Macbook Pro 2011 Logic Board/GPU failures

[christianronnel]

I know this is not camera related but I think there would be a lot of CR readers/members who may be using a Mac for post processing. I just want those Mac owners to be aware of this issue.

My 17-inch MBP (2011) just started showing GPU glitches when I use Photoshop or Lightroom. I cannot do anything to it but shutdown wait a few minutes before restarting. The machine would be usable for a few minutes only just enough time for me to get a file so that I can use in another computer. I'm out of warranty so I was getting ready to fix it. While I was searching for logic board repairer I stumbled on a thread from Apple forum https://discussions.apple.com/thread/4766577?start=3375&tstart=0regarding high failure rate on this model of MBP. Another google search and I found this article on MacRumors http://forums.macrumors.com/showthread.php?t=1697394&highlight=2011+macbook+pro

I'm curious if anyone here had the same problem? I know the photography/videography community would probably have a high percentage of Mac power users that would push the GPU/CPU to its limits.

 

[ksagomonyants]

I have exactly the same issue with my early 2011 Macbook Pro 15", i7, 8 gb RAM and 512 SSD. Started having this issue around 3 weeks ago or so. I looked at the pictures posted on that link you provided and they look similar to mine. Do you see something like this? Apple said it's corrupted RAM but I doubt it.

 

[christianronnel]

The issue is related to the discrete AMD graphics chip. Does your MBP have one? If it does then it's a related issue otherwise it may actually be software or RAM related.

 

[dgatwood]

@ksagomonyants

What I'm seeing in those photos—content shifted by a few pixels—is almost certainly a software bug. A hardware failure would almost always give you a block of absolute garbage. It's remotely possible that there's a defective register in the GPU, resulting in the low-order bit sometimes reading out a zero when it should read a one, which might cause that if that register were used as the source or destination address for a block move, but the odds of something like that happening are pretty small, and the odds of the chip being even halfway usable in that sort of state are remarkably close to zero. But you never know.

I would suggest that you install gfxCardStatus. This menu bar extra does two things:

• It provides notification when your system switches between the integrated GPU (Intel) and the discrete GPU (AMD or NVIDIA). If you experience problems at those transitions, it is almost guaranteed to be a software bug.
• It lets you pin the GPU mode to either the integrated GPU or the discrete GPU. This can tell you which GPU or GPU driver is broken.

Note that if you decide to pin the integrated GPU, you'll probably have to do so while Photoshop is not running, as most of those sorts of apps are on a blacklist that pins the discrete GPU automatically on launch.

With that said, let me be perfectly blunt: GPU drivers seem to almost invariably suck (and have sucked for as long as I can remember). For the first several months after I got my previous MacBook Pro, I experienced *constant* graphics corruption problems. After a little research, I determined that everyone was seeing them, and that it only occurred when using the integrated GPU, so I configured my system to use the discrete GPU exclusively. Eventually, after filing lots of bugs and doing lots of swearing, they finally got it straightened out, but it isn't uncommon. At all.


@christianronnel

As for the logic board failures, that's usually a thermal problem. Someone probably didn't assemble the computer with the right amount of thermal paste between the GPU and the heat sink, resulting in overheating when the GPU is under load. You might be able to work around the problem by using one of various tools that let you tweak the fan calibration.

If that doesn't help, then the GPU's heat is probably causing one of the BGA solder balls to not make proper contact with the logic board, which means that your logic board needs to be reflowed. I'd expect that to be cheaper than a replacement logic board, though if the problem is caused by using the wrong kind of solder and/or not keeping the chip cool enough, there's no guarantee it won't just fail again after another year.

But before you go to such extremes, call Apple anyway. Then press zero until you get a live human. Ask to speak to Customer Relations. Tell them that you've seen the thread, and that you're experiencing the same failure out-of-warranty. Ask them to make a warranty exception. They're the only team that can typically do that, so you might as well skip all the tier 1/tier 2 tech support and go straight to the people who can actually help.

