No compat check is only needed when smbios is from real unsupported Mac.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
macOS 10.14 Mojave on Unsupported Macs Thread
- Thread starter dosdude1
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
Ah, so this time XNU is doing the compat check? It used to be boot.efi that would show the
"This version of macOS is not supported on this computer!
Reason: Mac-F422..."
Well, now we know.
[doublepost=1528685879][/doublepost]
Makes sense, I get it now. You've done an amazing job, thank you a lot, truly.
[doublepost=1528686122][/doublepost]
Oh, so Fermi loads GeforceGLDriver.bundle? I mean, tesla loaded GeforceTeslaGLDriver.bundle, and stock macs used tesla and kepler. Interesting, it might be that Apple (or nvda) compiled in support for Fermi in GeforceGLDriver.bundle by accident, supporting fermi in addition to the kepler used in stock macs.
We may not be able to rely on Apple's continued support of Fermi in GeforceGLDriver... it's just shocking that they would include support for a GPU arch that's not stock... Does Fermi currently have metal support? I don't think so, b/c Fermi and Kepler are completely different, completely. I think that GeforceGLDriver rather contains code for both archs. I don't think fermi loads the kepler code, but rather the stock kepler GLDriver included code for Fermi separately.
If fermi has metal, that's great. But does it? Or just OpenGL and macOS renders with OpenGL?
P.S please excuse the long, complicated questions, yes I know they probably take a bunch of time to answer. It's a lot to understand... and sorry for making you read all of that if you did!
Last edited:
1) it's always been boot.efi doing that check. It just reads the area of XNU boot args.
2) no Fermi does not have metal.
3)
[doublepost=1528687268][/doublepost]
I assume you mean fermi, fermi, Kepler? lol
Crap, oops, big time... thanks for catching that, fixed! lol...
So the fermi code is the loophole for us, to (1) present the OpenGL-only kexts and bundles to Mojave and (2) have WindowServer invoke/hook our GLDriver and (3) render the UI using the OpenGL code that it uses in Fermi? (it makes me wonder why apple/nvda would supply code for fermi still, I think it's a mistake)
If that's the trick/loophole, then instantly HD Arrandale, HD 3000, all those open up! Just gotta binpatch WindowServer now!
Thanks for the answers again, they clear everything up! You're the best!
Last edited:
I have Core2 Duo E8400 (SSE4.1 only, no SSE4.2) + GT 640 1G, after installation success, boot to system restarting on gray screen and mouse. (GPT + APFS, Clover 4522. Works fine with 10.13.6)
May I ask for your /EFI/CLOVER/config.plist and FakeSMC.kext? Thanks.
Dear Dosdude1 or Parrotgeek,
I have a Macbook Air mid 2011 that is currently on High Sierra 10.13.5. I wish to use this to install Mojave Beta build 1. At first I installed Mojave on a VMware Fusion vm machine and it seems to be working (but slow). Performance might be an issue due to the memory allocation (2GB only) I have setup. So I decided to install it on my MBair (instead of a VM).
My MBAir mid 2011 is not under the compatibility list for Mojave. But based on your (dosdude1's quote on the wiki thread), I will modify the kext file mentioned on the quote. Is this kext flle inside the Mojave install file or the high sierra patcher file?
Thanks (Mahalo) as always for your help and insights)
Danno
Let me expand/clarify on my questions above:
Can I just run the Mohave install produced by the macOSDownload app from dosdude1?
Mahalo
Last edited:
I would also ask, I have the same problem, CPU E8400 sse4.1, GT630 Fermi, high sierra works great and Mojave reboots like a colleague, on a gray background in the final phase of the installation, I tried everything, whether it's the fault of the CPU or gpu?
Yes, that is the general idea, although we aren't really using any Fermi code directly. We're just trying to figure out what makes it "happy" with Fermi and not other GL-only GPUs.
As I got the boot stick showing a GUI on an AMD HD5870 on a MP3.1 it could be something in boot.efi ?
