iM 4,1-14,4 EnableGop: Pre-OpenCore GOP support for EFI-era iMacs and Mac Pros

[monoton]

sorry for my stupidity...i have all the files in a folder named GOP and placed it on the desktop.
i open a terminal window, change to the folder which contain following files:

- EFIRom
- EnableGop.efi
- UefiRomExtract
- vBiosInsert.sh

at least, i placed the vbios ROM image inside the folder...

i run

jorghenninges@iHackPro ~ % cd desktop/gop
jorghenninges@iHackPro gop % ./vbiosinsert.sh -a gk107.rom enablegop.efi gk107gop.rom
EfiRom not available!
UEFIRomExtract not available!
jorghenninges@iHackPro gop %

what am i doing wrong?

UNIX is case sensitive. EnableGop.efi is not the same file as enablegop.efi

 

[h9826790]

UNIX is case sensitive. EnableGop.efi is not the same file as enablegop.efi

According to his discription. Most likely he didn't add that GOP folder into the "path", but not case sensitive issue. Otherwise, there is no vbiosinsert.sh in that folder (if case sensitive).

 

[macgirl-mx]

Success on Mac Pro 4,1>5,1 AMD RX560 4GB, thanks

 

[iModFrenzy]

Managed to get it working on my 5,1 as well. Tested it with a RX580 Pulse and 5700XT Pulse, so we can add that to the list of working hardware

 

[Henninges]

finaly i managed to get the vbiosinsert script running...thanks to @h9826790 + @Bmju.
as i tested several rom's, which some of was successfull, but some are not...

getting a bit offtopic here, so i put the story in a spoiler...

Spoiler

as i have a wall full of bricked and brocken hardware, there was a still known as good hd6670 which produce some red lines when insert in the cMP. just as a test objekt, i unpinned it from the wall and read the rom with gpu-z

f.e., an HD6670 which does not have any GOP part inside of the vbios. conclusion was for me, to update the GOP of the vbios with GOP_Updater ( success ), but the file size truncate from 131.072 byte to 121.344 byte.

vbiosinsert script refuse to insert the driver and reports :

File size of 121344 bytes must be at least safe size of 131072 bytes; use -s or check file

same is with an firepro v4800 cards vbios...

the nvidia quadro k600 vbios does a bit better, the size of the rom file expanded from 205.312 byte to 217.600 bytes, the file could be flashed and shows the native bootpicker when insert into my cMP, but not when plugged into my xserve2,1...

is there any solution to get the ancient ATI/AMD cards running with an modded vbios instead of the modded firmware of the mac?

i'd like to run the firepro or the quadro inside of the xserve...

 

[Borowski]

Xserve is closer to older Mac Pro and GOPEnabler incompatible. I guess it has also EFI32-firmware.

 

[Macschrauber]

Xserve is closer to older Mac Pro and GOPEnabler incompatible. I guess it has also EFI32-firmware.

Xserve 2,1 is a little comparable to a Mac Pro 3,1
Xserve 3,1 is a little comparable to a Mac Pro 4,1

so they both should have a 64 bit EFI

 

[Bmju]

finaly i managed to get the vbiosinsert script running...thanks to @h9826790 + @Bmju.
as i tested several rom's, which some of was successfull, but some are not...

getting a bit offtopic here, so i put the story in a spoiler...

Spoiler

as i have a wall full of bricked and brocken hardware, there was a still known as good hd6670 which produce some red lines when insert in the cMP. just as a test objekt, i unpinned it from the wall and read the rom with gpu-z

f.e., an HD6670 which does not have any GOP part inside of the vbios. conclusion was for me, to update the GOP of the vbios with GOP_Updater ( success ), but the file size truncate from 131.072 byte to 121.344 byte.

vbiosinsert script refuse to insert the driver and reports :

File size of 121344 bytes must be at least safe size of 131072 bytes; use -s or check file

same is with an firepro v4800 cards vbios...

the nvidia quadro k600 vbios does a bit better, the size of the rom file expanded from 205.312 byte to 217.600 bytes, the file could be flashed and shows the native bootpicker when insert into my cMP, but not when plugged into my xserve2,1...

is there any solution to get the ancient ATI/AMD cards running with an modded vbios instead of the modded firmware of the mac?

i'd like to run the firepro or the quadro inside of the xserve...

