Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I would be interested in getting Windows 7 x64 to work. I think I have everything in line for it to work, including pure GPT and MSR, however I can't find the PCI registers for the Bridge and ATI 5870. How do I locate these and link these together?

Mac Pro 5,1

Thank you
 
Does someone test the new windows 8 drivers provided with the new bootcamp from 10.8.3. I could not test because the new manager said my iMac is only compatible with Win7 ! Liar !

Does someone could uploaded the Windows 8 files ?
 
Does someone test the new windows 8 drivers provided with the new bootcamp from 10.8.3. I could not test because the new manager said my iMac is only compatible with Win7 ! Liar !

Does someone could uploaded the Windows 8 files ?

I tried it on my iMac 2011, everything works fine except there's still no sound.
 
I tried it on my iMac 2011, everything works fine except there's still no sound.

What about your graphic card (ATI ? /NVIDIA ? /INTEL ?) ? Does it work without exclamation mark or other problem ? Which driver are you using (the one provided by Apple or for the manufacturer ?) You are in EFI mode ?

Does someone are tested the Leshcat’s Catalyst Version ? (http://leshcatlabs.net/index.php/category/unifl/)
 
Windows8 native EFI boot for MacBookAir2,1?

Model Identifier: MacBookAir2,1 (mid 2009)
The EFI is 64bit as shown by
Code:
Iioreg -l -p IODeviceTree | grep firmware-abi .... "firmware-abi" = <"EFI64">

There is no official support for my model, as stated here, hence I think the bootcamp package of Windows drivers will probably not contain all the drivers that I need. That may not be a problem because I might be possible to get them elsewhere.

Will there be any difference i the end result (except the windows drivers) if the Windows8 installation media is created in bootcamp compared to for example the Microsoft Windows7/8 USB Tool?

Previously I have managed to run windows8 32bit with no problems.

I have tried to prepare a USB flash drive formatted in both FAT32 and HSF with the windows 8 x64 installation files - but it didn´t boot using Apple´s bootloader.

Is there anything left to try?
 
Bootcamp 5.0 + Windows 8 EFI

Alright its bee awhile that I make post on this thread.

Since I got new MBP 13.3" 2012, I decided to install Windows 8 in EFI mode

Installing is very easy, no magic required (nothing has been changed from MSDN ISO image)

Bootcamp 5.0 driver installed fine as well.


Conclusion.
Everything works "EXCEPT" sound

even brightness can be changed (not like win7 + efi) but sound just not work!

Anybody have found solution for this problem?
 
Alright its bee awhile that I make post on this thread.

Since I got new MBP 13.3" 2012, I decided to install Windows 8 in EFI mode

Installing is very easy, no magic required (nothing has been changed from MSDN ISO image)

Bootcamp 5.0 driver installed fine as well.


Conclusion.
Everything works "EXCEPT" sound

even brightness can be changed (not like win7 + efi) but sound just not work!

Anybody have found solution for this problem?

I have a 13" rMPB, latest one released this year. Would your efi method work for mine? If so, could you link/describe the process for dualbooting OSX? Install win8 like normal and partition OSX,then install OSX? Thanks a ton!!
 
Is it possible to run Windows 7 in EFI mode on Mac Pro 2009 ?

I was able to run Windows 8 successfully, but I hate it :/
 
Is it possible to run Windows 7 in EFI mode on Mac Pro 2009 ?

I was able to run Windows 8 successfully, but I hate it :/

Have you managed to install Windows 8 to a 2009 MBA in EFI mode? I'm currently trying...

UPDATE
Installed successfully Win8 in EFI mode on MacBookAir 2009 (2.1). Only problem is still audio!
 
Last edited:
Have you managed to install Windows 8 to a 2009 MBA in EFI mode? I'm currently trying...

UPDATE
Installed successfully Win8 in EFI mode on MacBookAir 2009 (2.1). Only problem is still audio!

I don't have MBA. Only Mac Pro.
 
Brightness control on Win7 + efi

I have researched the reason why the brightness control doesn't work on win7 + efi, and found it is caused by an absence of SMI handler.
Can anyone tell me how to dump SMI handler from SMRAM in bios emulation mode?

According to my research on my macbook pro (9,1 Mid 2012),
the process of the brightness control when F1/F2 key pressed is as follows:
  1. OS (win7) detects F1/F2 key pressed.
  2. OS calls ACPI function "_BCM" in SSDT.
  3. "_BCM" writes a byte which indicates a brightness level in I/O port 0xB3.
  4. "_BCM" also writes 0xBF in I/O port 0xB2.
  5. Writing in 0xB2 generates SMI.
  6. SMI handler is called and writes in I/O port 0x774 which is apple gmux port for Brightness control.

