You can make a difference in the Apple Support Community!
When you sign up with your Apple Account, you can provide valuable feedback to other community members by upvoting helpful replies and User Tips.
When you sign up with your Apple Account, you can provide valuable feedback to other community members by upvoting helpful replies and User Tips.
After installing the OS X Yosemite my MBP Retina starts freezing due to graphic problems. The only option is to restart the Mac
How to find the problem and solve it?
MacBook Pro (Retina, Mid 2012), OS X Yosemite (10.10)
My 2012 non-retina MBP is on repair since last week, I was told I need a new logic board. This was before the repair program was announced but the non-retinas are not included so I probably will have to pay for something that doesn't fix the problem anyhow.
While waiting for the repair I'm using and older 2010 MBP, much less RAM, less CPU (2 cores), less disk but with Mavericks instead of Yosemite. The sad thing is that it in almost all ways feels like a much faster machine, much snappier feel.
Upgrading to Yosemite is something that I in many ways really regret. I've also had dns-problems with my MacPro (that otherwise could run for years) and had to reboot at several occasions (although the dns-issue should be solved by the latest Yosemite-update). The MacPro also feels really sluggish.
"Dont bother, its not hardware defect. Software. The repair program does`t do anything."
Nooooooooooooooo!
Folks, please, don't believe comments like this. Look, I'm as infuriated about this issue as anyone. Like all/most of you I paid $3k for my MBPr and expect the best performing laptop on the planet for the most expensive laptop on the planet. This issue was a nightmare for me....when it was an issue. But after a logic board replacement I'm 99% crash free. I think I had one GPU crash (can't remember if it was a freeze or automatic reboot) since the repair in late December. That's two full months in which I've been using the machine identically to how I used it before. I have posted previously about my thoughts on this being a combination of hardware and software. I'll reiterate what I see to be hard facts here that support this idea and that support the fact that you're a moron to not take advantage of this and get a new motherboard.
First, yes, I admit, these issues didn't come up before Yosemite so that leads us to the issue being only in software. BUT...if it were only a software issue, everyone using Yosemite with our MBPs would have this issue and it would be a much bigger deal. Sorry, but 36 pages in this thread are a tiny fraction of the worldwide MBP community. Second, my problems went away with a motherboard replacement. I didn't reformat after the repair. Same software. As I said, I think I had one crash a few weeks ago, but that's compared to regular, countless crashes before the motherboard replacement. If this is a software issue, why the did the hardware replacement work? Because it's not. As I have said before, this is a hardware issue that only manifests itself on Yosemite. That is not unprecedented. If you are having this issue, go get the motherboard replaced. Now that Apple has admitted it, I'm sure they have cleaned up their existing stock of replacement boards and people who have had replacements in the past without resolving the issue will get proper, working replacements.
Lastly, I'll say, this doesn't "restore my faith" in Apple as some have said. I'm still furious. I still have wifi issues with Yosemite. And this admission of the hardware defect should have been admitted months ago. And the folks with the other video issues that affect older MBPs (I have a 17" MBP that was repaired before they admitted this because it was barely still in the the 3-year Applecare term) have suffered for far longer. Inexcusable. Yes they did the right thing, but it's too late to be completely forgiven. These things cost two to six times more than PC laptops. Unacceptable!
Has anyone experiencing the symptom(s) described in this thread updated their MBP or MBPr's NVIDIA driver after confirming that their installed version was an older one?
If so, what happened after updating?
Here's the link to NVIDIA's download page for the latest and (hopefully) greatest driver.
http://www.nvidia.com/object/macosx-cuda-6.5.46-driver.html
New Release 6.5.46
Supported MAC OS X
10.10.x
10.9.x
10.8.x
Version: |
6.5.46 |
|
Release Date: |
2015.01.28 |
|
Operating System: |
Mac OS |
|
Language: |
English (U.S.) |
|
File Size: |
53.7 MB |
|
Supports all NVIDIA products available on Mac HW.
Note: this driver does not support the older generation GPUs with SM1.x. Please download the equivalent CUDA driver 6.5.45 which supports SM1.x.
This is software for CUDA (http://www.nvidia.com/object/cuda_home_new.html), NVIDIA's programming language that lets you run general-purpose application on their GPUs. I don't believe it changes the GPU driver that's installed in MacOS.
Are you sure that this actually installs a new graphic card driver?
OpenGL reports the same GL_VERSION (4.1 NVIDIA-10.2.1 310.41.15f01) as before running the CUDA installer.
Yours,
Stefan
Still a software issue.!
I do not write things just for fun. Software and Hardware are connected to each other, if there is a bug or issue in the software (like in Yosemite with the freezing of the desktop), this will be reflected in hardware not working properly.