Try the boot.efi from the bootstick or at least compare them.
Maybe its the boot-arg -no_compat_check I needed...
Try the boot.efi from the bootstick or at least compare them.
Maybe its the boot-arg -no_compat_check I needed...
it shows a gui but 3d/gaming does not work, or videos
booted the mac pro 3,1 with the stick, -no_compat_check flag and both tesla/8800gt and kepler/gt710, each connected to their own screen. formatted an ssd apfs and got this message: "your mac needs a firmware update in order to install to this volume. please select a mac os extended (journaled) volume instead". formatted the ssd hfs+j and then got this message: "this version of mac os cannot be installed on this computer".
Well bugger me... Looks like we've all come a long way since yesterday jesus. If I had a mac on me right now I'd be testing that theory with no_compat_checks arg, but I'm at work lol. Keep up the good work.
Why Fermi ? (Quadro 4000 mac) Why GT120 ? why partial OpenGL see my Thread :
https://forums.macrumors.com/threads/mac-pro-passthrough-efi-boot-output-fermi-etc-4mojave.2122760/
Please comment my post about Mac Pro 2010,2012 supported Metal GPUs
Last edited:
Right, and all that would work fantastic under Linux (I have done exactly that, actually with an X58 + xeon + linux 4.17 + 2 gpus) but the macOS kernel (Darwin/XNU) does not support PCI passthrough. We could patch the sources of XNU when released (Darwin Kernel version 18.0.0: .......) but XNU sources are only released around GM, and we'd also be missing XCPM (not open source) and a few other GPU libs that ARE in XNU but NOT in the open source tarball releases. I compiled my own kernel for Mavericks, and it does not have any XCPM, so we'd have to load AppleIntelCPUPowerManagement.kext instead (not optimal) XCPM=Xnu Cpu Power Management.
It'd be great if it were that easy, and trust me, we all want it to be, but Apple hasn't implemented that in Darwin...
[doublepost=1528730180][/doublepost]
Here's my hunch: for the same reason WindowServer wouldn't load my AppleIntelBDWGraphicsGLDriver.bundle in Mavericks in my backport of a broadwell macbook air: it probably has the driver names hardcoded. Just run a grep -a GLDriver WindowServer :
$ grep -a GLDriver WindowServer
-a: allow printing binary files to stdout
GLDriver: string to search for (add -i for case-insensitive)
WindowServer: File to search string in
You can probably run it recursively (-r) in the entire /system directory to find out where it might be hardcoded in files. Just run grep on * instead:
$ cd /System
$ grep -ra GLDriver *
(although you probably already knew this, no doubt, some others might not, lol)
Last edited:
Hope the community could find the way to use all loaded kexts for Intel and AMD to enable hw acceleration!
I mean, my MBP is old, sure; but it's not that bad even by modern standards.
I mean, my MBP is old, sure; but it's not that bad even by modern standards.
Has anyone tried this on a MacBook Pro (12", mid 2010) base-model (Core 2 Duo, 4GB Ram, NVIDIA 320/M)
Interesting. Your config.plist does not contain -x , but it contains a long boot flag I do not recognise.
Same problem.
Still no lucky for me, after installation success, boot to system restarting on gray screen and mouse cursor.
Thanks anyway!
Yes that is what I mean (ignore model; It's a hackintosh)
I know this is slightly off topic but could you inform me of the parts and specs? I'm building a hackintosh and I want it to be able to support Mojave
or if I tell you the specs I wish to use, could you tell me if they would be okay?
These are:
AMD Ryzen 5 1400 - Processor
Ballistix Sport LT 8GB - RAM
NVMe Samsung Evo 960 SSD (250GB) - Storage
Gigabyte GA-AX370-Gaming 5 - Motherboard
BitFenix Whisper M 650W - Power Supply
EVGA GeForce GTX 1050 SSC - Graphics
Gigabyte - GC-WB867D-I - Wi-Fi
I'm 18, earning £200 a week so it's going to be hard but does that sound good enough?
Last edited: