macOS 10.15 Catalina on Unsupported Macs

[joevt]

I have a Mac Pro 3,1 (2008) patched with Dosdude's patcher and running Catalina 10.15.7. My CPU is a NVIDIA GeoForce GTX 680. I installed recently Adobe Photoshop 2021 to replace version 2020 who was perfectly running on the MacPro. Install worked but when I start the program its crashing and the CrashReport says that SSE 4.2 is required to run the 2021 version and my CPU can't run it. Or in the patcher and I installed it there is included a MouSSE extension who is supposed to emulate SSE 4.2 !!!!! Has somebody an idea how to make Photoshop 2021 recognize this emulator ????? Thanks in advance. On the Adobe's forums other people even on Windows have this problem but the reply from Adobe is : Change your CPU.............we will not change our specs.

SSE 4.2 does not emulate the entire SSE 4.2. The emulator does not make the CPU advertise SSE 4.2 support. The emulator only works for code that does not check for SSE 4.2 support but instead just assumes that the support exists and uses a couple of the instructions. The emulator does not emulate all the SSE 4.2 instructions just those couple of instructions.

 

[Blackriton]

SSE 4.2 does not emulate the entire SSE 4.2. The emulator does not make the CPU advertise SSE 4.2 support. The emulator only works for code that does not check for SSE 4.2 support but instead just assumes that the support exists and uses a couple of the instructions. The emulator does not emulate all the SSE 4.2 instructions just those couple of instructions.

Thanks for these infos. Which CPU other than the NVIDIA GTX 680 would support SSE 4.2 and be compatible with a 2008 Mac Pro 3,1 or is there a way to workaround the Adobe Photoshop 2021 specs in the installer ???

 

[0403979]

Thanks for these infos. Which CPU other than the NVIDIA GTX 680 would support SSE 4.2 and be compatible with a 2008 Mac Pro 3,1 or is there a way to workaround the Adobe Photoshop 2021 specs in the installer ???

I’m not sure about the CPUs but depending on how the installer works, you can probably workaround the requirements. I doubt it would actually work after installation though.

 

[Larsvonhier]

I’m not sure about the CPUs but depending on how the installer works, you can probably workaround the requirements. I doubt it would actually work after installation though.

The original question is somehow mangled: It‘s not the case that a GPU has to (or can) support SSE instructions! The other way round: Some GPUs need drivers that contain certain SSE instructions (i.e. SSE4.2).
Your machine‘s CPU has to support them!

 

[trifero]

Thanks Trifero
- Boot pressing alt --> ok i see my efi boot from my USB key created by Dosdude1
- Check the "Disable APFS booter" when making the USB --> this, i am not understand ? where ? when at the creation of bootable key ?

last question : what is exactly the process of "APFS Rom Patcher". May i see a progression bar or something elses ? because me only what i explain in message up #12,621

 

[Blackriton]

I’m not sure about the CPUs but depending on how the installer works, you can probably workaround the requirements. I doubt it would actually work after installation though.

The requirements are NOT in the installer are install is finishing without any fuss. Its on start that the app is crashing with a crash log saying SSE 4.2 is missing.............and in the app requirement its mentioned also that SSE 4.2 is mandatory which was not the case in the last version.

 

[Syncretic]

The requirements are NOT in the installer are install is finishing without any fuss. Its on start that the app is crashing with a crash log saying SSE 4.2 is missing.............and in the app requirement its mentioned also that SSE 4.2 is mandatory which was not the case in the last version.