Regarding xserve, I've never used it. The basic rule for EnableGop is, if the card _can_ be made to work once OpenCore is running, on that machine, then there is some hope. Does that apply?

Regarding the short sized ROM, please can you attach or PM me the ROM? I will have a look. The general idea is that AMD cards have data which they need above the 128k limit, so a card with nothing above it sounds strange to me. Perhaps it is just left alone and won't be overwritten, if done correctly. Or perhaps (less likely) there isn't anything. Can you specify what tool(s) you are using to read from and write to the vBIOS on that card?

 

[Bmju]

Xserve 2,1 is a little comparable to a Mac Pro 3,1
Xserve 3,1 is a little comparable to a Mac Pro 4,1

so they both should have a 64 bit EFI

Based on this, there is more hope for the Xserve 3,1 - though I would need to get a copy of the (most recent) firmware; and less hope for the Xserve 2,1.

 

[Macschrauber]

Based on this, there is more hope for the Xserve 3,1 - though I would need to get a copy of the (most recent) firmware; and less hope for the Xserve 2,1.

have sent it in the conversation ;-)

 

[Henninges]

Regarding xserve, I've never used it. That basic rule for EnableGop is, if the card _can_ be made to work once OpenCore is running, on that machine, then there is some hope. Does that apply?

Regarding the short sized ROM, please can you attach or PM me the ROM? I will have a look. The general idea is that AMD cards have data which they need above the 128k limit, so a card with nothing above it sounds strange to me. Perhaps it is just left alone and won't be overwritten, if done correctly. Or perhaps (less likely) there isn't anything. Can you specify what tool(s) you are using to read from and write to the vBIOS on that card?

on the way...

 

[LightBulbFun]

Based on this, there is more hope for the Xserve 3,1 - though I would need to get a copy of the (most recent) firmware; and less hope for the Xserve 2,1.

funny this talk of Xserve's I was just wondering myself what compatibility with the Xserve3,1 might be like

as mention Hardware/firmware wise its a lot Like MacPro4,1 complete with a later BootROM that was never issued as an update for existing machines

https://forums.macrumors.com/threads/xserve-bootrom-upgrade.2224874/

 

[Bmju]

funny this talk of Xserve's I was just wondering myself what compatibility with the Xserve3,1 might be like

as mention Hardware/firmware wise its a lot Like MacPro4,1 complete with a later BootROM that was never issued as an update for existing machines

https://forums.macrumors.com/threads/xserve-bootrom-upgrade.2224874/

If someone has an xserve3,1, with that very latest bootrom, and preferably with OpenCore, so that we can determine that we are starting in the right place (i.e. with a GPU which works pre-boot once in OpenCore, but not before OpenCore), then I can look further... I don't know if anyone is actually in that situation?

 

[LightBulbFun]

I someone has an xserve3,1, with that very latest bootrom, and preferably with OpenCore, so that we can determine that we are starting in the right place (i.e. with a GPU which works pre-boot once in OpenCore, but not before OpenCore), then I can look further... I don't know if anyone is actually in that situation?

well I have exactly that hence my pondering guess Ill have to drag it out and get a test bench setup!

 

[Macschrauber]

Based on this, there is more hope for the Xserve 3,1 - though I would need to get a copy of the (most recent) firmware; and less hope for the Xserve 2,1.

I did a quick look setting the generic empty firmware of
MP 3,1 and Xserve 2,1 and
MP 4,1 and Xserve 3,1 side by side. They have a lot in common including most of the GUIDs

need some populated dumps to see more.

 

[LightBulbFun]

I did a quick look setting the generic empty firmware of
MP 3,1 and Xserve 2,1 and
MP 4,1 and Xserve 3,1 side by side. They have a lot in common including most of the GUIDs

need some populated dumps to see more.

thats im wondering has anyone tried this on a stock 4,1? or has all the testing just been on flashed 4,1's?


also to anyone else out there with an Xserve, do feel free to also drag yours out and test away, it may take me a little bit of time to drag mine out so dont wait for me so to speak!

 

[roobarb!]

also to anyone else out there with an Xserve

Oh, no. Mine is lying in the corner of a datacentre and now I'm tempted to bring it home.

If I bring it home it can't go back, and if I bring it home I don't have room for it. I feel like it's become some sort of Schrödinger's Xserve at this point.

