MBP 1,1-10,2 Force 2011 MacBook Pro 8,2 with failed AMD GPU to ALWAYS use Intel integrated GPU (EFI variable fix)

[Alex Crane]

Yes, gpu-policy=%01 remains after 3 restarts and 2 shutdowns.

By the way, my sudo nvram fa4ce...gpu-power-prefs... remains after 1 month of restarts, shutdowns and 6 system installs El Capitan, Yosemite and Sierra.

Added sudo nvram gpu-policy=%01 to my script. Still not survive normal reboot.

Could you please explain how did you apply this settings exactly? Linux, Single User Mode or normal macOS boot? Do you have SIP enabled or disabled?

P.S.: I haven't moved ATI kexts to the "backup" folder.

 

[MikeyN]

My current working solution is

• to make following shell-script (gpufix.sh)
  

  sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
  sudo nvram SystemAudioVolume=%00

  
  Second command is for silent boot as you know.
  
• CMD + S during boot
  
• running my script
• reboot

For now biggest problem is self-restoring of nvram.

What content has your file?

/private/var/db/.com.apple.iokit.graphics :

$ cat /private/var/db/.com.apple.iokit.graphics

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IGPU@2/AppleIntelFramebuffer@0/display0/AppleBacklightDisplay-610-9cb7</key>
<dict>
<key>version</key>
<integer>2</integer>
</dict>
</dict>
</plist>


Since TheFix it stays this short way.

There used to be several entries about several things in there.


Also keep an eye on the output of
nvram -p
https://github.com/ah-/nvram-dump/blob/master/nvram-dump.m

Added sudo nvram gpu-policy=%01 to my script. Still not survive normal reboot.

Could you please explain how did you apply this settings exactly? Linux, Single User Mode or normal macOS boot? Do you have SIP enabled or disabled?

P.S.: I haven't moved ATI kexts to the "backup" folder.

You run a script that also sets the chime volume to zero every time?
The nvram settings are all cleared?

 

[Alex Crane]

You run a script that also sets the chime volume to zero every time?

Yes. If chime sound isn't muted, then I understand that dGPU isn't disabled.

Since TheFix it stays this short way.

What is "TheFix"? %))


My /private/var/db/.com.apple.iokit.graphics

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>Alias:0/AppleBacklightDisplay-610-9c98</key>
    <dict>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAMAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAACA2i0JAAAAAIDaLQkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAIAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
    <key>Alias:0/AppleBacklightDisplay-610-9ccd</key>
    <dict>
        <key>cyuv</key>
        <integer>268435456</integer>
        <key>pscn</key>
        <integer>10000</integer>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAIAAEACAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAAA/KBoJAAAAAD8oGgkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
    <key>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IXVE@10/IOPP/IGPU@0/NVDA,Display-A@0/NVDATesla/display0/AppleBacklightDisplay-610-9c98</key>
    <dict>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAMAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAACA2i0JAAAAAIDaLQkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAIAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
</dict>
</plist>

Actually, a had MacBookPro5,2 that died several days ago. Now I bought MacBookPro8,3 and just moved my SSD from old machine to the new one. Maybe this is my problem (NVDA string)? I'm afraid to reinstall macOS due to corrupted dGPU. What if it is needed during macOS install?

 

[MikeyN]

Yes. If chime sound isn't muted, then I understand that dGPU isn't disabled.



What is "TheFix"? %))


My /private/var/db/.com.apple.iokit.graphics

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>Alias:0/AppleBacklightDisplay-610-9c98</key>
    <dict>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAMAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAACA2i0JAAAAAIDaLQkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAIAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
    <key>Alias:0/AppleBacklightDisplay-610-9ccd</key>
    <dict>
        <key>cyuv</key>
        <integer>268435456</integer>
        <key>pscn</key>
        <integer>10000</integer>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAIAAEACAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAAA/KBoJAAAAAD8oGgkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
    <key>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IXVE@10/IOPP/IGPU@0/NVDA,Display-A@0/NVDATesla/display0/AppleBacklightDisplay-610-9c98</key>
    <dict>
        <key>startup-timing</key>
        <data>
        AAAAAAAAAMAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAAAgNotCQAAAACA2i0JAAAAAIDaLQkAAAAAgAcAAKAAAAAwAAAAIAAAALAE
        AAAjAAAAAwAAAAYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        AAIAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
        </data>
        <key>version</key>
        <integer>2</integer>
    </dict>
</dict>
</plist>

Actually, a has MacBookPro5,2 that died several days ago. I bought MacBookPro8,3 and just moved my SSD from old machine to the new one. Maybe this is my problem? I'm afraid to reinstall macOS due to corrupted dGPU. What if it is needed during macOS install?