MouSSE comes with an auxiliary command-line program called moussestats. Depending on how MouSSE was installed, it might already be in your path, or you might need to find it on the installation media you used. In any case, please run moussestats from a shell prompt (does not need to be sudo), and post the results. Output will look something like this:

Sun Nov  1 10:54:32 2020 EST (GMT-5) :: System Up 16d 7h 21m 05s
Darwin Kernel Version 19.6.0: Mon Aug 31 22:12:52 PDT 2020;
root:xnu-6153.141.2~1/RELEASE_X86_64 (Catalina) !suid
MouSSE 0.95 appears to be loaded and functioning correctly.
Instructions parsed since last kext load:
   popcnt:                   6987
   pcmpgtq:                   423
   crc32:                      15
   Unknown:
     VEX (AVX/AVX2):          871
     EVEX (AVX-512):            0
     pcmp?str?:                 3
     Other:                   192

Also, as @joevt correctly points out, MouSSE does not "advertise" SSE4.2, meaning it does not hook the CPUID instruction or alter the MacOS flags indicating SSE4.2 support. Programs are expected to check for SSE4.2 support before using SSE4.2 instructions, but a surprising number (including the MacOS AMD driver) do not - they simply use the instructions and cause a fault or panic when there's a problem. If your Adobe software is testing for SSE4.2 as it should, MouSSE won't help you. However, if they are instead trapping faulty instructions and letting you know there's a failure, there's at least a chance MouSSE will be of some use.

Please post the moussestats output so we can catch a glimpse of what's going on "under the hood."

 

[Blackriton]

MouSSE comes with an auxiliary command-line program called moussestats. Depending on how MouSSE was installed, it might already be in your path, or you might need to find it on the installation media you used. In any case, please run moussestats from a shell prompt (does not need to be sudo), and post the results. Output will look something like this:

Sun Nov  1 10:54:32 2020 EST (GMT-5) :: System Up 16d 7h 21m 05s
Darwin Kernel Version 19.6.0: Mon Aug 31 22:12:52 PDT 2020;
root:xnu-6153.141.2~1/RELEASE_X86_64 (Catalina) !suid
MouSSE 0.95 appears to be loaded and functioning correctly.
Instructions parsed since last kext load:
   popcnt:                   6987
   pcmpgtq:                   423
   crc32:                      15
   Unknown:
     VEX (AVX/AVX2):          871
     EVEX (AVX-512):            0
     pcmp?str?:                 3
     Other:                   192

Also, as @joevt correctly points out, MouSSE does not "advertise" SSE4.2, meaning it does not hook the CPUID instruction or alter the MacOS flags indicating SSE4.2 support. Programs are expected to check for SSE4.2 support before using SSE4.2 instructions, but a surprising number (including the MacOS AMD driver) do not - they simply use the instructions and cause a fault or panic when there's a problem. If your Adobe software is testing for SSE4.2 as it should, MouSSE won't help you. However, if they are instead trapping faulty instructions and letting you know there's a failure, there's at least a chance MouSSE will be of some use.

Please post the moussestats output so we can catch a glimpse of what's going on "under the hood."

I just run moussestats in Terminal and that's the result :

 

[philk34]

View attachment 977195

Thank You for your help
i did it finally… works !
i made a new bootable key with an old hard drive disk and finally works and finalize installation
Now i have trouble with the display port of my principale screen, i've try with DVI-Dcable and work fine
There is a bug with Display port ?

 

[Syncretic]

I just run moussestats in Terminal and that's the result :
View attachment 977254

That output suggests that the Adobe software is checking for the existence of SSE4.2 before trying to execute any instructions, which is the correct thing to do. Unfortunately, the current version of MouSSE won't help you here. A future version will likely "advertise" SSE4.2 support, but that version won't be available for some time (weeks/months/maybe never).

It sucks that our old machines are slowly becoming obsolete, but at least it's for a valid technical reason, and not Apple blacklisting old computers just to encourage the purchase of new computers. Broader use of AVX will make the MP4,1/5,1 troublesome, and broader use of AVX2/AVX512 will cause MP6,1 users heartburn. It's just a matter of time.

 

[Blackriton]

That output suggests that the Adobe software is checking for the existence of SSE4.2 before trying to execute any instructions, which is the correct thing to do. Unfortunately, the current version of MouSSE won't help you here. A future version will likely "advertise" SSE4.2 support, but that version won't be available for some time (weeks/months/maybe never).

It sucks that our old machines are slowly becoming obsolete, but at least it's for a valid technical reason, and not Apple blacklisting old computers just to encourage the purchase of new computers. Broader use of AVX will make the MP4,1/5,1 troublesome, and broader use of AVX2/AVX512 will cause MP6,1 users heartburn. It's just a matter of time.

