Q: 2011 MacBook Pro and Discrete Graphics Card

Don't get your hopes up. I've been there. Taking it apart, cleaning everything and replacing the thermal paste helps. For a while. The. It's back to corruption.

And no, SSDs have nothing to do with it, nor do the drivers or firmware.

(Oh, it has nothing to do with 10.8.5 either.)

<Edited By Host>

One more thing. Reballing only helps for a while. I've reflowed - not reballed - my GPU myself. It worked fine for about a month. Today i'm back on square one.

One odd thing i've noticed - when you get to the point where you can't boot anymore (corrupt boot screen or stuck on gray screen with fans on full blast), shut your MBP down, wait a second, yank out the power cord, then boot on battery. You'll magically be able to boot fine again, with no corruption, even if you plug the external power back again. Until some random moment later, when it'll go bad again.

Yes. Everyone should realize that this problem is totally caused by gpu failure, neigher firmware nor operating system. If you boot into other systems, for example, Windows, you will understand a hardeware problem crashes your system as long as the gpu is activated, no matter which operating system and version you are using. Everytime I boot into Windows 8 (not safe mode) it will show me a blue screen with thin black vertical lines.

Its apple's fault we are trying to make them fix it check this pages: (the first is posted by Dreamson)

https://discussions.apple.com/thread/4637833?start=135&tstart=0

<Link Edited By Host>

Hi Orcinus

Your method seems to be specific to your mac case. It doesn't work for me

It's not a "method". Nor a fix, it sometimes helps, before the issue degrades beyond some "nothing helps" point, after which the only thing that will make your machine work is a reball/reflow, and that's only for a while again.

I'm mentioning it not as a fix, but as a symptom - booting on battery appears to *sometimes* force the machine to boot on integrated (Intel) GPU without trying to initialize the discrete GPU. At least on Mavericks.

Did you create the survey?

Can whom ever did, update it to included the late 2011 models and 6770m cards? Right now it only focuses on the early models, but the late models are experiences this too.

Seems to be true - crashing again this morning.

Also seems odd to me that you can improve the situation since a hardware issue should be "fixed" and continuously getting worse rather than getting a little better for a while.

Looks like fixing my 19 month old laptop will be my early Christmas present this year.

<Edited By Host>

I have same issues as everyone else. Took it into Apple store and they said $310 flat rate fix. I should get it back in a couple days and will post an update. This was the first and last time purchashing a Mac computer.

Its not my creatin but i'll check who start it and tell them to include the all 2011 becuase looks its the same problem

I'm beginning to think that the GPU crash happens during the switchover from integrated to discrete and back and isn't strictly related to discrete GPU being defective. I.e. that the controller responsible for switchover is messed up / faulty, not the actual GPU.

Consider the following:

- forcing it from integrated to discrete also sometimes causes a crash / glitchout.

- removing AMD6000 kext sometimes also causes glitching during boot

- removing AMD6000, AMDX3000 and other related kexts prevents glitching, but not because the discrete GPU is not being used anymore, but because you've removed other supporting drivers (including those responsible for the switch) as well

- i am writing this with discrete GPU forced on, despite the fact i've had major glitching a few minutes ago, with dynamic switching on

In fact, i'm now semi-convinced the issue might be with the GPU power supply and related power management. Power supply circuitry often has non-solid state capacitors (even though most are solid state, i.e. tantalum). Non solid state (electrolytic) capacitors go bad. And it's not unheard of whole batches being factory-botched and going bad at the same time.

In fact, famously, a whole series of Antec branded desktop PC PSUs once went bad for a *crapload* of people nearly simultaneously, due to a bad set of "counterfeit" (knockoff) chinese caps.

Orcinus wrote:

 

I'm beginning to think that the GPU crash happens during the switchover from integrated to discrete and back and isn't strictly related to discrete GPU being defective. I.e. that the controller responsible for switchover is messed up / faulty, not the actual GPU.

 

You may be right. But how do you explain the failure mode where running on the discrete GPU will suddenly "glitch" and give visible visual problems after several minutes or hours?

When I described my problem originally to a coworkers who's a hardware hacker whose hobby is resurrecting old dead Macs, he immediately said, "Sounds like a bad GPU voltage regulator." So he'd agree with your assertion that the problem is related to power.

But at the end of the day, it doesn't really matter whether it's defective voltage regulators or counterfeit capacitors or evil ghosts. What matters is that Apple designed, built, and sold a product with an unusually low lifespan, and they should do better.

Of course it's Apple's fault, that's not even debatable.

However, pinpointing the cause might help - both us, users, if it leads to some kind of a stop-gap measure that makes it possible to use our machines while we wait for Apple to react, and Apple, if they keep tabs on discussions (fat chance, i know, but better than no chance).

As to sudden glitching after hours of work, there are a few possible explanations:

- voltage fluctuation due to some other factor (high CPU load combined with disk spinning up, for example)

- quick "drop" back to integrated GPU for some reason (i've had my MBP "flicker" between integrated and discrete on a few occasions, which wasn't detected by gfxCardStatus, but was visible by the shift in gamma)

- discrete GPU switching to a different power state (or clocking up beyond some "magical" threshold, due to increased load)

The reason i'm suspecting capacitors (or -or) is because heat tends to fix the issue *for a while* (and by heat i mean reballing or reflowing). Dodgy caps tend to respond to drastic shifts in ambient temperature.

Orcinus wrote:

 

 

Of course it's Apple's fault, that's not even debatable.

 

 

Maybe. Apple bought these GPUs from AMD. There are usually requirements that have to be met for Apple to accept these parts. Apple has final system responsibility, but getting bad parts is getting bad parts.

My guess is Apple and AMD are looking at the data and trying to see if it meets Apple's requirements, and if not, assign responsibility. Is this a systematic problem? A manufacturing problem? Root cause is critical for assigning blame.

Neither Apple or AMD want to pay for a recall/repair program if they don't have to, so the blame game becomes important. If Apple shows the part doesn't meet requirements, then AMD gets a majority of the cost. If AMD shows that Apple is misuing the part (not enough cooling, current, whatever), then they can claim it wasn't their fault.

If Apple sold 10,000,000 Macbook Pros for 2011, then 200 bad units works out to be about 20 ppm (parts per million). In the auto industry, that would be high, though not crazy high. Single digits are usually normal targets.

This number would indicate how big of a problem this issue is. That's going to be Apple's first step. I'm not sure if there's enough of a blip for a two-year-old product yet. A nice class-action lawsuit would catch their attention.

I'm not defending Apple, and if this causes people to stop buying Apple products, I don't really care. But if tires start exploding on Fords, it's not necessarily Ford's fault, unless they were using the tires inappropriately. Same thing here. I don't know the true issue here, but I'm looking hard at AMD.