macOS 10.14 Mojave on Unsupported Macs Thread

[MichaelDT]

Potential Fix for UI Glitches:
If the transparency related UI glitches are caused by a lack of correct drivers to render it then perhaps it would to possible to combine the drivers from Mojave and High Sierra in order to get proper rendering. Just an idea.

No. It’s glitching because they’re calling a Metal API which is calling an instruction on the GPU that doesn’t exist even with the correct driver, hence the cut off. It is likely possible to simulate the instruction but replace all of its calls with the older equiv but good luck without the source code, or it may be that it is being simulated and it’s just not posssible to do so correctly.

 

[0403979]

No. It’s glitching because they’re calling a Metal API which is calling an instruction on the GPU that doesn’t exist even with the correct driver, hence the cut off. It is likely possible to simulate the instruction but replace all of its calls with the older equiv but good luck without the source code, or it may be that it is being simulated and it’s just not posssible to do so correctly.

So in short this issue is almost impossible to fix?

 

[jackluke]

I just tried this, and I don't see any difference really. Still the graphical glitches on corners of windows, horrible black blocks in the Dashboard, etc.

What's others' experience with this? @jackluke, what exactly did you notice being better?
[doublepost=1531377017][/doublepost]Super small and not particularly relevant thing, but anybody else notice open -R /path/to/folder (reveal in Finder) not working in Mojave?

The IOAccelerator unix exec I uploaded is from the first absolute Mojave beta, I often keep copies of first seeded betas not only of MacOS.
I don't know I feel the entire system a bit faster, and consider that this unix exe from beta 1 is the smallest compiled by apple as far I remember, I do like forked code doing the same function.

You mean bad black blocks on Dashboard's Weather bar, yes, there are.
About Finder's windows artifacts didn't notice maybe because I use dark mode appearance.

I am still on Mojave beta 2 (18A314h) and my isight camera works OOB, just with your IOUSBHostFamily.kext from HS.

Maybe since Mojave beta 3 some frameworks have been updated so that's why some new patches could be needed.

While Finder's Go to Folder to me is working perfectly (since I replaced IOAccelerator from Mojave beta 1)

No root folder called com.apple.AdLib into my /Volumes/Mojaveb2/ just all the typical temporary and hidden folders.


Advise for everyone with a Penryn C2D Mac: do not forget that after any macOS upgrade you have to replace again IOUSB, telemetry plugin, GPU family kexts and so on and maybe something else new.


edit:

I have read on posts before that many of you have saturated memory usage in Activity monitor.
Here are my 8 GB DDR3 1067 Mhz Macbook7,1 memory usage (with Safari opened and Dictation enabled)

Physical Memory: 8,00 GB
Memory Used: 3,96 GB (after 3 hours of Mojave use this value increased to 4,45 GB)
Cached Files: 2,23 GB
Swap used: 0

App Memory: 2,26 GB
Wired Memory: 1,77 GB
Compressed: 0

Sleepimage: 4,29 GB

All these values just increase a bit every half hour.

I do have two different RAM sticks: a crucial 4 gb 1066 Mhz and a samsung 4 gb 1600 Mhz, of course they work at lower frequency, never had an issue since years.

[doublepost=1531383023][/doublepost]

No. It’s glitching because they’re calling a Metal API which is calling an instruction on the GPU that doesn’t exist even with the correct driver, hence the cut off. It is likely possible to simulate the instruction but replace all of its calls with the older equiv but good luck without the source code, or it may be that it is being simulated and it’s just not posssible to do so correctly.

Your explanation does not make a fold, anyway I report that on dark mode appearance every transparency seems work fine, with very minimal glitches, instead to those who prefer light mode appearance simply System Pref / Accessibility / Display / reduce transparency.

 

[0403979]

The IOAccelerator unix exec I uploaded is from the first absolute Mojave beta 1, I often keep copies of first seeded betas not only MacOS. I don't know I feel the entire system a bit faster, and consider that this unix exe from beta 1 is the smallest compiled by apple as far I remember, I do like forked code doing the same function.

You mean horrible black blocks on Dashboard's Weather bar, yes they are.

I am still on Mojave beta 2 (18A314h) and my isight camera works OOB, just with IOUSBHostFamily.kext from HS.

And Finder's Go to Folder to me is working perfectly (much better since I replaced IOAccelerator from Mojave b1)

Maybe from Mojave beta 3 some frameworks have been updated so that's why some new patches could be needed.

