MacBook Pro (Retina, Mid 2012) freeze with Yosemite

I have a Macbook Pro Retina (mid 2012) 2,3ghz, 8GB Memory, 256Gb SSD.

Last July, I started to have a lot of kernel panics. Long story short, I went to a genius, they diagnosed a dead GPU and replaced the logic board under warranty + the entire screen for a faulty hinge.

Since then my computer was doing better even If I still had some random freezes (one a week on average). At that time I thought it was because of my messy OS X system but, two weeks ago, I decided to take care of this issue by doing a clean install.

However, less than 10 days after, I got another freeze. To me, this was a strong indication that my hardware was somewhat faulty.

I went to the genius last Thursday and luckily, I find out that It was my last day of warranty. They suggested to run some hardware tests and call me with the results.

Last Saturday, they called me and said that my hardware passed all the tests. Next step would be to do a clean install to make sure the problem could be reproduced on a clean system. Even though this is what already did, I told them to do it and that I would pick up my computer on Monday.

Today, I went to the genius bar and started to chat with the genius. I showed him this thread about the issue but he wasn't open to discuss about it. He told me the only way to get addtional help would be to show some proofs (on videos or at the store) that my issue could be replicated. I decided to stick around at the Apple Store to restore my datas and "luckily", during the process, my computer had a freeze.

The genius looked at it, agreed that something was faulty and decided to order parts for replacement. Unfortunately, they don't have them in stock so I need to wait a couple days but early next week, I should have the logic board and SSD replaced under warranty. Hopefully this will fix the issue.

On a side note, I agree that the freezes appears when the graphic card is switching from the Intel to the Nvidia. Why? Because, at work, I have an external display (which requires the NVIDIA) and my freezes only appears when I'm not at the office working with a single screen.

My MBP is now suffering from graphics distortion when switching from the internal card to the NVIDIA card. It did NOT have this problem until the 10.10.2 update. Now half of the time I open any program (iPhoto, iTunes, Safari with video...) that uses any graphics the screen distorts. It seems like the background is dimming and it's cutting out half the resolution. Once I exit out of the application the display returns to normal. Restarting the computer does not help. I am also still having intermittent WiFi issues with TimeCapsule and OSx and iOS too. Hopefully these problems get fixed soon as 10.10.x and iOS 8.x have both been a Windows like frustration.

As has been previously brought up, a logic board replacement seems to either reduce or solve the issue. In my case, it seemed to mitigate the issue. I spoke with a senior support specialist who didn't really offer much help today. Eventually she arranged another logic board replacement. She did mention that after 3 major repairs, it was possible the machine could be entirely replaced. I'm hoping this finally solves the issue, but seeing as even under 10.10.2, the issue still persists after months I'm loosing hope. The specialist tried to get me to again reinstall OS X with no 3rd party software, I've already done this before with the crash still occurring. It also isn't fair to expect a user to reinstall OS X multiple times within the past few months, especially students or professionals that rely on their machines. Hopefully a firmware update, OS X patch, or a combination comes soon to resolve the issue 😕

Macbook pro retina 15 mid-2012 display problem : solution :)

Read my third post 🙂

I don't see, in my case at least, that this is a hardware problem.

I got my MBP just after they launched and never had a single issue with it. As soon as I upgraded to the Yosemite Developer Preview it started hanging and has been that way ever since.

Agreed, this is a Software problem that occures on most of the MBP with dual graphic cards running yosemite. A major swich bug.

NO hardware problem, do not use time on services that dont work (read in the other comments), everyone that have replaced parts on their MBP still have the same problems!!!

Solution: switch off the Automatic graphic-card switcher! Wait for a newer/updated version of yosemite. You'll have crapy battery life, so plug in the power cord, dont vine, nothing else to do. Or downgrade to osx 10.9.

Agreed. Let Apple know:

Apple - Mac OS X - Feedback

For three months, I’ve been in the same boat as many of you—2012 Retina Macbook Pro with increasing GPU panics after installing Yosemite—but I’ve held back posting until I could provide the full results of my testing. Details below, but first my high level conclusions.