 

[ksagomonyants]

@ksagomonyants

What I'm seeing in those photos—content shifted by a few pixels—is almost certainly a software bug. A hardware failure would almost always give you a block of absolute garbage. It's remotely possible that there's a defective register in the GPU, resulting in the low-order bit sometimes reading out a zero when it should read a one, which might cause that if that register were used as the source or destination address for a block move, but the odds of something like that happening are pretty small, and the odds of the chip being even halfway usable in that sort of state are remarkably close to zero. But you never know.

I would suggest that you install gfxCardStatus. This menu bar extra does two things:

• It provides notification when your system switches between the integrated GPU (Intel) and the discrete GPU (AMD or NVIDIA). If you experience problems at those transitions, it is almost guaranteed to be a software bug.
• It lets you pin the GPU mode to either the integrated GPU or the discrete GPU. This can tell you which GPU or GPU driver is broken.

Note that if you decide to pin the integrated GPU, you'll probably have to do so while Photoshop is not running, as most of those sorts of apps are on a blacklist that pins the discrete GPU automatically on launch.

With that said, let me be perfectly blunt: GPU drivers seem to almost invariably suck (and have sucked for as long as I can remember). For the first several months after I got my previous MacBook Pro, I experienced *constant* graphics corruption problems. After a little research, I determined that everyone was seeing them, and that it only occurred when using the integrated GPU, so I configured my system to use the discrete GPU exclusively. Eventually, after filing lots of bugs and doing lots of swearing, they finally got it straightened out, but it isn't uncommon. At all.


@christianronnel

As for the logic board failures, that's usually a thermal problem. Someone probably didn't assemble the computer with the right amount of thermal paste between the GPU and the heat sink, resulting in overheating when the GPU is under load. You might be able to work around the problem by using one of various tools that let you tweak the fan calibration.

If that doesn't help, then the GPU's heat is probably causing one of the BGA solder balls to not make proper contact with the logic board, which means that your logic board needs to be reflowed. I'd expect that to be cheaper than a replacement logic board, though if the problem is caused by using the wrong kind of solder and/or not keeping the chip cool enough, there's no guarantee it won't just fail again after another year.

But before you go to such extremes, call Apple anyway. Then press zero until you get a live human. Ask to speak to Customer Relations. Tell them that you've seen the thread, and that you're experiencing the same failure out-of-warranty. Ask them to make a warranty exception. They're the only team that can typically do that, so you might as well skip all the tier 1/tier 2 tech support and go straight to the people who can actually help.

Hi, thanks for your message. I did go to the Apple store, showed the tech guys the same pictures I posted here. They said it's logic board/graphic card failure (I hope I understood them correctly). I told them about this thread and the one on the apple.com website. They said they never heard about this issue with early 2011 models. The funny thing was they said, well you know people say all different kind of stuff on forums

So, I paid $310 to get the new logic board + 8gb RAM, although they said it usually costs ~$500-600. I didn't really care as I'm covered with Squaretrade warranty. I picked up the laptop today, so far it's working fine.

 

[christianronnel]

@christianronnel

As for the logic board failures, that's usually a thermal problem. Someone probably didn't assemble the computer with the right amount of thermal paste between the GPU and the heat sink, resulting in overheating when the GPU is under load. You might be able to work around the problem by using one of various tools that let you tweak the fan calibration.

If that doesn't help, then the GPU's heat is probably causing one of the BGA solder balls to not make proper contact with the logic board, which means that your logic board needs to be reflowed. I'd expect that to be cheaper than a replacement logic board, though if the problem is caused by using the wrong kind of solder and/or not keeping the chip cool enough, there's no guarantee it won't just fail again after another year.