No root folder called com.apple.AdLib into my /Volumes/Mojaveb2/ just all the typical temporary hidden folders.

Advise for everyone with a Penryn C2D Mac: do not forget that after any macOS upgrade you have to replace again IOUSB, telemetry plugin, GPU family kexts and so on and maybe something else new.
[doublepost=1531383023][/doublepost]

Your explanation does not make a fold, anyway I report that on dark mode appearance every transparency seems work fine, with very minimal glitches, instead to those who prefer light mode appearance simply System Pref / Accessibility / Display / reduce transparency.

It’s not a complete solution. It’s just a workaround.

 

[indigotheory]

Massive thanks for getting this onto older Macs. I installed it last night on my Late 2008 Macbook. I installed all the patches bar the WiFi patch as it was unchecked by default

Legacy drivers work fine, I dont see any graphical glitches in dark mode. Light mode just seems a bit of a darker grey in the left of finder windows and the menu bar.

Also is there a patch updater? Do we even need one?

 

[ASentientBot]

Hello everybody! Sorry to potentially bring bad news, but, using Mojave as my primary system on MacBook7,1 for a day I've definitely started to see a problem. Basically, it runs perfectly for a while (30 minutes - a couple hours) and then suddenly gets insanely slow, to the point where it's not always possible to even shutdown properly.

I think I've found the explanation: in Console, right as this happens, there are messages like
unable to create a basic OpenGL renderer
CoreImage switching to software rendering mode
(Not the exact messages - writing them from memory as I haven't gotten a screenshot -- the sudden lag is that bad.)

So it seems like my sketchy clientClose() patch is coming back to bite me in the form of OpenGL failures after long term use.

Anybody else experiencing this?

In any case I'm going to bed now but I'll check in with this tomorrow. Fellow nVidia Tesla guys, our work might not be quite done yet...

 

[jackluke]

Hello everybody! Sorry to potentially bring bad news, but, using Mojave as my primary system on MacBook7,1 for a day I've definitely started to see a problem. Basically, it runs perfectly for a while (30 minutes - a couple hours) and then suddenly gets insanely slow, to the point where it's not always possible to even shutdown properly.

I think I've found the explanation: in Console, right as this happens, there are messages like
unable to create a basic OpenGL renderer
CoreImage switching to software rendering mode
(Not the exact messages - writing them from memory as I haven't gotten a screenshot -- the sudden lag is that bad.)

So it seems like my sketchy clientClose() patch is coming back to bite me in the form of OpenGL failures after long term use.

Anybody else experiencing this?

In any case I'm going to bed now but I'll check in with this tomorrow. Fellow nVidia Tesla guys, our work might not be quite done yet...

Thanks for the flash info, honestly I still use mainly High Sierra, have used Mojave just few hours per day, and I would say that I noticed a similar slowing (not insanely slow) when I keep many Safari panels opened especially the youtube ones (sometimes have to force closing in activity monitor).

But I recall I am still on Mojave beta 2.
So I would think that previous patching schemes are slight better stable than Mojave beta 3.

I wouldn't focus too much on Console monitoring, I often even on supported High Sierra see many many errors messages while the system in better or worse is working very good.

Regard clientClose() function, maybe you could try to cut less binary code from your GeforceTesla unix exec, maybe is a trunkated function that you left hanging, even if this time I guess could be also to Metal API missing related.

Using OpenGL.framework from HS doesn't work with your GeforceTesla unix patched on my Mojave beta 2, but maybe on beta 3, who knows.

Anyway, I am writing from Safari's Mojave beta 2 (with IOAccelerator from beta 1) on MB7,1 since a couple of hours and still working fine.


edit:

Another advice for everyone: when I had a good working beta system on an unsupported Mac, I made almost ever a full compressed dmg backup with read/write permissions of my disk, possibly deleting first /var/vm/sleepimage


edit2:

Ok, I have opened Console and monitoring Mojave beta 2, read these messages (they occur not very often):

Unable to create basic Accelerated OpenGL renderer.
Core Image is now using the software OpenGL renderer. This will be slow.

but I repeat the system is very usable (not for high-end user), but a bit slower than High Sierra of course.

I am using IOAccelerator beta 1 that I posted some pages ago.