First, yes, I admit, these issues didn't come up before Yosemite so that leads us to the issue being only in software. BUT...if it were only a software issue, everyone using Yosemite with our MBPs would have this issue and it would be a much bigger deal.
You do actually mention that your problem began after upgrading to Yosemite, so the new software had some issues with the compatibility of the hardware. So not hardware defect from factory, but software bug. And NO, there would not be more people discussing it here, thats because the issue is just for Macbook pros with dual graphic cards, (Intel HD graphics and Nvidia Geforce), and not all macbook pros come with these specifications (just the high-end ones.)
I dont understand how people can think its a hardware defect from factory, its so weird. I have mentioned it before. If there was a hardware defect from factory, you as an consumer would have noticed the issue the first week of using the brand new 3K MacBook Pro Retina Mid 2012, not 3 years later after you upgraded to a new buggy OS X software.
On scenario:
1. Lets say in the year 2012 - 2013, apple made 2 batches of the macbook pro with dual graphic cards.
2. Nvidia made also to batches of the Nvidia GeForce GT 650M.
a. The first batch of Nvidia GeForece GT 650M v.1, no bugs or nothing.
b. The second batch of Nvidia GeForece GT 650M v.2, no bugs or nothing, just some new coding or components (needs some modifications i OS X to
integrate just right)
2. The fist batch of macbook pros retina was equiped with v,1 of Nvidia cards
3. The second batch of macbook pros retina was equiped with v.2 of Nvidia cards, and some new patch of OS X software was released to integrate the new components.)
4. In 2013 newer macbooks was released with newer Nvidia cards and Intel Integrated cards, the GT 650M v.1 and v.2 is history.
5. Modification in software must be done to integrate the new cards and still support older cards like GT 650M v.1 / v.2.
6. In 2014 apple are working on a new platform Yosemite, a new recoded operating system. They will support all macs down to the 2011-2012 versions, so coding to integrate all versions of graphic cards and external components must be done. The developer have made a mistake early on, in the core/kernel function that supports the GeForce GT 650M v.1 - Intel HD Graphics switching.
7. They Release Yosemite, the customers with MacBook pros retina 2012 "first batch" with "first batch of Nvidia GT 650M v.1" notices that they get freezes, but the people with second batch with Nvidia GT 650M v.2 does not notice it.
The case above is a real life scenario, and it shows how a glitch in coding (software) have an affect on hardware. There is a silver line between software and hardware. So if a compony or developer have made an error in integration process, one or more products with the spesific types of hardware will fail (NOT problems with hardware from factory).
If they run a diagnostic and no problems is shown, you'll get you mac back, no replacements parts. My mac passed all tests, got it back nothing replaces.
If they send it to repair and change the logic board with the same logic board produced when you MAC was bought (Batch 1 Macbook, Batch 1 of Nvidia Geforce GT 650M v.1), the problem will occur.
BUT, if you get a newer logic board and graphic card,from ex: "batch 2 macbook pro retina 2013, batch 2 of geforce gt 650M v.2" this may solve the problem. If this is the case, apple development, have made a mistake down in the Core/kernel that is so expensive to correct (needs many man hours and, total rewrite of code) so the easiest way is to replace some parts. (Still a software problem, no hardware defect).
I am still waiting for an update of Yosemite.
I defer to those who are more knowledgeable than me, but here's a copy-paste of info from the linked page in my previous post.
Supports all NVIDIA products available on Mac HW.
Note: this driver does not support the older generation GPUs with SM1.x. Please download the equivalent CUDA driver 6.5.45 which supports SM1.x.
How about the downloaded file from this page? Is this a driver?
http://www.nvidia.com/download/driverResults.aspx/81365/en-us
CUDA Application Support:
In order to run Mac OS X Applications that leverage the CUDA architecture of certain NVIDIA graphics cards, users will need to download and install the 6.5.46 driver for Mac located here.
New in Release 343.02.02f01:
My questions remain…
Has anyone in this thread who has experienced the symptom(s) updated their NVIDIA driver from an older one to a newer one?
If so, what was the outcome?
I tried.
It tells me after opening: This computer is not compatible with NVIDIA web driver.
I have MPB late 2014,16GB, Nvidia GT750M
I'm a software engineer. I write drivers for a living. There is so much wrong in mariusa's last post that I felt the need to dive back in here one more time. Good luck to everyone.
You do actually mention that your problem began after upgrading to Yosemite, so the new software had some issues with the compatibility of the hardware. So not hardware defect from factory, but software bug.