TheFix is shorthand for what's described in the first post of this thread, or what nsgr said:
sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
issued from anywhere and then preventing AMDRadeonX3000.kext from loading directly at boot.

The SSD or an old OS on it is quite likely no problem. Although, what is "NVDA-string"?
With TheFix in place an OS install will proceed to a certain point and then reboot and crash (here) since it then will have tried to load X3000.
Then it it's time for the SIP-dance and move the kext. Can be done. I installed High Sierra only after the dGPU died.

If everything from NVRAM is set to default then it does not point very much in the direction of GPU hacks applied wrongly or any software acting up.
Make several runs of

nvram -p
https://github.com/ah-/nvram-dump/blob/master/nvram-dump.m

Are there differences in behavior between reboot or shutdown and cold boot?
What happens starting the system from a fresh/clean install?
Clean install, same behavior you describe:
Then for me it starts to look as if the SMC itself or something leading up to it is a bit shaky here.
The battery, the backup-capacitor for the NVRAM, the onewire-circuit…
Never actually happened to me. Is the battery is in fair condition?
My next step would be to disconnect power cord, open case, disconnect battery from rest, let it sit for some time to drain the capacitor, reconnect all, see what happens.

 

[JohnMacAndRow]

My 2011 17" MBP's seems to be having problems with the discrete graphics card. Striped veritical lines at bootup and then white screen with fans blowing forever.

Before trying your fix to switch to integrated card at bootup, I'd like to know if it is easily reversible if I later decide to get the logic board or graphics card fixed. And if so, how do I do it?

 

[MikeyN]

My 2011 17" MBP's seems to be having problems with the discrete graphics card. Striped veritical lines at bootup and then white screen with fans blowing forever.

Before trying your fix to switch to integrated card at bootup, I'd like to know if it is easily reversible if I later decide to get the logic board or graphics card fixed. And if so, how do I do it?

Yes. it is reversible.
The trick is in writing a variable to the NVRAM. If you reset the SMC/PRAM that variable gets deleted and then recreated with default values.
But if your logic board is replaced then the new one already comes with these default values in place. So there is no need to worry or reverse anything.

 

[Alex Crane]

The SSD or an old OS on it is quite likely no problem. Although, what is "NVDA-string"?

My /private/var/db/.com.apple.iokit.graphics

...
OService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IXVE@10/IOPP/IGPU@0/NVDA,Display-A@0/NVDATesla/display0/AppleBacklightDisplay-610-9c98
...


Your file contains

...
<key>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IGPU@2/AppleIntelFramebuffer@0/display0/AppleBacklightDisplay-610-9cb7</key>
...


Can I somehow reset this file?

 

[MikeyN]

My /private/var/db/.com.apple.iokit.graphics

...
OService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IXVE@10/IOPP/IGPU@0/NVDA,Display-A@0/NVDATesla/display0/AppleBacklightDisplay-610-9c98
...


Your file contains

...
<key>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IGPU@2/AppleIntelFramebuffer@0/display0/AppleBacklightDisplay-610-9cb7</key>
...


Can I somehow reset this file?

Ah! NVDA.
First I moved then I deleted the file, rebooted and arrived at the short one above.

 

[nsgr]

Added sudo nvram gpu-policy=%01 to my script. Still not survive normal reboot.

Could you please explain how did you apply this settings exactly? Linux, Single User Mode or normal macOS boot? Do you have SIP enabled or disabled?

P.S.: I haven't moved ATI kexts to the "backup" folder.

Forget Linux mode. NVRAM command inside Linux do not work for this purpose.

1 - Always use Mac OS mode -> boot Safe Mode (Press SHIFT key at boot) -> (commands typing inside Mac OS boot Safe - Graphical Interface -> Applications -> Utilities -> Terminal

sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00

sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-active=%01%00%00%00

sudo nvram gpu-policy=%01

sudo nvram boot-args="-v"


2 - Then restart the system with this command:

sudo shutdown -r now


3 - Boot Normal

No move AMDRadeonX3000.kext from /System/Library/Extensions then freeze at IOConsoleUsers IOLockScreenState 3


You could use Mac OS Single User mode (command + S at boot), but the probability that you will miss some letter or number of the GUID would increase.

