Q: 2011 MacBook Pro and Discrete Graphics Card

It's not strictly the quality of the paste that's the main problem with MacBook Pro's, these 2011's especially.  Apples stock paste is a Shinetsu type, a pretty good one compared say to Sony and Dell. It's the massive over application of the material and the heatsink surface plate being not completely smooth that gives rise to excessive heat.  A little bit of metal polish to smooth down the surface of plate before reseating the heatsink making it shiny helps a lot to keep these models cooler, and checking using a metal rule and magnifying glass to make sure the heatsink itself isn't slightly bent and twisted. I've found many have needed bending slightly to make them level! With the other models of unibody which don't generate as much heat it may not be as crucial to do but with these Sandy Bridge 2011 models I firmly believe it is!

It's not strictly the quality of the paste that's the main problem with MacBook Pro's, these 2011's especially.

You can lament as much as you like about the quality or the quantity of the applied thermal paste and while keeping the chips cooler might reduce the physical strain on the connections this is not the primary problem here because the GPU is not overheating. In fact if excessive heat was the problem then first component to be dead would be the CPU because it is stressed far more than the GPU, has a higher TDP and is part of every model.

If you read back over the past 3 months on this thread I've said more than once bad pasting is only part of the problem with these models.  It's a combination of a high power and heat, larger sized CPU die to cool plus hot GPU, exemption on lead solder use ending in 2011, excessive paste and uneven contact plate, plus if you stress test your CPU when in discrete mode and use a monitoring program you can see the temps rise on the GPU sensor very noticeably. Rises far more noticeable than with the 08-10 or 12 models.  The 2012 models are far less affected as the ivy bridge CPU runs far, far cooler than the 2011 Sandy bridge CPU plus it has an nvidia Kepler GPU which also generates far less heat than the AMD in the 2011.

But that's just an opinion of an anonymous bloke who's been working on mac and PC kit for just gone 25 years!

After more than a month of grey screens, I finally decided I could no longer wait and took my MacBook pro into Apple's Fith Avenue store in NYC. 

I showed them the problem, all the posts in the forum and the many internet articles.  Despite a long conversation with the manager they would not at all acknowlede the gravity of the problem.

They quoted me $700 to replace the logic board instore and only when complaining more vigoursly did they lower it to $310.

I decided to bite the bullet and have it repaired but am very upset at Apple and their lack of support.

Hopefully this GPU will prove better than the last one!

My card failed and I decided to research further and found hundreds (perhaps thousands) are dealing with the same issue. Two articles on issue over the past two to three months and still nothing official from Apple. I'm very disappointed. I have called AppleCare and there has been no official recall. I did have them log my issue and give me a ticket number. I recommend you do the same. We're on our own at this point but we should let Apple know so they can see the large number of failures. Apple should take this as an opportunity to acknowledge a hardware defect and do the right thing.

The Sage continues.

First, the background.  Both my wife and I own 17" Macbook Pro's.  Hers was an early 2011 model and mine was a late 2011.  About 2 weeks ago, hers started exhibiting the symptoms of this GPU problem.  Boot to gray screen.  Unexplained random halts.  Occasionally, boot into a lovely shade of royal blue screen with dark vertical lines.  The usual.  Mine was still running ok.  I went through the usual diagnostic type of things.  ADT couldn't find anything wrong but in it's defense, it did give me the assurance that the power was on.  I was able to get it to boot normally a couple of times by resetting the PRAM or the SMC or both.  Nothing consistent worked though.  Finally, I got tired of trying to fight with it and decided to do a swap with her.  Clone her failing machine to my later MBP and I'd take her earlier one.  That way, I'd have to fight with it and she could keep working.  At least for now.  So, out came CCC and a couple of hours later, I'd managed to swap our two machines.  Now I had the early 2011 MBP and she had the later 2011 one.  Took the one that was now mine down to my basement office.  This was last weekend.  All this past week, the earlier one that I'm now using has performed well.  No apparent problem.  At least until this weekend.

Fast forward to this weekend.  My normal job keeps me in my basement office during the workday all week.  So, I've got my Windows work machine on one desk and my Macbook Pro (now the early 2011 one) on the credenza.  On the weekends, I bring my MBP upstairs so I can be with my wife while we do our computer things.  Yesterday morning, brought the machine upstairs, set it up on the kitchen table, hit the power button, and as soon as it  came  up, it went straight to the solid gray screen.  Do not pass go.  Do not collect $200.  Repeated attempts to get the machine up have yielded little in the way of success.  I did boot into single user mode and moved all the AMD and ATI kernel extensions out of /System/Library/Extensions.  This allowed the machine to boot but the performance was so abysmal that I gave that idea up as a bad job and have resigned myself to the fact that the machine is going to have to go in for the usual repairs.

The dilemma.  Having read all 285 or so pages of this thread, it seems that here are basically two options open to me to get this fixed.  I can:

1) Take it to the Apple store, argue with the genius, and try to get it repaired through them.  Reports in this thread however indicate that people are having mixed luck with this approach.  It seems to fix the issue but in a percentage of cases, it seems to only be a temporary fix that dies again within a brief period of time.