No. It's possible the problems some of the people on this thread are reporting are due to a bug in Yosemite, but it's also possible that they have other causes. To the best of my knowledge, no one here has actually demonstrated that the bug is in Yosemite. You are relying on correlation, but you haven't demonstrated causation. I find this frustrating, because while it may take some time, it's pretty simple to do, and it may actually provide information that Apple needs to address the issue. Here's how:
If you follow these steps, there are three possible outcomes:
I dont understand how people can think its a hardware defect from factory, its so weird. I have mentioned it before. If there was a hardware defect from factory, you as an consumer would have noticed the issue the first week of using the brand new 3K MacBook Pro Retina Mid 2012, not 3 years later after you upgraded to a new buggy OS X software.
Wrong. Manufacturing defects of all sorts can manifest themselves months or years after the fact. As can design defects. Hardware can work perfectly when you buy it, it can work perfectly for months, and then it can stop working properly. That's why Apple warranties its Macs for 1 year. And that's why many of us buy AppleCare to extend that warranty, especially for laptops.
Besides the general case, as I've already documented in my particular case, I started having GPU Panics and resets in early 2014, a year after I bought my laptop, which were completely fixed (for 9 months) by a logic board replacement. And then I started having them again—with similar symptoms—and they were again fixed by a logic board replacement two weeks ago. In both cases, I could readily reproduce the problem before the logic board swap, but not do so once fixed. No change to software was made either time.
I'm not suggesting that any one person's particular problem has to be caused by hardware, but the notion that it's not possible to see hardware issues develop over time due to original manufacturing or design defects is nonsense. In my particular case, I still suspect a malfunctioning GPU temperature sensor resulted in thermal damage to my board over time, something that was exacerbated by the increased GPU usage in Yosemite.
And NO, there would not be more people discussing it here, thats because the issue is just for Macbook pros with dual graphic cards, (Intel HD graphics and Nvidia Geforce), and not all macbook pros come with these specifications (just the high-end ones.)
Apple sells about 4 million Macs a year, and a sizeable percentage of those are MacBook Pros. Both the Retina and non-Retina mid-2012 15" MacBook Pro used the NVIDIA GeForce GT 650M GPU, and Apple didn't rev these lines again until late 2013, so almost a year and a half. It seems reasonable to assume that at least 1 million of the mid-2012 15" MacBook Pros were sold. So yes, if this was a problem with all mid-2012 15" MacBook Pros because of that Nvidia chip, we should be seeing a lot more people discussing the issue.
Furthermore, if you want to say that the issue I experienced is different than what is being discussed on this thread—as you have done—then that means that I am definitely not and have never experienced the issue you are describing, despite using the same model and GPU running on Yosemite. Specifically, I use my MacBook Pro for graphics-intensive work and external video all the time, but I've never seen a situation where the screen locked and the cursor kept moving.
The case above is a real life scenario, and it shows how a glitch in coding (software) have an affect on hardware. There is a silver line between software and hardware. So if a compony or developer have made an error in integration process, one or more products with the spesific types of hardware will fail (NOT problems with hardware from factory).
I think everyone understands that a glitch in coding can affect the operation of hardware. But it's also true that a defective installation can affect the operation of hardware. And a non-bug change in software behavior can expose existing design or manufacturing issues in hardware. To claim this is due to a Yosemite bug that affects all of the mid-2012 15" MacBook Pros models, you have to explain why most of 1 million users (myself included) aren't seeing it. What is different about your computers -- in software or hardware or usage -- that is causing you to see the problem?
And if your response is that it may be particular batches of Nvidia GPUs that have the problem, so only some mid-2012 MacBook Pro logic boards have an incompatibility with Yosemite, then it's very possible that swapping your logic board _could_ fix the problem.
I am still waiting for an update of Yosemite.
Since you haven't demonstrated that this is a bug in Yosemite, you may have a long wait. Personally, if I _really_ believed this was a software issue, and if it was clear that most mid-2012 15" MacBook Pros are probably not experiencing the problem, then I would start with figuring out what might be different about my software installation than the majority. I'd follow the steps above to see if my specific Yosemite installation was the culprit, and then move on to testing with Mavericks. Which incidentally, is exactly what I did, and for my computer, I demonstrated the GPU Panics I was experiencing persisted in Mavericks and were indeed caused by hardware.
You start by demonstrating that you can reproduce the bug reliably in Yosemite without using an Apple app like Safari or iPhoto etc. For instance, if your problem is caused by graphics switching, you should be able to use gfxcardstatus to switch back and forth and show that you can reliably demonstrate the problem 10 times of 10 within X minutes. Why is this important? Because in the next steps, you need to be able to reliably show that you can or cannot reproduce the problem with the OS as the only piece of Apple software you are using, in case Apple made other changes to their apps between Mavericks and Yosemite that could affect your ability to reproduce the problem.
I did it in Genius Bar. I reproduced the problem on their clean system, which they booted from their device (by switching on/off Automatic graphic switcher)
I could not reproduce the same bug on clean Mavericks that was booted from their device.