Copy and paste the commands would be better.


Update:

I forgot to put the main command first. Now I put it as the first item.

sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
[doublepost=1502060913][/doublepost]

Ah! NVDA.
First I moved then I deleted the file, rebooted and arrived at the short one above.

NVDA is Nvidia kext.

2º - Intel + Nvidia = Remove the Nvidia kexts with these commands:
http://www.insanelymac.com/forum/topic/299794-how-to-delete-kexts-for-enable-video/


Update2:

If you have the problem of the AMD chip being primary (red screen with white stripes) then you will have problems with the Recovery Mode Graphical Interface (freeze on grey screen).

Use only the Single User mode or Safe Mode or Recovery Mode Text Only in Mac OS to enter the NVRAM commands.

Recovery Mode Text Only = Command + R + S at boot.

 

[to*y]

Hi AppleMacFinder, I have tried following the steps in your solution but without success.

When I list the contents of EFI vars i do not see an entry for "gpu-power-prefs-..." nevertheless I flowed the rest of the instructions but not sure if I should have or have missed something?

Also, in the "printf..." step, I entered this on the command line but is it a change I should make to a specific file, and if so, how?

Thanks for any help you can give, desperate to get my MBP working again.

Toby

 

[SirMeowington5]

Check my post #528 for simpler instructions.

Hi AppleMacFinder, I have tried following the steps in your solution but without success.

When I list the contents of EFI vars i do not see an entry for "gpu-power-prefs-..." nevertheless I flowed the rest of the instructions but not sure if I should have or have missed something?

Also, in the "printf..." step, I entered this on the command line but is it a change I should make to a specific file, and if so, how?

Thanks for any help you can give, desperate to get my MBP working again.

Toby

 

[idoit]

Leave only the black color mirroring chip on the green (most probably) pcb. Dont forget to cover EVERYTHING in aluminum foil of tripple or more layers apart from that black chip. Search some videos there are on youtube.

Thanks for your reply. Can you explain a bit more? What do you mean by "leaving" ? Should I take all the shields off from the 4 chips? or I can leave the other 2 untouched (just take the heat sink off)?

 

[JohnMacAndRow]

Yes. it is reversible.
The trick is in writing a variable to the NVRAM. If you reset the SMC/PRAM that variable gets deleted and then recreated with default values.
But if your logic board is replaced then the new one already comes with these default values in place. So there is no need to worry or reverse anything.

Thank you for the quick reply!

I just tried the fix and it appears to be working now!!! I really appreciate everyone's contributions to this thread to help make it so useful.

COMPUTER:
Late 2011 17" MBP that had the thick vertical stripes at bootup and then would go into a blank white screen forever. Had tried booting in various modes and resetting stuff per Apple's suggestions and nothing was working.

WHAT I DID:
Followed the ArchLinex / EFIVARS instructions.
When I went to try and remove gpu-power-prefs* nothing showed up. I went ahead with the rest of the instructions and then rebooted and it booted fine, but then within a couple minutes it would appear to freeze. The mouse curser would disappear and everything was frozen.

Then I tried moving the AMD* files per the instructions in the first part.

After doing this it now boots up fine and was able to try a few apps without problems as well.

 

[MikeyN]

Check my post #528 for simpler instructions.

Wouldn't this be still simpler and faster for a first time fix?
(Can't test this for ElCap and later right now. Maybe the boot drive has to be mounted first while in single user recovery?)

– Reset SMC
– Reset PRAM
– Boot into recovery single user mode (<Cmd> + <R> + <S>)
– csrutil disable # on ElCap and later; skip this on Yosemite
– nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
– mkdir /Volumes/<HD>/AMD_Kexts # replace <HD> with the name of your boot drive
– mv -v /Volumes/<HD>/System/Library/Extensions/AMDRadeonX3000.kext /Volumes/<HD>/AMD_Kexts/
– reboot

*) Boot into OS X normally

– Start Terminal
– sudo kextload /AMD_Kexts/AMDRadeonX3000.kext

 

[SirMeowington5]

Sure, just remember to turn SIP back on again.

Wouldn't this be still simpler and faster for a first time fix?
(Can't test this for ElCap and later right now. Maybe the boot drive has to be mounted first while in single user recovery?)

