Q: 2011 MacBook Pro and Discrete Graphics Card

The next error messages directly pointed me to GPU

"kernel[0]: GPU Hang State = 0x0000000"

"kernel[0]: Trying restart GPU"

Same to you I did not get the SMC related messages too serious

as the system was behaving absolutely normal.

And thats all I noticed in system.log

If you saw/see above message your discrete GPU is most likely gone @all and

not only have cracked solder joints.

In next steps the error behavior can be quite different over several units as no one

can forcast which way the heat took through the system or which component have lost fight first

and will loose next.

If BGA repair is done correct old chip and ALL old solder will be removed.. contacts cleaned

and NEW chip comes with new solder balls.

To positioning new chips @ 0.01 mm level and connect it, exactly temparature time specification

need to be met and quite expensive equipement is required.

You definitely won't get satisfied on trying to reflow only (reheating old solderballs under old chip)

which usually do all baking and heat gun methods

Additionally I can participate in expiriences 3 of my friends had with their MBPs A1260 and earlier.

2 of them had BGA chip replacement over 2 years ago.. and they are still running fine.

3rd already passed 15month line after repair.

My friends and my expiriences made me quite comfortable to go that way.

@Will-NY

addition to your 1st question.

system usually archives its log in gzip containers.

Inside console.app you can find them in bottom section /var/log

named  "system.log.0.gz" ...

If want to go manual way either in terminal oder single user mode

cd /private/var/log

ls -l system.log*

zgrep <searchstring> system.log.*.gz

I would like to help you (like me, there are many users with the same problem too in Spain). I did login a few days ago with the user name Miguel. Please, contact me by mail (I supose you can see it as administrator of the page).

<Link Edited By Host>

Finally my mac couldn't boot anymore.

Did every hard reset, pram, smc etc.

Hey apple guys! Take a look on this http://www.youtube.com/watch?v=ELE-R6-dZGQ

What number will connect me with someone in apple customer service? I tried calling some numbers but i have to pay to talk to anyone about this problem.

I did not seen, but if there is vertical blue line - probably you as many other people here, have defect in GPU chip... For me, it was fixed only when GPU chip was replaced by new one.

But if it is just blue screen - maybe it is software problem. Try to boot in safe mode and repair disk permissions (from disk utility), if it will not help - you can try to boot in verbose mode and try to figureout what is going wrong and then fix it from command line in single user mode(but it can be hard if you are not familiar with os internals).

But the most simple way, espessialy if you have backups, is - just boot into recovery and reinstall OS X.

If you have no backups and want to copy something - you can do this using: safe mode, single user mode, or target disk mode and anouther mac.

Thanks but first time it happened I did everything you're telling me. Clean install of Mavericks and it could boot, 5 minutes after blue screens, split image, system crashes. hard resets, installed gfx card status, used smc fan to boost fans to keep it cool... Tonight my mac couldn't boot even safe mode. Thankfully I backed up my mac yesterday.

Seems to be final stage for my macbook pro

Thanks anyway!

I dis permissions repair several times also.

Maybe you can not boot because something was corrupted on file system because of hard resets(I have something like, before GPU replacement).

But blue screens, split image and system crashes afrer clean install - looks more like hardware defect.

What do you have in system log? - Maybe there can be some usefull info.

If you have something like:

kernel[0]: GPU hang:
kernel[0]: Trying restart GPU …
kernel[0]: GPU Hang State = 0×00000000

Than, I am almost sure, that it is hw problem.

Thanks, but how do I get that info?

If you can boot - you can look in Console.app.

If not - you can view log files in single user mode: they are located in "/private/var/log/" folder.

You can view them typing 'less /var/log/system.log' or 'zless /var/log/system.log.0.gz' if you want view older logs.

Also if you want just test then for GPU errors - you can use 'grep' or 'zgrep' for this: 'cat  /var/log/system.log | grep GPU', 'zgrep GPU system.log.*.gz'

The only fix for this is a logic board replacement (or a reball type fix). Maybe you'll get lucky and it'll last? Not everone has had their replacement fail.

The issue on my late 2011 started failing the day after I installed 10.8.5. Mavericks didn't make much of a difference with the issue.

I did all kinds of hardware swaps (harddrive, optibay drive, and ram), and OS installs. It "fixed" itself for a week and then was suddenly totally usuable. I redid the hardware swaps and it never fixed it self again. Logic board repair and it's been ok for a few weeks now.

I did email Tim Cook and work with a corp relations person and a sr tech. Had to take it to the Genius Bar for and ask for and got the depot option. And I didn't have Apple Care. Explained all about this discussion and the commonalities with everyone I spoke to. After expressing doubts about the replacement lasting, she said that it would definately last. I hope so since I plan to raise heck if it doesn't!

19 months is too short a life for a computer I expected to last 5+ years.

jensgoehler wrote:

 

...I advised the shop to do the repair only after calling me back.

They sent analysis report and microscope photos of faulty solder joints and heat marks

round the GPU die back and offered me repairing as followed:

 

- new GPU                                                       148,00   <- GPU burned + cracked solder joints

- new GPU Vcore                                               48,60

- SMC resoldering                                                0,00   <- no replace required - "only" few solder balls cracked

- new thermal paste by "Prolimatech"                  0,00

- clean up                                                             0,00

- component stress test

- 48 hours longterm test                                       

- shipping                                                              0,00

 

...

Thanks for the detailed report.  Would you mind posting the imaging that they sent you? 

Prolimatech makes nice stuff.  Do you have before & after GPU temps to compare?

My MacBook Pro Early 2011 17-inch just stopped working today.

II'm running the lastest OS X (10.9). I'm pretty sure all the firmwares were updated.

Here's a photo of the pink vertical stripes:

-SMC reset, PRAM reset didnt fix this issue.

-Tried booting from the recovery volume, and safe boot. I got a black screen with no mouse pointer.

-I also noticed the GPU section (the part between two fans) of the logic board got really hot if I left it on.

-After reading some posts, I also have the hard freeze issue from time to time.

-Plus, I remembered sometimes few black rectangles with small colored dots will appear on the screen. not sure if anyone have the same scenario.

Im taking my MacBook Pro to the Apple Store tomorrow.

I hope Apple acknowledged this issue.

odarellmc wrote:

 

Thanks for the info. Can u apply paste yourself?

Agree with previous response that one should be cautious if never before having done anything of this nature.  That said, you can look at the iFixit links and some videos and get a feel for that yourself.

IMHO, the biggest danger is the same for both the novice and the enthusiast.  That's ESD.  The novice often thinks that such warnings are overstated and the enthusiast, or even seasoned professional, can often be on autopilot when it comes to ESD protection.

One thing missing from the popular links on this subject is the procedure for reattaching the heatsink, so I'll post that again for those planning to do this.

The popular instruction links end with something like:  "To reassemble your device, follow these instructions in reverse order."  That's not correct.  It should be done like this:

Screws should be reinstalled in the order shown and tightened down half way.  Then, following the same pattern, they should be tightened down the rest of the way.

For anyone considering a reball or GPU replacement, it's a given, of course, that this will be included with that service.  The thermal compound needs to be cleaned up and reapplied whenever the seal between die and heatsink is broken.  But it might be a good idea to ask them to verify that the heatsink fits properly (that both sections of the heatsink fit flat against their corresponding dies).