The above #6 is my guess.
I monitored I/O and Memory space by using RW Portable "Read and Write Utility" http://rweverything.phpnet.us/
and compared it when booted via efi and bios emulation then it was same except #6.
I think, in bios emulation mode, SMI handlers are installed in SMRAM while booting, but they are not in efi mode.
If SMI handler can be installed in efi mode, I think the brightness control works.
 
Display brightness can be controlled via Apple System Management Controller. This requires very complex programming, but as the result this works in EFI mode. And actually I recently created such an utility for Macbook Pro: it controls the display backlight both in the BIOS emulation mode and EFI mode, it provides ultra-smooth, analog-like control (1024 levels). I made it for myself, but I think I can make it available to public later. It's a bit raw, as of today.

And by the way, my alternate trackpad driver for Boot Camp works well in EFI mode. See my topic at https://forums.macrumors.com/threads/1497761/
 
  • Like
Reactions: Teutone
Anybody solved Intel High Definition Audio problem with EFI?

I tested iMac12,1(2011 21.5") and MacBookPro10,2(2012 13.3") and both of them are having same problem

High Definition Device is unable to start (error code 10)

However, iMac 2012, and MacBookAir 2013 are fine with this issue as far as I know.


Someone please tell me why this is happening.

I have no idea why this is happening so I cant even start researching.

Thanks.
 
Windows 8 EFI on Macbook Pro 5,1 (Late 2008)

While waiting for good Haswell ultrabooks, I decided to play with my Macbook Pro 5,1 (Late 2008) and I want to share my (partially) successful experience with Windows 8 EFI:
- From another mac I backup the OS X Partition and wiped the disk and created pure GPT table and restored the partition but also created two more for the windows installation
- Slipstreamed the expanded nVidia drivers into an .iso and created autounattend.xml in the root and copied the .iso content on FAT32-formatted bootable usb
- Downloaded Tianocore EFI Shell and created startup.nsh (see https://forums.macrumors.com/showthread.php?p=14075475#post14075475)
- I renamed the Microsoft’s BOOTX64.EFI to MS-BOOTX64.EFI and pointed the efi script to start it
- Windows 8 installs smoothly and also Bootcamp 5 installs using msiexec /i and all the drivers are loaded except one “Coprocessor”. All Bootcamp controls are working except brightness.
- At this point I experienced very much random freezes per day. Internet is full with proposed fixes, but what helped me: I downloaded the latest nForce 980a SLI driver from nvidia site and installed in Win7 compatibility mode. The installer also updates the AHCI controller. I then manually updated the driver of the coprocessor from the folder created by the installer and since then Win8 is running very stable (no single crash) and very fast!

Thank you all for the hints in this thread!

So here the three issues:
- Hibernation: The pc wakes from hibernation with turned off black screen. I must close the laptop to trigger sleep and then wake to have a video output. But if I try to hibernate again, windows 8 simply restarts and says an error has occurred. I am only allowed to hibernate once between the restarts!
I tried to change the startup.nsh before hibernation and sometimes the computer wakes up even with video output. But this behavior is not reproducible. I experienced similar hibernation problems with win7 efi.
- After a sleep the expected battery time is little reduced. (Maybe the 9600gt is waked too).
- Brightness control: is not working. I tried various proposed fixes but none does. V-l-a-d-i-m-i-r, would you share your brightness control with us? I can live without hibernate but without brightness control, the macbook is pain in the... eyes!

EDIT:
Actually someone did a pretty good job on creating an Apple SMC Control at http://lwn.net/Articles/226624/ and this driver is used by a MacFan for Windows http://sourceforge.net/projects/macfan/ It gives access to fans, light sensor, keyboard and display brightness controls and says it doesn't need installed bootcamp drivers, so, theoretically, with a little programming one can even create an adaptable brightness on an efi windows.

EDIT2:
Newer Version here: http://stuff.mit.edu/afs/sipb.mit.edu/contrib/linux/drivers/hwmon/applesmc.c

EDIT3:
sleep battery issue fixed: reduced baterry expected time after wake can be fixed by sending 2 to port 0x750 (subscribe to WM_POWERBROADCAST or send it every 50s)
hibernate issue: I suspect the tianocore shell for initializing some pci values. OS X won't start after I exit the shell. If a self created efi sets the needed pci bits and starts the win bootloader, it may start working.
brightness issue: I tried different things such as sending bytes to ports 0x10724/0x774 or 0x52e/0x52f, which actually works under bootcamped win7, but without success. I can communicate with the SMC: read light sensors and change the keyboard led state. But I tried different SMC keys for the brightness and none is working.