And I did installed Mojave beta 1 then upgraded to beta 2 from a supported mac, and now booting Mojave beta 2 from an external USB 2.0 SSD.


edit3:

After 3 hours, when I open together some apps there is a delay/freeze of about 5-10 seconds, even when close/minimize/maximize a window with the red/orange/green dots, but I would say after that delay/freeze in rendering the animation, graphically the opened process is stable and usable. There is no need to force closing the processes (except Safari if you opened many panels as I do often) just waiting that delay.

So I confirm system became slower after some hours, but not insanely slower on Mojave beta 2 with IOAccelerator beta 1 (sorry I know I have repeated thousands time it's annoying).

I will try to binary edit the GeForceTesla unix exec, hoping to understand ASAP what the clientClose() function does and possibly leave that function where is and inspecting and deleting some lines inside.


edit4:

Ok I'm in, successful decompiled GFT unix binary taken clean from HS, but do you have removed all the assembler code related to function IONVGLContextTesla::clientClose() ?

OK, I've seen your file, you NOP'd every memory addresses except the last one, but how did you modified the code inside pseudo function in return rax ?
I mean I can't modify inside the code, only NOP.


Maybe if we compare the memory addresses push, call, mov, pop, add and so on with the Kernel panic log could keep almost all memory addresses.
They are not so much I counted 21 in columns.

I saw also that the 7th memory address je points to the 16th address xor where is referral to
; CODE XREF=__ZN18IONVGLContextTesla11clientCloseEv+16
that is the exactly binary name translation of IONVGLContextTesla::clientClose() string.

I'll wait for your opinion, thanks.
Meanwhile I try to custom break some lines.

edit5:

Tried nop first six lines, and still land QE/CI to desktop, but after LoginUI got the known KP, in the KP log this time I get no kexts in backtrace but only an invalid memory address pointer.

After 2nd tried nop from the 7th to 16th, same results, same KP log.

3rd tried nop from 17th to 21st, same results but this time got the already known KP Log with the Tesla kext in backtrace, so I guess in this memory address range there isn't the KP guilty or is there!?, I am confused.

I know why I am NOP zones without criteria, but really can't nop completely a zone, I mean it remains ever a DWORD memory address residual.

My aim is to avoid NOP completely IONVGLContextTesla::clientClose() memory addresses.

Then this is mostly experimental, based on ASentientBot file, I don't know exactly I have followed the last memory address he left and this brought me to a IOGPURestartReportTesla function, then I NOP'd that memory address and recompiled binary. It seems that no more delay/freeze. I could get wrong, but maybe to those experienced, give it a try. You should chown/chmod the file and rebuild kextcache.

No, I've double checked the file, the NOP'd address is still there, so same issue delay/freeze after couple of hours, it is similar to the one posted by ASentientBot.
I think we must start again from GeForceTesla unix binary untouched from HS, I removed it.

I attach the entire untouched extracted assembly code and his relative function responsible of Mojave QE/CI KP on Legacy Nvidia Tesla.

Ok this post is becoming tedious, I will use a new one if had any news.

 

[0403979]

Thanks for the flash info, honestly I still use mainly High Sierra, have used Mojave just few hours per day, and I would say that I noticed a similar slowing (not insanely slow) when I keep many Safari panels opened especially the youtube ones (sometimes have to force closing in activity monitor).

But I recall I am still on Mojave beta 2.
So I would think that previous patching schemes are slight better stable than Mojave beta 3.

I wouldn't focus too much on Console monitoring, I often even on supported High Sierra see many many errors messages while the system in better or worse is working very good.

Regard clientClose() function, maybe you could try to cut less binary code from GeforceTesla unix exec, even if I guess could be also to Metal API missing related.

Using OpenGL.framework from HS doesn't work with your GeforceTesla unix patched on my Mojave beta 2, but maybe on beta 3, who knows.

Anyway, I am writing from Safari's Mojave beta 2 (with IOAccelerator from beta 1) on MB7,1 since a couple of hours and still working fine.


edit:

Another advice for everyone: when I had a good working beta system on an unsupported Mac, I made almost ever a full compressed dmg backup with read/write permissions of my disk, possibly deleting first /var/vm/sleepimage


edit2:

Ok, I have opened Console and monitoring Mojave beta 2, read these messages (they occur not very often):

Unable to create basic Accelerated OpenGL renderer.
Core Image is now using the software OpenGL renderer. This will be slow.

but I repeat the system is very usable (not for high-end user), but a bit slower than High Sierra of course.