I've changed my mainboard and bug is still there. So, i think Mariusa is right
This is really confusing. I have two user accounts on my MacBook Pro (15" Retina, Early 2013). The first account, which I mainly use, is extremely UI laggy and produces freezes sometimes.
Today I used the other account the whole day and I experienced a much more stable and fluid UI and there were no freezes. I know this is only a one day experience but maybe the problems could be caused by certain user-settings, -trash, -apps or something similar.
Maybe some of you can try creating a new user account and see what happens?
I tried according to your advice to create additional account. And i tried to freeze MBP by switching ON/OFF Automatic Graphic switcher.
It needed more time than usual. May be about 100 switching. But i could do this. So, this is not a solution.
frookt wrote:
I did it in Genius Bar. I reproduced the problem on their clean system, which they booted from their device (by switching on/off Automatic graphic switcher)
I could not reproduce the same bug on clean Mavericks that was booted from their device.
I've changed my mainboard and bug is still there. So, i think Mariusa is right
I agree completely that if I had the experience you had at the Genius Bar, I'd also think the problem was a Yosemite-related issue, or at least a Yosemite-related issue that was specific to my computer. And if I got my logic board swapped and I still had the problem, I'd start to wonder if the problem could be caused by something not on the logic board, or suspect that the problem was more widespread.
But I don't think the problem is that widespread. For instance, I just switched the Automatic graphic switcher on & off 200 times on my MacBook, and I had no GPU Panics nor did I see the screen lock while the mouse kept moving (nor have I ever seen that). Since we're both using mid-2012 Retina MacBook Pros with Yosemite (I'm currently running 10.10.2), we have to explain the difference in incidence by something beyond "a bug in Yosemite".
If the difference was due to your software installation, we'd expect to see that disappear when you booted off of the clean Yosemite disk at the Apple Store (which it didn't). And if the difference due to something with your logic board, we'd expect it might disappear when you got a new logic board (which it didn't). So what is the difference?
And for that matter, why would you see any difference in behavior at all running Yosemite under a different user, as kungfunefar and you just reported. If this is a low-level GPU issue with Yosemite, I don't see how which user you are running should make any difference in the incidence rate. But I think exploring these sorts of avenues is exactly what's going to lead to a solution.
BTW, I am curious about one thing. Back in January, you reported that even when booting clean on Mavericks, in some cases you still saw the screen freeze for a couple of seconds, but not completely like you were seeing on Yosemite. Is that correct? Did you remember having that problem on Mavericks before you upgraded to Yosemite?
I read this thread a few days ago. Have been getting crashes with a 10.10.2 system (late 2013 RMBP) for the last 4 weeks or so. Frequency of crashes is about 5-10 a day. Beachball would appear randomly on somewhat graphic intensive sites (like pages with flash animations). Only recourse was to do a hard shutdown. Mouse continued to work. Everything else, unresponsive.
The only oddity in my situation is that I can't remember it ever happening on the laptop itself. It would only happen when connected to 2 thunderbolt displays. Since connection to Thunderbolt displays automatically implies use of the discrete Nvidia GPU, therefore the "switching" of the GPUs is not the likely cause of my problem (it could be, I don't know).
Took it to an authorized service center. Here's the diagnostic report from them:
MRI passed
Storage Diagnostic passed
ASD EFI passed
ASD OS failed on multiple video controller testsThey concluded that the Logic Board was faulty. Supposed to get a replacement. YMMV.
I'm waiting to hear back from them when the new Logic Board is installed (supposedly will take 2 days to arrive).
My late 2013 retina MBP is not part of the GPU recall, BTW.
BTW, I am curious about one thing. Back in January, you reported that even when booting clean on Mavericks, in some cases you still saw the screen freeze for a couple of seconds, but not completely like you were seeing on Yosemite. Is that correct? Did you remember having that problem on Mavericks before you upgraded to Yosemite?
Yes, i had freezes for a couple seconds. I think driver interpretation of this freezes is the reason of the problem. Yosemite thinks that this is GPU panic. Mavericks increase time of some calculations, but as a result doesn't give an error.
I never had even 1 full freeze (where i need reboot) of the MBP on Mavericks.
But I don't think the problem is that widespread. For instance, I just switched the Automatic graphic switcher on & off 200 times on my MacBook, and I had no GPU Panics nor did I see the screen lock while the mouse kept moving (nor have I ever seen that). Since we're both using mid-2012 Retina MacBook Pros with Yosemite (I'm currently running 10.10.2), we have to explain the difference in incidence by something beyond "a bug in Yosemite"
Please be sure that changing is really happens . Use gfxCardStatus. Cause in other way there could be processes that use NVidia (like Photoshop or some parts of Chrome) and switching this switcher makes nothing.
MacBook Pro (Retina, Mid 2012) freeze with Yosemite