EDIT4:
brightness issue fixed: no smc programming needed: the brightness value for stone age nvidia adapters must be send to the gmux port differently:
short GMUX_PORT_BRIGHTNESS = 0x774;
OutPort(GMUX_PORT_BRIGHTNESS, brightness);
OutPort(GMUX_PORT_BRIGHTNESS + 1, brightness >> 8);
OutPort(GMUX_PORT_BRIGHTNESS + 2, brightness >> 16);
OutPort(GMUX_PORT_BRIGHTNESS + 3, 0);
Newer nvidia drivers use only the newer option.
Next is to map the brightness keys to trigger the changes and I am done.
Code still needs cleanup. But if someone is interested pm me with mail adress and I will send the code.

EDIT5:
hibernate issue: The problem is sending the 2 to port 0x750: the gmux powers down the 9600m gt and it is not visible to windows anymore. After a restore from hibernation the device is powered up again and windows thinks a new device is installed and throws error. If I make the both graphic cards bus masters, they are visible in the device manager and I can even set the Physics to the 9600m gt, but I can't get video output from it. In this case hibernation always works (as long as 0x750 stays untouched) but the notebook starts with black screen and I only hear the starting sound. It is impossible(?) to prevent PCI devices from discovering by the OS. I think a virtual windows or efi driver must be created to power down the 9600m and its bus on each start / wake. Than maybe everything would be ok. Switching the adapters could even be possible if one understands which ports are triggered under osx.
 
Last edited:
I've found some interesting reading for those wanting to know why it is so difficult/impossible to get Windows 7 working in (u)efi mode on macs as opposed to Windows 8 - Windows 7 is not fully uefi compliant.

In the article "UEFI Support and Requirements for Windows Operating Systems"
http://www.microsoft.com/whdc/system/platform/firmware/uefireg.mspx it mentions that for Windows 7, "existing Windows dependencies on INT 10 video BIOS functions also require a CSM". In its advice to firmware manufacturers, it recommends that when secure boot is disabled ie for booting versions of windows other than Windows 8 (such as Windows 7), the firmware should

• Enable the CSM for VGA support, though not BIOS mode emulation.
• Enable messages during the POST process to show which keys open the boot menus.

In other words, Windows 7 still requires a "Compatibility Support Module" for basic (VGA) video functions in its "hybrid" uefi mode.

On macs, it is all or none: CSM is provided for video and everything else if you boot legacy BIOS emulation via bootcamp and a hybrid mbr disk. If you boot in pure (u)efi mode from a pure GPT formatted disk, no CSM is provided and hence Windows 7 uefi install hangs. Windows 8 is OK since it uses GOP for video instead of int10.

On PCs such as Lenovo X1 Carbon described in this thread http://forums.lenovo.com/t5/X-Serie...install-without-CSM-mode-possible/td-p/911353
there is an intermediate (u)efi mode with CSM provided for video only. This allows installation of Windows 7 in (u)efi mode.
 
Last edited:

It may boot with the CSM for video disabled but there will be no screen output until VGAE registers are set by the firmware and the uefi compatible video driver (if it exists) finally loads - see Devil's post https://forums.macrumors.com/posts/13255278/.

This may be done by the firmware automatically on Windows PCs (maybe a requirement of windows certification as the msdn article I quoted) but doesn't happen for macs, hence the great difficulty to install Windows 7 on macs.

I have a few examples where CSM has to be provided to Windows 7 to enable screen output during the initial installation procedure:

For his HP Probook 4540S, this user noted that "Windows 7 will only boot in efi mode if the bios is set to efi legacy mode" http://simonhawketts.wordpress.com/2013/05/28/probook-4540s-hackintosh-os-x-install/

On PCs such as Lenovo X1 Carbon described in this thread http://forums.lenovo.com/t5/X-Serie...D=10594818&PID=3160356&pubname=VigLink&nid=cj
there is an intermediate (u)efi mode with CSM provided for video only. This allows installation of Windows 7 in (u)efi mode. As noted here http://forums.lenovo.com/t5/T400-T5...hout-CSM-support-enabled-in-BIOS/ta-p/1038089 , "while it is technically true that Windows 7 itself can support a legacy-free (CSM disabled) UEFI installation, not all OEM component drivers were updated for legacy-free" ie no guarantee that the oem provides a (u)efi compatible video driver and the default Windows 7 VGA (basic) driver fails to work during windows 7 setup in uefi mode.
 