This problem is due to a progressive hardware failure of the Nvidia GPU or its interface to the logic board. Given that the problem appears to affect only some MacBookPro10,1 laptops, and has been also reported on 2013 and 2014 MacBook Pros, I think it’s likely that the root cause is a manufacturing defect with the Nvidia GPU or the MacBook Pro itself. But I think it’s also very possible that the issue only manifests itself and gets progressively worse when the MacBook Pro exceeds some thermal limit.

While I know it's tempting to blame this problem on Yosemite given the time frame—I was there myself—from my evidence it’s very clear this is not a problem with Yosemite. I rolled back to Mavericks and the problem persisted. With my new logic board, Yosemite is running fine.

I should also mention that I was able to ultimately fix the problem by replacing the logic board. Like many people here, an Apple Store rejected my unit for warranty repair claiming liquid contact, despite the fact that I know this has never happened. I spent $50 to have an Independent Apple Service Provider clean out dust and replace the thermal paste (per vadeskoc, although this did not fix the problem), and in the process gathered evidence that refuted the Apple Store’s claim. With the photographic evidence I provided (http://imgur.com/a/cHDOS), I went back to the Apple Store and they replaced the logic board, both fans, and heatsink under warranty.

• I bought the MacBook Pro in Jan 2013. Just a year later, while still running Mavericks, I experienced what I believe was likely the same issue. Apple fixed it by replacing the logic board and heatsink. Here’s an excerpt from the problem description filed by the Apple Genius: “Customer reports that during use screen will go black. He will have to do a hard shut down … Customer had a error report that showed GPU panics”
• After this, my MacBook Pro worked flawlessly until mid-Nov 2014, a month after installing Yosemite. Safari rendering was wonky, and iPhoto would occasionally bomb with GPU panics. By mid-Dec 2014, this had progressed to frequent GPU panics in other apps, and after each GPU panic I’d have to wait 5-10 minutes before the laptop would boot properly again.
• I also determined that I could consistently reproduce the problem within 5 minutes by running the “furry donut” stress test in the popular “GPU Test” app. And I determined that I could prevent the problem completely by using gfxCardStatus to only use the integrated Intel GPU. In other words, I don’t believe the core issue is caused by switching between GPUs, although that could be a second, related failure mode.
• Using Mac Fan Control, I determined that manually boosting my fan speeds to max did not fix the problem. I did notice one anomaly that I think may be important: the temperature shown by the “GPU diode” was all over the place, especially as I started running the Nvidia GPU: I would see fluctuations up and down of 15°C in a single second. In contrast, on my newly fixed logic board, I don’t see any such fluctuations: when I run the stress test, it starts down around 40°C and then rises steadily to 90°C, at which point the fans are on full and bring it down to a steady 80°C.
• Like many of you, I strongly suspected this might be a software issue, so I rolled back and did a clean install of Mavericks (completely clean, just built-in apps and no custom data or additions). The problem persisted. If this is a software issue as many are still claiming, it also exists in Mavericks. If you don't believe, I recommend you try the same.
• I also suspected that the problem could be caused by the new EFI boot ROM surreptitiously installed with Yosemite (MBP101.00EE.B05). Apple’s EFI boot ROM page shows that the MacBookPro10,1 should be using MBP101.00EE.B02. So I rolled back to this boot ROM. The problem persisted. (I reported to Apple three months ago not only that this page was out of date, but that the B03, B05, and (now) B07 boot ROM updates for the MacBookPro10,1 are not available online anywhere, and they still haven't addressed this)
• Having exhausted likely software options, I attempted to eliminate thermal issues as the culprit by implementing the vadeskoc fix: using an IASP (so the MacBook Pro remained under warranty), I had them clean out the dust and replace the thermal paste over the GPU and CPU. The Mac ran much cooler, but the problem persisted. I agree that the thermal paste application—undoubtedly performed during the 1/2014 repair—looked sloppy; in fact, what Apple was claiming as liquid contact appeared to be thermal paste.
• When Apple returned my repaired laptop with new logic board, it was still running my Mavericks install, although the new logic board had the B03 boot ROM. First thing, I set gfxCardStatus to run the Nvidia-only and then ran the GPU stress test for an hour, no failures. I then upgraded to Yosemite, which also updated the boot ROM to the new B07 version, and repeated this test; no failures.