But before you go to such extremes, call Apple anyway. Then press zero until you get a live human. Ask to speak to Customer Relations. Tell them that you've seen the thread, and that you're experiencing the same failure out-of-warranty. Ask them to make a warranty exception. They're the only team that can typically do that, so you might as well skip all the tier 1/tier 2 tech support and go straight to the people who can actually help.

Thanks for the info. I agree with you, the GPU needs to be reballed. But it's too late for mine, I need to find a place that can replace the GPU with a new one and actually solder the GPU properly to the logic board. The problem with Apple is that their service centers are not equipped for board level repairs. They just replace the parts (i.e., logic board). The problem with that is they just replace with refurb or new logic board with the same flaw. It will just have the same problem again after a short time.

 

[christianronnel]

Hi, thanks for your message. I did go to the Apple store, showed the tech guys the same pictures I posted here. They said it's logic board/graphic card failure (I hope I understood them correctly). I told them about this thread and the one on the apple.com website. They said they never heard about this issue with early 2011 models. The funny thing was they said, well you know people say all different kind of stuff on forums

So, I paid $310 to get the new logic board + 8gb RAM, although they said it usually costs ~$500-600. I didn't really care as I'm covered with Squaretrade warranty. I picked up the laptop today, so far it's working fine.

Good luck with your newly fixed computer, I hope that it lasts a lot longer than what other people had experienced. And of course they've never heard of it before, suuure (wink wink). Hopefully we'll get refunded the repair cost once there's a class action lawsuit. There is now a petition page on FB with 919 members so far. https://www.facebook.com/groups/2011mbp/

 

[sagittariansrock]

I had logic board failure with my early-2011 Macbook Pro (15", 2GHz Quad i7, AMD 6490), which didn't involve the graphics card (the audio would turn off and the control would be greyed out and unresponsive). Now, the one they put in had a similar graphics problem to what the OP describes crop up immediately. I joined the petition on FB.

 

[nWmR12]

While I don't have a that year model, Mine is a Mid MBP 2010 with a faulty GPU!
Mine was far worse, it would go black and then I would have to restart it, I couldn't put any workload with PS, Lr was okay for awhile.

I went last year to a store and they did their thing because I was tired of dealing with it. It was all paid for by them under some extended warranty which apparently would run out after 3 years after you bought it (they never asked). It said it was like $200+ to fix...

Although at least from my research this affected Mid 2010-Mid 2011 MBP.

 

[dgatwood]

Hi, thanks for your message. I did go to the Apple store, showed the tech guys the same pictures I posted here. They said it's logic board/graphic card failure (I hope I understood them correctly). I told them about this thread and the one on the apple.com website. They said they never heard about this issue with early 2011 models. The funny thing was they said, well you know people say all different kind of stuff on forums

No matter what, there will always be a small percentage of laptops whose GPUs fail because of solder ball failure. Keeping that number low is certainly a goal, but it is impossible to make the number be zero, given the temperature extremes involved. Assuming Apple has maintained the 70/30 laptop/desktop split that they've talked about in the past, then Apple sold somewhere around 12 million laptops in FY2011. 1,000 failures is less than one hundredth of one percent. If the failure rate really is at that order of magnitude, I wouldn't expect any sort of recall, because it's likely a fluke caused by inconsistent manufacturing.

With that said, assuming everyone who signed that petition is, in fact, an affected customer, then a thousand signatures is somewhat alarming. For every person who complains about a problem with a company, statistically speaking, there are usually anywhere from 10–100 affected people who didn't complain, with the average being about a 25:1 ratio. So 1,000 signatures could really mean that there are on the order of 25,000 affected 2011 MacBook Pro laptops that failed just outside of warranty, which is a different story altogether.

So, I paid $310 to get the new logic board + 8gb RAM, although they said it usually costs ~$500-600. I didn't really care as I'm covered with Squaretrade warranty. I picked up the laptop today, so far it's working fine.

For folks who don't have any warranty, the good news is that Apple recalls tend to be retroactive, so if it isn't a fluke, you'll probably get that money back.