2) Contact one of the reballers that are on eBay and have the GPU replaced and reballed.  Seems there are several of them out there.

So, I'm soliciting input on what the consensus would be on the best course of action.  Should I take what's behind door #1 above, or go for door #2?  If door #2, does anyone have any experience (good or bad) with any of the reballers that advertise on eBay.

I'd really like to have my barely 3 year old MBP back in working order.  It really should have the same life expectancy as the machine I'm currently typing this on.

Oh, that machine?

It's a 2003 Powerbook G4.  Works fine.  Lasts a long time.  My 2011 Macbook Pro should too.

Thanks for your time!

GavMackem wrote:

 

If you read back over the past 3 months on this thread I've said more than once bad pasting is only part of the problem with these models.  It's a combination of a high power and heat, larger sized CPU die to cool plus hot GPU, exemption on lead solder use ending in 2011, excessive paste and uneven contact plate, plus if you stress test your CPU when in discrete mode and use a monitoring program you can see the temps rise on the GPU sensor very noticeably.

All of that doesn't matter as long as the temperatures stay under the junction temperatures of the individual chips which is 100°C for the Sandy Bridge CPU and probably 90°C (or even higher) for the 6750M GPU. Before they reach the temperature they're going to throttle. Even when really stressing the system with FurMark for a long time I've never seen the temperatures exceed 80°C for the GPU and 90°C for the CPU which is still well within the specs. Here in normal operation the GPU doesn't do much while the CPU is quite a bit at maximum resulting in a very smooth 50°C at the GPU most of the time. Still I've already had 2 logicboard failures due to a broken GPU. Go figure.

If a system always operated within the specs failes with broken joints that's clearly a processing/manufacturing defect and not a result of incorrectly applied thermal paste. Fixing that slight mishap is a nice thing for everyone but not the cause of this problem.

MacBook Pro 15 2GHz early 2011
AMD Radeon HD 6490M
10.9.2

this happens when i close the cover and it goes to sleep
unable wake my MBP back up
so i reboot my mac and the apple screen has a pink tint to it and is pixelated
after the spinning wheel then it goes blank, fans and optical drive are still running
first time i try to boot from Recovery (command-R ), no luck

so boot from the install DVD and verify permissions and disk repair
disk is ok
every time when i fix permissions in disk utility i get this
Permissions differ on “private/var/db/GPURestartReporter”; should be drwxr-xr-x ; they are drwxrwx---

also if unable to boot from DVD i had to use other startup combination keys
like, safe boot, recovery and even Apple hardware test! for some reason it toke me directly to my login screen, weird!!!
in verbose mode the screen was red with white font

when my boot option failed i saw the Mac Blue Screen of Death or the Grey Screen of Death !! SCARY!!!! when your apple care expired!!!

so far i did not experienced any screen issues when MBP is awake

AND thank god i was able to login and BACKUP my mac !!

i was thinking my hard drive was failing, because of age. that i could have fix myself for cheap!!! but logic board fix $$$$ !!!! and not guaranteed to work

Honestly, you saved my live! Bachelor Thesis isn't gone! Thank you so much! Is there any way to do the same with the Bootcamp driver?

beniroquai wrote:

 

Honestly, you saved my live! Bachelor Thesis isn't gone! Thank you so much! Is there any way to do the same with the Bootcamp driver?

Sorry, I do not have experience with Bootcamp, so I cannot help much there. Glad that the Mac OSX side of things is working better for you.

degger wrote:

 

GavMackem wrote:

 

If you read back over the past 3 months on this thread I've said more than once bad pasting is only part of the problem with these models.  It's a combination of a high power and heat, larger sized CPU die to cool plus hot GPU, exemption on lead solder use ending in 2011, excessive paste and uneven contact plate, plus if you stress test your CPU when in discrete mode and use a monitoring program you can see the temps rise on the GPU sensor very noticeably.

All of that doesn't matter as long as the temperatures stay under the junction temperatures of the individual chips which is 100°C for the Sandy Bridge CPU and probably 90°C (or even higher) for the 6750M GPU. Before they reach the temperature they're going to throttle. Even when really stressing the system with FurMark for a long time I've never seen the temperatures exceed 80°C for the GPU and 90°C for the CPU which is still well within the specs. Here in normal operation the GPU doesn't do much while the CPU is quite a bit at maximum resulting in a very smooth 50°C at the GPU most of the time. Still I've already had 2 logicboard failures due to a broken GPU. Go figure.

 

If a system always operated within the specs failes with broken joints that's clearly a processing/manufacturing defect and not a result of incorrectly applied thermal paste. Fixing that slight mishap is a nice thing for everyone but not the cause of this problem.

For as much as I'd like to concur completely with you and for everyone affected by this (including most likely myself in the next year or two) I couldnt agree it's 'clearly' a lead free solder defect, my experience working on the whole Mac range throughout the years and having a physics brain I will only go as far as 'probably very likely' it's the root cause with the others playing their roles in making the problem worse.