Thanks for your expertise Syncretic. It looks we will have to think more rapidly than expected to get new computers to run the most up-to-date software and systems like Big Sur.........!!!!!!!!

 

[trifero]

Thank You for your help
i did it finally… works !
i made a new bootable key with an old hard drive disk and finally works and finalize installation
Now i have trouble with the display port of my principale screen, i've try with DVI-Dcable and work fine
There is a bug with Display port ?

I´m glad it worked. With certain cards, GTX 680, for example, the display port does not work. It worked in Mojave, but not in Catalina. I´m using now Atlona AT-DP400 for my Led Cinema Display.

Mojave to Catalina Displayport Not Working

Hi All, Seems to be a lot of people having issues with DisplayPort not working after upgrading to Catalina, but HDMI still does - and I'm one of them. Everything else works - just not DisplayPort Have been struggling to find answers on any other threads as to how to fix, and I can't work out...

www.tonymacx86.com

 

[Dayo]

Thanks for your expertise Syncretic. It looks we will have to think more rapidly than expected to get new computers to run the most up-to-date software and systems like Big Sur.........!!!!!!!!

That sucks but I suppose it was inevitable.

One question though, is this affecting anyone using OpenCore?

 

[Peeboo]

Macbook Pro, 2011. Core I7. I upgraded to Catalina 10.15.1 last year with the dosdude1 patcher. It will not "update" to later versions of 10.15 . Have tried to use the latest version of the dosdude1 patcher to install 10.15.7 from a USB stick and/or the hard drive(SSD) No joy. Both attempts get stuck in the Apple boot screen. USB method progresses to the end of the progress bar but never completes. Waited many hours. The hard drive(SSD) has been APFS since the 10.15.1 install/patch process. Is there a log file "somewhere" that might indicate where/why it is getting "stuck".

Another day, another thought? I have been reading about Opencore loader. Impression... Opencore may provide the ability to run and update Catalina. Does that mean I should give up on the dosdude1 patcher and invest the time to try opencore? At this point in time, opencore "configuration" looks complex and I am concerned about ending up with a Macbook brick.

 

[joevt]

Thank You for your help
i did it finally… works !
i made a new bootable key with an old hard drive disk and finally works and finalize installation
Now i have trouble with the display port of my principale screen, i've try with DVI-Dcable and work fine
There is a bug with Display port ?

I´m glad it worked. With certain cards, GTX 680, for example, the display port does not work. It worked in Mojave, but not in Catalina. I´m using now Atlona AT-DP400 for my Led Cinema Display.

Mojave to Catalina Displayport Not Working

Hi All, Seems to be a lot of people having issues with DisplayPort not working after upgrading to Catalina, but HDMI still does - and I'm one of them. Everything else works - just not DisplayPort Have been struggling to find answers on any other threads as to how to fix, and I can't work out...

www.tonymacx86.com

MacPro3,1 with EVGA Nvidia GTX 680 Mac Edition testing notes at #546 (about DisplayPort and other ports, Atlona, macOS versions, boot screens)

 

[Generic_bug]

Hi all,

I try to do a clean install of macOS Catalina 10.15.7 on a MacBook 7,1 but every time I get a message saying "Installation of macOS could not continue. Installation requires downloading important content. That content can't be downloaded at this time. Try again later." I have already tried to re-download Catalina from Dosdude AND Apple and I get the same error both times. At first I was getting the message that the installer is damaged, but this got solved by removing installinfo.plist from SharedSupport. I have also tried to modify the DATE on the installer via terminal and I still get the same error. The Time command from terminal always worked... Now even this doesnt... Can anyone help please ? I really need to install Catalina. Thanks !

 

[sipony]

guys I must open my Imac 9.1 early 2009 , to be sure of apply the correct APFS firmware ? Did someone apply succesfully the catalin patcher on this model?

 

[trifero]

MacPro3,1 with EVGA Nvidia GTX 680 Mac Edition testing notes at #546 (about DisplayPort and other ports, Atlona, macOS versions, boot screens)

I missed that. Great work!