I am using IOAccelerator beta 1 that I posted some pages ago.

And I did installed and now booting Mojave beta 2 from an external USB 2.0 SSD.


edit3:

After 3 hours, when I open together some apps there is a delay/freeze of about 5-10 seconds, even when minimize/maximize a window with the orange/green dot, but I would say after that delay/freeze in rendering the animation, graphically the opened process is stable and usable. There is no need to force closing the processes (except Safari if you opened many panels as I do often) just waiting that delay.

So I confirm system became slower after some hours, but not insanely slower on Mojave beta 2 with IOAccelerator beta 1 (sorry I know I have repeated thousands time it's annoying).

I’m starting to wonder if I should stay on High Sierra.

 

[netsrot39]

Hello everybody! Sorry to potentially bring bad news, but, using Mojave as my primary system on MacBook7,1 for a day I've definitely started to see a problem. Basically, it runs perfectly for a while (30 minutes - a couple hours) and then suddenly gets insanely slow, to the point where it's not always possible to even shutdown properly.

I think I've found the explanation: in Console, right as this happens, there are messages like
unable to create a basic OpenGL renderer
CoreImage switching to software rendering mode
(Not the exact messages - writing them from memory as I haven't gotten a screenshot -- the sudden lag is that bad.)

So it seems like my sketchy clientClose() patch is coming back to bite me in the form of OpenGL failures after long term use.

Anybody else experiencing this?

In any case I'm going to bed now but I'll check in with this tomorrow. Fellow nVidia Tesla guys, our work might not be quite done yet...

I've been playing around with it and you're right it becomes slower after 30 minutes or so but not to the extent that it makes it unusable (but for every app one has to wait 20 seconds or so, so for productive use I wouldn't recommend it in this state). The only thing right now to regain a faster system is a reboot, right? I'm on Beta 3.

Edit: If you leave the application open in the background it is usable but quitting it and relaunching is kinda painful. So I just will let my applications "rest" in the background but of course that is not a real solution but I want to test Mojave so badly (on an unsupported machine even though I have a MacBook Pro from 2012 haha)

 

[scrub10rd]

I keep getting install errors. I’ve got a 16gb USB I did around 1am cst (it’s 7am cst right now, I can’t really check which version of the patcher I have at the moment but I got the newest one at that time)


It’ll say 29 minutes remaining and then it’ll crash it seems due to NSException with a crash log that takes up the entire screen with only the options to let me shut down or restart. I’m trying it again but how would I fix this (assuming this next try doesn’t work)?


Edit: didn’t work, posted log below. I can’t boot into my high Sierra install either. MacBook 5,2

 

[Starplayr]

I've been playing around with it and you're right it becomes slower after 30 minutes or so but not to the extent that it makes it unusable (but for every app one has to wait 20 seconds or so, so for productive use I wouldn't recommend it in this state). The only thing right now to regain a faster system is a reboot, right? I'm on Beta 3.

Edit: If you leave the application open in the background it is usable but quitting it and relaunching is kinda painful. So I just will let my applications "rest" in the background but of course that is not a real solution but I want to test Mojave so badly (on an unsupported machine even though I have a MacBook Pro from 2012 haha)

I haven't experienced any slow down yet on my MacBook5,5. It goes to sleep fine and wakes up fine. I will check it again today.

For those getting a slow down. I wonder if it's garbage collection that's not happening?

I do know that in clamshell mode while linked to an external display, the laptop's screen is not lit up all the time, but keeps running like its the internal display is in use.

 

[schwab]

Thank you so much Dodsdude1 and colleagues for bringing Mojave 10.14 to many unsupported Macs.

I am looking forward to installing it as I did with Sierra & High Sierra on my 17" MacBookPro 4,1 (A1261-early/late 2008
https://everymac.com/systems/apple/...ore-2-duo-2.5-17-early-2008-penryn-specs.html.

For those of us who wish to contribute a "thank you donation" for your efforts to what email address/account can a PayPal payment be made?

 

[Sevenfeet]

First install issue with the 2010 Mac Mini (and this may be obvious to the programmers and regulars here)...

Do not install this over an existing High Sierra volume with HFS+. The installer crapped out midway with what looked like an error about the volume not being APFS. Doing the install this way was a mistake on my part since I had intended to wipe the hard drive clean anyway and I just forgot. But at some point in the beta process, I'd think a APFS conversion should work on the way to doing a proper High Sierra upgrade.