I cannot be definitive because of the mitigating factors to be added in:

Using a thermal system originally desinged for a Core2Duo and Nvidia 9400/9600 GPU, far lower power consumption and die size. They had issues with the GPU failing in the 2010 15 inch Arrandale/Nvidia 3 series GPU which is on a recall, it got hotter in there. The following year ours with the most heat of all, followed by far less thermal problems in 2012 cos they used cooler silicon.  Totally redesigned, far larger and more efficient layout in the retina models subsequently.

The fact the iMac 2011 with similar GPU generation is on a recall, but with far easier to replace MXM module than our 2011 models as a repair program for the 2011 will cost far, far more than the iMac, and it was largely Nvidia's liabilty insurance, not Apple who footed the bill for the defective Nvidia 8 series GPU models.

I have seen many MBP/iMac/Mini models hit 100c on both parts, cutting out, some with blockages but many also down to too much paste which has dried out far away from the contact plate and die. The over pasting of the Shinetsu paste which I have seen literally cover the entire GPU part, around the sides even, 75% the area of the Sandy Bridge CPU which isnt small either. The fact that with this heatsink layout the CPU can transfer heat to the GPU, and with the globs of thermal paste far away from the die it's helping to transfer that heat where it shouldn't be going!

As computers have a fixed TDP and will throttle at the high end accordingly I prefer to use idle temps as a yard stick for how effiicent the system is working.  As I type this on my 2011 17 my idle CPU hovers in OSX between 34-40c, before re-pasting a month ago it was 20c hotter.  The GPU when active in OSX or Windows 7 with maximum performance power settings idles at 43-47C, the CPU about the same.  The temperatures have dropped quite a but with AS-5 paste they usually do drop after a couple of hundred hours. 

I've never been a fan of Apple's thermal stepping and fan control triggering in the unbody range one bit, I believe they are set far too high.  They prefer the arc of having low noise, sizzling hot left speaker side and a very warm leg whereas as an engineer give me a cooler but noisier running Mac laptop any day of the week.  I use MacsFanControl triggering mine far earlier with the fans set for the correct CPU/GPU layout to keep them how I like it, and not Apple.

For as much as I'd like to concur completely with you and for everyone affected by this (including most likely myself in the next year or two) I couldnt agree it's 'clearly' a lead free solder defect, my experience working on the whole Mac range throughout the years and having a physics brain I will only go as far as 'probably very likely' it's the root cause with the others playing their roles in making the problem worse.

Semiconductors do not randomly work, fail, work except for contact problems. Also there're plenty of people who got a reflow/reball (or god forbid a bake job) after which the same GPU is working just fine.

I think the thermal solution (despite the lousy paste job) is working perfectly fine and also keeping the components within their thermal specifications very much unlike many other vendors using weaker components but still facing the plague of emergency system shutdowns due to overheating where even massive throttling couldn't help.

The reasons why other devices received a recall and even much earlier than the time that has passed til now is manifold but some of which are certainly:

• A company to blame for defective components or wrong manufacturing instructions
• Easier to replace components
• Early catching of the problem far before end-of-manufacturing so many devices never left the shop with a defect and for others there were new spare parts
• legal threats
• more publicity

> MacsFanControl triggering mine far earlier with the

> fans set for the correct CPU/GPU layout

Could you specify which settings you like?

I'd like to like them myself.

Contact problems - you've also just reminded me of the lousy uneven surfaces of the heatsink plates to add to the list, I polish every one shiny with metal polish to optimise that contact.  I have got two successful baked 2011's out there for poor clients working fine, now 10/11 months and 7 months past though they are pasted my way and not via Apples official guidelines to engineering, which I firmly believe are even far more stupidly wrong for the 2011 than the others past and future,

I think that the thermal solution worked fine in the 2008/9 models despite the usual bad paste job, worsened with 2010 with more heat to dissipate and our 2011's fared the worst of all with superb but very hot and large sized silicon dies inside. Fixed partially with cooler parts in 2012.

As for your last part - my thoughts hypothetically largely concur with yours but I am fully aware of the guidelines of posting in these forums through experience if you catch my drift.  I know for a fact that in 2011 cooling semiconductors was a big problem due to the lead free exemption expiry as I have friends who manufacture kit which have to dissipate thousands of heat watts in rack mounted boxes. They ran into problems in a matter of months and not a couple of years, most likely because of the heat causing the solder to break much sooner than with chips in a MBP.  I'm not a believer of co-incidences being accidental like you but like everything it's not black and white but a shade of grey. Gunmetal or graphite in this case I think ;-)

It's past midnight here in the UK, everything is shut down and I'm in the smoking garage with the iPad, but iirc the left fan is set triggered for CPU0 and the right for the GPU, the presets macsfancontrol use are pretty good when you select it for standard settings but I adjust the lower band down because of my lower idle temps with what I define and not Apple as an optimum cooling setup inside for my MBP 2011. Bear in mind if you have a 15 the temps will be 2-4c higher than my 17 with less room inside.

It's a great utility on OSX and even more so for their Windows version on bootcamp, as their all round support for Windows is putting it mildly barely adequate.  An issue which I could rant on about a lot over all these years so I will say no more...!