– Reset SMC
– Reset PRAM
– Boot into recovery single user mode (<Cmd> + <R> + <S>)
– csrutil disable # on ElCap and later; skip this on Yosemite
– nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
– mkdir /Volumes/<HD>/AMD_Kexts # replace <HD> with the name of your boot drive
– mv -v /Volumes/<HD>/System/Library/Extensions/AMDRadeonX3000.kext /Volumes/<HD>/AMD_Kexts/
– reboot

*) Boot into OS X normally

– Start Terminal
– sudo kextload /AMD_Kexts/AMDRadeonX3000.kext

 

[MikeyN]

Then I tried moving the AMD* files per the instructions in the first part.
After doing this it now boots up fine and was able to try a few apps without problems as well.

Congratulations on this success.
But please, keep on reading more of this thread. It is very likely that this is a working but not an ideal solution.
To avoid future and further temperature trouble it might be advisable to move all AMD-kexts back, except AMDRadeonX3000.kext. And then loading that lonely kext manually only after booting has finished.

Forget Linux mode. NVRAM command inside Linux do not work for this purpose.

1 - Always use Mac OS mode -> boot Safe Mode (Press SHIFT key at boot) -> (commands typing inside Mac OS boot Safe - Graphical Interface -> Applications -> Utilities -> Terminal

sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-active=%01%00%00%00
sudo nvram gpu-policy=%01
sudo nvram boot-args="-v"

2 - Then restart the system with this command:

sudo shutdown -r now

3 - Boot Normal

No move AMDRadeonX3000.kext from /System/Library/Extensions then freeze at IOConsoleUsers IOLockScreenState 3

You could use Mac OS Single User mode (command + S at boot), but the probability that you will miss some letter or number of the GUID would increase.

Copy and paste the commands would be better.

Update:

I forgot to put the main command first. Now I put it as the first item.

sudo nvram fa4ce28d-b62f-4c99-9cc3-6815686e30f9:gpu-power-prefs=%01%00%00%00
[doublepost=1502060913][/doublepost]

NVDA is Nvidia kext.

2º - Intel + Nvidia = Remove the Nvidia kexts with these commands:
http://www.insanelymac.com/forum/topic/299794-how-to-delete-kexts-for-enable-video/

Update2:

If you have the problem of the AMD chip being primary (red screen with white stripes) then you will have problems with the Recovery Mode Graphical Interface (freeze on grey screen).

Use only the Single User mode or Safe Mode or Recovery Mode Text Only in Mac OS to enter the NVRAM commands.

Recovery Mode Text Only = Command + R + S at boot.

Still looking for the optimal solution.
What happened to agc=0 in your setup?
I now tried all of these added settings. That is setting gpu-active, setting gpu-policy. Latest boot was with all of these and agc=0.
All the while keeping a close eye on GPU temps with every combination.
On Yosemite I do not see a systematic and significant difference for them compared to simply setting the gpu-power-pref. GPU-diode now: 3°C.

 

[nsgr]

Congratulations on this success.
But please, keep on reading more of this thread. It is very likely that this is a working but not an ideal solution.
To avoid future and further temperature trouble it might be advisable to move all AMD-kexts back, except AMDRadeonX3000.kext. And then loading that lonely kext manually only after booting has finished.




Still looking for the optimal solution.
What happened to agc=0 in your setup?
I now tried all of these added settings. That is setting gpu-active, setting gpu-policy. Latest boot was with all of these and agc=0.
All the while keeping a close eye on GPU temps with every combination.
On Yosemite I do not see a systematic and significant difference for them compared to simply setting the gpu-power-pref. GPU-diode now: 3°C.

I temporarily removed agc=0 for other tests.

I would like Apple to release a document on the settings agc, gpu-policy, gpu-active. That would make our job a lot easier.

Unfortunately I could no longer make agc=0 work with AMDRadeonX3000.kext inside /System/Library/Extensions.

I noticed that with gpu-policy=%01 no more appeared the message:

Without gpu-policy=%01 in nvram
AGC: booted to IG, policy disable!!
AGC: 3.12.8, HW version=1.9.23, flags:0, features:20600

With gpu-policy=%01 in nvram
AGC: 3.12.8, HW version=1.9.23, flags:0, features:20600


You have not tested with El Capitan?


Update:

Macbook Pro 2016 using gpu-policy=%01 . This variable is still used by Apple.

https://macperformanceguide.com/blo...Apple2016MacBookPro-enable-startup-chime.html

 

[to*y]

Check my post #528 for simpler instructions.