 

[iModFrenzy]

Oh, no. Mine is lying in the corner of a datacentre and now I'm tempted to bring it home.

If I bring it home it can't go back, and if I bring it home I don't have room for it. I feel like it's become some sort of Schrödinger's Xserve at this point.

I mean I could take that piece of junk off your hands (the noise would probably get me killed)

 

[Bmju]

Oh, no. Mine is lying in the corner of a datacentre and now I'm tempted to bring it home.

If I bring it home it can't go back, and if I bring it home I don't have room for it. I feel like it's become some sort of Schrödinger's Xserve at this point.

Maybe I shouldn't have mentioned that this might be possible (and it is only might!) - the advantage of bringing the MacPros and iMacs back to life (via OpenCore and OCLP, of which EnableGop is just a small part) is that a lot of people are still using them... rather than having to dig them out of a corner just to try something! 😉

 

[Bmju]

f.e., an HD6670 which does not have any GOP part inside of the vbios. conclusion was for me, to update the GOP of the vbios with GOP_Updater ( success ), but the file size truncate from 131.072 byte to 121.344 byte.

vbiosinsert script refuse to insert the driver and reports :

File size of 121344 bytes must be at least safe size of 131072 bytes; use -s or check file

@Henninges - Thanks for PM-ing the files!

Just to update for anyone else reading this: if you do have an AMD file below 128k in size (e.g. after the output of GOP_Updater) then you can just pad it with FFFFFFFF.

However, be aware that you after doing that you _may_ (depending on your ROM) get the different error - Not enough space within 128k limit - aborting!. This is not so easily fixed; it would mean you need to go onto the iMac threads and find a ROM for the card with VGA stripped (so that there will be enough space for EnableGop) and GOP added (optional, since you already know how to do this yourself, if you are in this situation). See the 'AMD' spoiler in the first post for more info.

BTW if you are on a Mac Pro (or on an iMac, but it's harder to do) you can avoid all this by installing EnableGop in your main firmware instead of your vBIOS. You still need to add GOP support to your card (if it doesn't have GOP already, and most slightly newer cards already do; you can basically tell this by whether it works in OpenCore or not), but you don't need any of the other harder-to-make changes mentioned above.

 

[719c6]

Don't know if this is a thing but I get a black boot screen after rebooting from OC and pressing ALT for the Apple boot picker. I get the white screen after non OC re-boots. Everything works as it should but the traditional white screen is black. 5,1 & RX580

 

[Henninges]

happend to me as well...not a big deal.

 

[Macschrauber]

Don't know if this is a thing but I get a black boot screen after rebooting from OC and pressing ALT for the Apple boot picker. I get the white screen after non OC re-boots. Everything works as it should but the traditional white screen is black. 5,1 & RX580

This is because OpenCore sets the background color to black in NVRAM, this is expected.

When I first saw this way back in time I was totally confused thinking how the hell is OpenCore loading even when all ESPs are out of the machine. It was just the black color

The version number in the lower right corner clearly shows if it's OpenCore or the native boot picker.

 

[719c6]

This is because OpenCore sets the background color to black in NVRAM, this is expected.

When I first saw this way back in time I was totally confused thinking how the hell is OpenCore loading even when all ESPs are out of the machine. It was just the black color

The version number in the lower right corner clearly shows if it's OpenCore or the native boot picker.

Glad to know. Thanks gentlemen for all you do!

 

[MacNB2]

Don't know if this is a thing but I get a black boot screen after rebooting from OC and pressing ALT for the Apple boot picker. I get the white screen after non OC re-boots. Everything works as it should but the traditional white screen is black. 5,1 & RX580

The default back ground colour that OpenCore uses is set in the config.plist NVRAM variable. As stated in the manual:

• 4D1EDE05-38C7-4A6A-9CC6-4BCCA8B38C14:DefaultBackgroundColor
  Four-byte BGRA data defining boot.efi user interface background colour. Standard colours include BF BF BF 00 (Light Gray) and 00 00 00 00 (Syrah Black). Other colours may be set at user’s preference.

So if you wish to see the classic light grey then set DefaultBackgroundColor to BFBFBF00

I find default black background useful..to know that I have booted via OpenCore (especially when the OpenCore version display in the OpenCore picker menu title feature is not enabled via ExposeSensitiveData setting)

So your RX580 is behaving as it should after EnableGop and booting via OpenCore.