Mount VMDK.
3) Figure out what /dev/diskX VMDK is
4) Figure out what target /dev/diskX real disk is (and diskXs1, diskXs2 for partitions 1 and 2, and so on)
5) dd if=/dev/diskVMDK of=/dev/diskReal bs=1m

I know, complicated stuff but doable

I use Paragon's free VMDK mounter (https://www.paragon-software.com/home/vd-mounter-mac-free/), and then disk utility