 

[Dayo]

SSE 4.2 does not emulate the entire SSE 4.2. The emulator does not make the CPU advertise SSE 4.2 support. The emulator only works for code that does not check for SSE 4.2 support but instead just assumes that the support exists and uses a couple of the instructions. The emulator does not emulate all the SSE 4.2 instructions just those couple of instructions.

Do you know which instructions it emulates?

It is possible to spoof SSE4.2 support using OpenCore so that the software query about support will be positive.
If the other instructions are uniquely specified, it might be possible to spoof them also.

Presumably the software would then issue the commands and MouSSE can kick in.

 

[Syncretic]

Do you know which instructions it emulates?

It is possible to spoof SSE4.2 support using OpenCore so that the software query about support will be positive.
If the other instructions are uniquely specified, it might be possible to spoof them also.

Presumably the software would then issue the commands and MouSSE can kick in.

From the MouSSE ReadMe:

*** What does it do?
MouSSE traps illegal instructions in both privileged (kernel) mode and unprivileged (user) mode, and emulates the POPCNT, PCMPGTQ, and CRC32 instructions. As of this writing, PCMPGTQ and POPCNT are the only problematic instructions used in the AMD drivers. (And, apparently, the only two currently used in World of Warcraft.)

*** What does it NOT do?
[snip]
At this point, only the SSE4.2 PCMP?STR? instructions remain un-emulated.

 

[Dayo]

From the MouSSE ReadMe:
*** What does it do?
MouSSE traps illegal instructions in both privileged (kernel) mode and unprivileged (user) mode, and emulates the POPCNT, PCMPGTQ, and CRC32 instructions. As of this writing, PCMPGTQ and POPCNT are the only problematic instructions used in the AMD drivers. (And, apparently, the only two currently used in World of Warcraft.)

Apologies for not remembering to check the readme first.

Using Intel's MacCPUID, POPCNT appears to be a discrete CPU feature but the others do not appear so (presumably covered under SSE4.2).

 

[Syncretic]

Apologies for not remembering to check the readme first.

Using Intel's MacCPUID, POPCNT appears to be a discrete CPU feature but the others do not appear so (presumably covered under SSE4.2).

Correct. POPCNT is a bit of an oddity; it, along with LZCNT, was released with SSE4a as created by AMD, but they have their own separate CPUID bits (rather, POPCNT has its own, and LZCNT is part of ABM), and technically aren't part of SSE (they don't work on SIMD registers), but most sources consider POPCNT to be part of SSE4.2 anyway.

 

[joevt]

Do you know which instructions it emulates?

It is possible to spoof SSE4.2 support using OpenCore so that the software query about support will be positive.
If the other instructions are uniquely specified, it might be possible to spoof them also.

Presumably the software would then issue the commands and MouSSE can kick in.

Spoofing SSE4.2 support is probably not a good idea - MouSSE doesn't do it because it doesn't do all of SSE4.2, plus if the macOS sees that bit, then it's going to want to use SSE4.2 for everything when it would be preferable to use the non-SSE4.2 code where available.

But for the Photoshop problem, spoofing SSE4.2 support, then using the MouSSE logging feature might be informative.

 

[Dayo]

if the macOS sees that bit, then it's going to want to use SSE4.2 for everything

That's definitely a potential issue in theory which may crop up later.

Not yet manifesting as an actual issue so far and if Mac OS had been looking closely at SSE4.2, the "Photoshop Problem" would have happened earlier at the OS Level and MouSSE would probably have not worked in the first place.

Still definitely something that may happen at some point later but this kicks the can down the road on the 3,1.

 

[rmb3218]

Hi all,

this is my first post here, I apologise as I have searched but can’t find the answer. I have just installed the patcher on my 2011, early, 13”, MacBook Pro, 16gb memory. The installation went without any issues. I had a codesign error on Coconut Battery, but used Kubadownload instructions to fix. The problem I have is that I have a codesign error with Parallels. I have done a search on this thread and a patch has been mentioned to fix this. Can someone point me to it?

thanks, Kevin