Last edited:
Macbook Pro 5.1 Vbios...

So no good haswell laptop around and I am still playing with my Macbook Pro 5.1
The current state:
- Hibernate always works if I send the computer to sleep some seconds after restore. If I forget it the saved state is lost
- “Fast Boot” works only 30% of the times. I don’t know why, but the fast boot is slower than the normal boot.
About the 0x750 gmux port: I browsed the Linux source for video switch and found out that one have to send
1 -> 0x750
0 -> 0x750
Source: http://lxr.free-electrons.com/source/drivers/platform/x86/apple-gmux.c
This increases the battery time more than just sending 2.
Discrete graphic card is still not working but there is an explanation:
http://linux-hybrid-graphics.blogspot.de/2012/04/apple-macbook-pro-and-linux-hybrid_21.html
If one dumps the video bios then copies it to the right memory address and sets one PCI register on each start, maybe the 9600m GT card will start functioning. It would be even possible to switch the graphic cards on the fly. I think this is also the root of the hibernate problem.
This is where I got stuck. If someone have done something or can help me with links I would appreciate it. @d3vi1 & others: would you share your apps with me?
 
Last edited:
If one dumps the video bios then copies it to the right memory address and sets one PCI register on each start, maybe the 9600m GT card will start functioning. It would be even possible to switch the graphic cards on the fly. I think this is also the root of the hibernate problem.

I've attempted that already and it's not enough. Apparently when the Nvidia OptionROM is POST-ed it changes itself on the fly assumably after detecting the configuration (outputs, RAM, frequencies, connected displays, etc.) and saves all that information in itself and in another two places in the lower memory area. That basically means that in order for VGA to work properly, I would have to do exactly what Linux and Xorg do and run the option rom in a virtual machine (x86emu) and then dump that 1'st MB of VM ram in the 1'st MB of the real system RAM. Which is where I abandoned the work (though there's room to pick it back up). Fortunately, EFI doesn't use that 1'st MB of ram and it's called a LEGACY MEMORY area which is protected and can be unlocked for writing. If you really want to do that, boot in Bootcamp, dump the 1'st MB of RAM, restore it in EFI, poke the correct IO ports that the VGA option rom pokes and it will work.
I already had the code to set the correct gmux values and dump the correct option rom at the 0xC0000 taken straight from the EFI_FIRMWARE_VOLUME. I implemented code that opens all the firmware volumes, goes through all of them in order to get all the option ROMS. Creates a vector with all of them and then, when you figure out which is the currently active VGA card (this is important!), load the option rom for it at the correct address.
I was thinking of picking it back up but in Apple EFI mode there are no ways of debugging other than printing on the screen, which tends to be quite broken if you suspend the EFI video driver in order to run the Option ROM. My suggested approach would be to do that on QEMU with OVMF which can be debugged running one of the classic VGA option ROMS. In QEMU with OVMF Windows < 8 doesn't run for the same VGA problem. It's all depending on availability and mood to be perfectly honest and I haven't touched the code in 12 months.
 
And there's one more thing that's crappy about the Apple implementation:
Some computers have BOTH option roms for both the discrete card and the integrated one and BOTH EFI Drivers. In theory, one could write an EFI driver that makes the system start up on one of the VGA cards or the other depending on user preference. However, others, like the MBP6,2 don't have an included Option ROM for the integrated GPU (Intel Ironlake for MBP6,2). The behavior is strange in other aspects:
MBP5,3 has both EFI and Option ROMs for both cards but starts from the integrated one.
MBP6,2 has EFI drivers for both cards and two identical Option ROMs for the discrete card and starts only from the discrete one.
MBP8,1 has EFI and Option ROMs for both cards and starts from the discrete one.