The fact that the GPU diode temperature had apparently failed on my unit is very suspicious, especially since it didn’t show up in Apple’s tests. While this could be coincidence or a symptom, what I’m wondering now is whether it could be a primary or secondary cause of the problem. If this temperature sensor was failing, as it appears, it’s possible that the GPU is getting too hot, either damaging the GPU itself or worsening a manufacturing defect in the GPU interface. If so, this problem would be exacerbated by everything reported: dust in the case, bad thermal paste job, and increased Nvidia GPU utilization in Yosemite.

I also wonder why Apple is letting the Nvidia GPU get all the way to 90°C on my repaired unit before bringing it down to 80°C; while this situation only lasts for less than a minute, it seems like they should be cranking the fans up higher faster to avoid thermally stressing this area, give the problems many of us are having.

I also wonder if the fact that a number of us shelled out $3300 for this laptop is relevant, since it means we bought custom units (16 GB RAM, 768 GB SSD, fast processor), which could mean we all have logic boards (and even replacements) from a particular, relatively small batch that could have had defects not seen in the standard SKUs. And, of course, it also means our laptops run hotter.

Finally, because I don’t think Apple has resolved this issue yet—although I suspect they are aware of it—I suspect I may be back in the same situation a year or less from now. What I’m going to do to try to prevent this is keep an eye on the GPU diode temperature sensor to see if it starts failing. I’m also going to buy a pentalobe screwdriver so I can open the bottom case up every month and blow out the dust.

Not the freezing problem this thread refers to.

+1 on my end, it is happened exactly on graphics card switching

It is definitely Yosemite issue, I met this issue in my MBP 15" 2012 only after installing Yosemite.

Mavericks, Lion, Mountain Lion - doesn't have had this issue

This thread is filled with people experiencing mid-2012 Retina MacBook Pro lockups due to GPU panics after installing Yosemite, which is exactly what I was experiencing. If you think their experience differs from mine in some key way, you need to be a little more explicit. Otherwise, it sounds to me like exactly the same problem.

Thanks for the detailed report, but you're describing a different problem. I get a freeze where nothing on the screen changes, except the mouse pointer still moves. My screen never goes black. I can reboot immediately and it happens even on a cold machine.

That's the difference between correlation and causation. The correlation is that many of us only experienced this after installing Yosemite. The only way to establish whether it's causation is to do what I took the time to do, which is to roll back to Mavericks and try to reproduce the problem by exercising the Nvidia GPU. Until I did that, I also believed and hoped that it was a Yosemite issue (despite some evidence to the contrary). It's not. You can argue otherwise, but until you test that out, you really don't know.

What the "software only" advocates in this thread have yet to explain is:

• Why does the problem get progressively worse over time, as reported by quite a few people on this thread? How would a software-only problem cause MacBook Pros to get more and more GPU panics over time? Or fail to restart even before Yosemite starts loading?
• Why does the problem go away entirely or get better (at least temporarily) by replacing the logic board despite running the same software, again as reported by a number of people on this thread? If the software hasn't changed, why the variation in behavior following a hardware change?
• How do homegrown mechanical fixes, like those from vadeskoc and nrj45, make the problem disappear if it's a software-only problem?
• If it's purely a software issue, why isn't the problem more widespread? While there naturally are hundreds or thousands of people having this problem for every person posting on this thread, that's still only a fraction of the people who bought this model. There has to be some hardware-level variation to explain why all users of this model (or other models using this GPU) aren't experiencing it.

What are you seeing to determine it's a GPU panic? There's zero evidence of this in any of my logs because no kernel or gpu panic is logged and there's no corruption anywhere - the system simply halts and reboots.