Thanks, will try that today.

 

[Savroula]

Thanks for your reply. Can you explain a bit more? What do you mean by "leaving" ? Should I take all the shields off from the 4 chips? or I can leave the other 2 untouched (just take the heat sink off)?

Heatsinks must be taken off. Just remember to put them back on, after the procedure. And this procedure is at your own risk. It mostly works, not always. Google some videos about it on the internet to get an impression of what should you do. Dont go yolo

 

[kohsamui100]

I had my 2011 MBP opened, cleaned, GPU reflown and new thermal paste placed. So far it works great and looks like the GPU isn't heating too much (the fans were totally clogged and the paste used before the repair was white cheap one).

As I'm not really using a dual monitor setup, and want to have my MBP running good for many years more, I wonder if it's a good idea to still let it only use the Intel GPU (the way described here). This will make sure the AMD GPU still works whenever I want to put it back to action...

Also, how do you reverse this change (moving out AMDRadeonX3000.kext from the Extensions folder) ? When I tried to move it back, Sierra didn't seem to load it, and it worked like it's in Safe Mode (slow, extremely sluggish graphics)...

Thanks!

 

[MikeyN]

I had my 2011 MBP opened, cleaned, GPU reflown and new thermal paste placed. So far it works great and looks like the GPU isn't heating too much (the fans were totally clogged and the paste used before the repair was white cheap one).

As I'm not really using a dual monitor setup, and want to have my MBP running good for many years more, I wonder if it's a good idea to still let it only use the Inter GPU (the way described here). This will make sure the AMD GPU still works whenever I want to put it back to action...

Also, how do you reverse this change (moving out AMDRadeonX3000.kext from the Extensions folder) ? When I tried to move it back, Sierra didn't seem to load it, and it worked like it's in Safe Mode (slow, extremely sluggish graphics)...

My current wisdom is that a working AMD is better than a non working one and booting in default mode without this hack beats integrated mode in regard to power and temperature management. Currently I fear that the bad temps with this hack in place still accelerate the deterioration of that chip.
Were I new to this I would advise against keeping this hack if the AMD is working at all. Instead using gfxCardStatus to immediately force on integrated only. That way it is also easier to use a second monitor or whatever for you is "putting it back into action".
However, I did exactly this, after the last replacement running integrated as much as possible, and it still brought me into this situation. Reflowing is a temporary fix. It will break again. Since you already had a clogged exhaust: open the unit up from time to time and use compressed air to blow the fans and fins clean.

To find out which kexts are loaded use kextviewr or the built-in options:
– sudo kextstat

How to move it back depends on how you moved it out. And only X3000 or all AMD kexts?
sudo mv on the command line should be easily reversible.
Just sudo mv again this time into /System/Library/Extensions then
sudo touch /System/Library/Extensions
and wait until the kextcache process has finished, only then reboot.

If you have moved it with e.g. Finder it is likely that the permissions are borked now. And the kext will not load because of that.
Repair Permissions with DiskUtility to fix that.
Otherwise just download and run the latest ComboUpdater that applies to your current system from Apple.
That will put all kexts back where they belong for sure.

 

[kohsamui100]

Thanks for the detailed explanation. I managed to "fix" the kext installation by running Sierra installation again from the recovery disk/usb.

When you say it will surely "break" again if been reflown... is it highly likely to happen?

My technician said he'd rather reflow than replace the entire GPU, as from his experience the act of replacing the GPU may even increase the chances that the problem returns... what do you think ?

How can I prevent it from failing again ? or what's the best I can do to make it work for as long as possible ?

Is failing again is a result of how much heat will be created inside the computer in the future ? what will happen if I run Macs Fan Control on high RPM constantly from now on, to keep it cooler than in automatic mode?

 

[Audit13]

According to Louis Rossman, reflowing is useless:

 

[kohsamui100]

Yes, I know this video. But the question is - which method will make it fail again sooner. Reflowing or removing the GPU and soldering in a new one ?

 

[Audit13]

I had the 6770m replaced in my MBP. The chip was already defective and reflowing would do nothing to resolve the issue. He said he replaced the existing chip with an updated revision of the same chip and it should be good for a few years.

Not sure why installing a new GPU would cause the new GPU to fail sooner. How did your technician make this determination? Has he replaced and reflowed many of these failed MBPs?

Does your technician have the equipment to remove the existing GPU or would s/he send it out?