There's absolutely no logic in this. In my ideal world, I would finish writing a driver that allows you to:
1) Select the boot-up video device and load the EFI driver for that one solely.
2) If desired (configurable with an EFI variable) also load the VGA ROM for that card. Only useful for Vista SP2 or Windows 7, but not 8 unless it's a badly written driver that depends on int10.
3) Select if the other (non-active) VGA card is available or not. You could boot Windows 8 from the Intel GPU and leave the Nvidia one powered on for CUDA and OpenCL applications, but not for graphics duties. One would need to write a Windows driver in order to get also graphics switching and that's not on my list of interests.
4) Patch the EFI Runtime Services Table to also include RS->QueryVariableInformation which is the missing function for the boot loader mess. I've written this driver and it passes all the EFI SCT Tests in BootServices mode but once you go to Runtime Mode (and particularly do RS->SetVirtualAddressMap()) it's broken. I've tried converting my pointers using RS->ConvertPointer() but it didn't help. Maybe someone could look into that as it would fix Windows 8 and newer.
 
welcome back devil.
Since this thread started, windows already had two major update (if i count windows 8.1 as major)
From windows 8, most of user is fine with VGA issue.
However, Some Intel north bridge based macs having issue with audio problem.
Macs with nVidia chipset dont have issue with this audio problem but I saw many users with complaining audio is not work when user installs windows 8 in efi mode.
There are few exception model (mac mini 2012, imac 2013, mba 2013) but between 2010~2012 models, audio is one last big problem with efi installation.

Do you have any idea why this is happening?

I tried MBP 2012 13" and iMac 2011 21.5"
Both of them has a problem of audio.

I tried overriding DSDT assuming it might be problem but the problem still exist.

If you can tell us at least why this is happening, maybe we can figure a way to resolve it.

Thanks
 
Managed to get my Early 2011 17" MBP booting Windows 8.1, though it did take a bit of witchcraft. Had to use my Late 2011 Mac mini to do the first half of the installation.

I've got my drives configured as such:

240GB SSD in HDD bay
750GB HDD in optical bay

Put SSD in a USB enclosure, connected to Mac mini. Booted from 8.1 EFI USB installer. Used command prompt to partition the SSD. 256MB for EFI, 123GB OS X, 128MB for MSR, and 100GB for the Windows. After running through the dism install method (no slipstreaming or unattend needed), I rebooted, waited for the "Getting devices ready - 100%", then at the following restart, relocated the SSD back into my MacBook Pro and completed the setup.

Everything works, short of brightness, sound and accelerated graphics, and integrated graphics of course causes the black screen of mystery.

After that, I installed OS X, created a Fusion drive with my OS X partition and my 750GB HDD, and everything is still working great.

Has anybody had any luck getting the Radeon 6750M running? Methinks I just have to get the PCI registers corrected, but haven't seen any good guides on accomplishing this. I could get by without sound, but would really like to have accelerated graphics going.
 
The gfxCardStatus var

Thank you for the fast replay, d3vi1.
I now have a stable running win8 on my MBP 5,1 and I am only missing the hibernate option and would like to replace the EFI shell start.nsh with own written bootable EFI in order to decrease the starting time. Now it is about 10s to login with an Intel SSD, but I hate the 5s count down. (Not bad for a 5 years old laptop.) But nevertheless I can lose some time in trying more stuff.
Like I said in previous post I managed to get both graphic cards shown in the windows device manager and one can use CUDA on the 9600m and it only works if the gmux ports stay untouched at startup.
I browsed the Ubuntu source and forums these days to see what actually happens under Linux and found this interesting info about which graphic cards is started under MBP 5,1: It depends on a gfxCardStatus EFI variable according to this thread: http://ubuntuforums.org/archive/index.php/t-1076879.html. Is it possible that your macbooks are with energy safe mode turned off in the OS X control panel?
Actually we don’t need VGA if we have an EFI ROM and driver under Win 8.x. MS published GOP tests for Windows 8.x http://msdn.microsoft.com/en-us/library/windows/hardware/hh997769.aspx so one can see what is actually missing in the Apple’s EFI 1.1 with some UEFI extensions. (http://www.computerbase.de/bildstrecke/13102/6/)
I have also a similar imagination about what a custom bootable efi should do at startup. We can use the gfxCardStatus to detect which graphic was initialized by the apple efi. I would give a try to find out what is actually missing in QueryVariableInfo. I will pm you as long as I reach 5 posts. Until then I am not allowed to use the messaging system. Forum rules…
Next week I will try to find out if I can make the 9600m work under windows and run the MS GOP test.
 
Has anybody had any luck getting the Radeon 6750M running? Methinks I just have to get the PCI registers corrected, but haven't seen any good guides on accomplishing this. I could get by without sound, but would really like to have accelerated graphics going.

I gave up trying on an early 2011 15" MBP. Have you tried the latest ATI drivers? I forgot which ATI driver kit I used, but only a specific kit type will update the GPU driver on a Mac.

I doubt it will work, but it's worth a try.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.