 

[scrub10rd]

It fixed itself by failing 3 times and then running the post install tool (I figured why not) and then a restart.

It runs nicely!

 

[Hack5190]

It fixed itself by failing 3 times and then running the post install tool (I figured why not) and then a restart.

It runs nicely!

Doesn't it feel good when you find a fix yourself - good job and enjoy.

 

[highvoltage12v]

Hello everyone,
I'm reporting on an issue I have on my 27 inch 2011 iMac with the GTX 780m MXM card installed. On build 18A326g when waking from sleep the machine will reboot with a kernel panic report related to NVDAResman and other extensions related to Nvidia and graphics. I have attached a screenshot of the panic and the log. I looked into kexts that could be related to the panic and added my iMac's board-id (942B59F58194171B) to them as we would in Sierra these kexts being "AppleGraphicsControl.kext" and "AppleGraphicsPowerManagement.kext" it didn't help with the kernel panic issue, however it improved the responsiveness of the card. I did not install Legacy Video Card Patch from the Mac OS Post Install.

 

[gugadj]

Hello everyone, I made the install with the almost normal 1b6 patch on a macbook late 2011 [macbook 8.2] after applying the patch if I checked the "legacy video patch" option the system does not load until the end when it will enter the machine restarts someone with this experience

I formatted the hd again and did not dial legacy video patch the system came through without graphics acceleration

 

[gabe12345]

Will MacOs 10.14 work on a mid 2007 Imac with a cpu upgrade without graphics acceleration? The early 2008 and mid 2007 are really similar to each other. Just wanted to ask if it will be possible

 

[dosdude1]

Will MacOs 10.14 work on a mid 2007 Imac with a cpu upgrade without graphics acceleration? The early 2008 and mid 2007 are really similar to each other. Just wanted to ask if it will be possible

Yes it would. No GPU acceleration though, like you mentioned.

 

[0403979]

There must be fix (albeit undiscovered currently) for these transparency glitches. It’s the only reason that I’m not using Mojave on my 2010 MacBook.

 

[videotape74]

I am still having the same issue on my MBP 5,4. Using the latest patcher, downloaded Mojave with it, it installs to my SSD, I apply the patches for my model, reboot. Still stuck on the Apple logo and full progress bar. Sooooo frustrating .

 

[rmeulen]

Curious to find a solution to get my iMac (27-inch, Mid 2011) with AMD Radeon HD 6970M on Mojave when released at fall this year. Therefore I add myself to this thread.

 

[Sevenfeet]

First install issue with the 2010 Mac Mini (and this may be obvious to the programmers and regulars here)...

Do not install this over an existing High Sierra volume with HFS+. The installer crapped out midway with what looked like an error about the volume not being APFS. Doing the install this way was a mistake on my part since I had intended to wipe the hard drive clean anyway and I just forgot. But at some point in the beta process, I'd think a APFS conversion should work on the way to doing a proper High Sierra upgrade.

OK, an update. I have Mojave beta running on this Mac Mini 4,1 (Nvidia 320m graphics). Performance is a bit sluggish but I'm chalking that up to an old 320 GB hard drive that was standard in this machine and only 4 GB of RAM. I have a flash drive and more RAM I can stick in this thing and I'll try to at least get the RAM done today. Graphics seems to be full speed....watching 1080p videos on Youtube now.

Otherwise, it looks like it works. The dock is showing question marks for there not being copies of iWork apps loaded so I don't know what that's all about after a clean install. Dark mode seems to work.

One issue...I don't seem to be getting any sound out of the built in speaker (but I do get sound out of the display through HDMI). Not sure if this is a patching issue or if there is something wrong with the Mac Mini itself....I'll check it out. [EDIT: Sound is working now....must have been something I did.]

 

[LegoNickD]

The new utility works well, there are a few issues of course, the expected graphics issues are here and normal text buttons seem cut off which is more noticeable in the installer. I also used the nightshift installer and while it worked I can no longer open the display settings.

 

[TimothyR734]

I am still having the same issue on my MBP 5,4. Using the latest patcher, downloaded Mojave with it, it installs to my SSD, I apply the patches for my model, reboot. Still stuck on the Apple logo and full progress bar. Sooooo frustrating .

is your SIP disabled if you recently did a nvram/pram reset control-option P and R while starting up sometimes turns the SIP back on