MacBook Pro 2011 17" hard freeze

Sid Merrett wrote:
I've had my MBP 2.3 15" for 5 days.
It has now frozen or locked up about 6 or 7 times.

Each time, the fan(s) were spinning audibly.
Last time it happened I was running Handbrake converting video.
Machine froze, no mouse movement - completely "dead" and had to hold down power button to shut off.

Twice it has happened when running BioShock - in this case, in one instance at least, I Apple/Tabbed out of Bioshock to go into finder and do something and the screen went black, OSX seemingly unable to "get out" of Bioshock and the whole thing froze - ALT/APPLE/ESC etc did NOTHING.
Had to hold down power button again.

Also, I have had similar behaviour, when fans were spinning loudly and thus, presumably the machine working relatively hard, but the mouse would stay responsive to movement but clicking did not work for anything. Again, Apple/Tab etc would not work, could not force quit anything - effectively the same problems as above, but the mouse would at least move.
Only way to fix = hold down power button.

In each instance, the machine started up fine, immediately and most recently (such as right now) I am running Handbrake again and typing this post, and the MBP appears fine.
Thus, the problem does NOT occur everytime the CPU/GPU is under load, but it HAS ONLY happened when the CPU/GPU is under load.

Rather worrying and I will be talking to Apple first thing Monday morning!

Same behaviour for me, if i video encode and leave the machine alone it completes successfully but if encode and do anything else like browse the web then i get the freeze, not always but usually when some flash content runs as well. I don't believe it is the flash or the video encoding that is the problem but loads they require together..

It just has to be a fault in the CPU/GPU..

I don't see how a software fix would help unless it intentionally throttles something back. Unfortunately I'm sure that would be the route other than a recall. I've never known Apple to admit to any fault.. iphone4 death grip being the last prime example.. Response? We're holding it wrong.

yikes. this reminds me of the old 2009 SATA thread.

anywho.. just adding to the bucket. experiencing a hard freeze here as well. Open Xcode4, build one of my apps, freeze. No KP, just mouse and graphics freeze. CPU usage is high according to activity monitor, kernel_task is using the most cpu when it happens. No GPU intensive activities are going on. It happens every single time without fail, even after a fresh boot. Fans going full blast.

I just opened this project tonight since getting this machine a few weeks ago so I could make some fixes. Pretty bummed.

Tried the suggestion in previous posts about disabling the ATI gpu via that tool, and disabling graphic switching so it always used the ATI gpu.. didn't seem to matter. I did discover that when connected to a 27" cinema display, the onboard graphics won't drive it, ie: force it to the intel graphics, and nothing happens when the display is connected. I don't know if that's true for smaller displays.

Hmmm... now I can't get mine to crash no matter what I do...

Been running the CPU at 100% (at 93deg) encoding video for the last hour and no crashes.

Seems the problem is intermittent?

Message was edited by: Madhat

I can't seem to re-create the problem once i un-tick 'automatic graphics switching' and go with the high end graphics

I video encoded and run some heavy flash websites and i could not re-create my problem thus far.

Is anyone else's HD 6750 or HD 6490 running with x8 PCIe lanes instead of x16? Or is it just me and 1 or 2 other people?

You can find out what yours is running at in System Profiler under Graphics/Display then selecting your Radeon chip.

I found out using Sisoft Sandra in Boot Camp that the Sandy Bridge platform in my CPU has been configured as two x8 PCIe ports, one for the GFX and the other with nothing attached. It should be configured as one x16 for the GFX only. It can also be configured as one x8 and two x4 ports according to the Intel website.

I don't know if I should take my MBP back or see if a firmware update with the correct configuration comes out, if one even does come out. Apart from the GFX running at half the bus speed everything else seems to be functioning normally.

8x may be correct...

Read this: http://forums.macrumors.com/showpost.php?p=12021055&postcount=7



"The Sandy Bridge platform, currently only has 16 available PCIe lanes. Edit: Thunderbolt conenctes via a 4x PCIe conenction. A quick google search provided me the following link http://electronicdesign.com/Content/...62390_fig3.jpg. Therefor the GPU cannot utilize the full 16 lanes as 4 are already being used, and there is no "12x" specification, so the GPU utilizes 8.

The 8x PCIe bandwidth for the GPU should not affect performance. In very few situational circumstances would High end desktop GPU's, perhaps in multi GPU configurations see some performance increases from 8x to 16x, but in every review that I have seen its only marginal (< 5%). The Notebook version of the 6750 GPU should not see any adverse effect at all as it does not provide the fill rate or bandwidth even close to high end desktop GPUs. "

Message was edited by: Madhat

That seems to make sense as the 2nd x8 port on my system in Sisoft says it is running at 4x with nothing attached. If only I had a Thunderbolt peripheral to see if it comes up there. It's to bad Thunderbolt cant use a port on the Cougar Point support chip, but it doesn't look like it will make much difference.

I just wanted to make sure there was nothing wrong with my system.

http://electronicdesign.com/Content/UserStorage/14946/62390_fig3.jpg
This diagram of Thunderbolt shows an x4 interface to the PCH which in the MBP is the series 6 (Cougar Point) chip. So it shouldn't use the lanes from the CPU unless Apple has done it a little bit different. I'll have to go back in Windows and see if there are any available PCIe ports on the Cougar Point chip and if not then Apple must have resorted to using some of the CPU's.

Ok well that's, that. Cougar Point has three PCIe ports, they are all x1 and each are used by airport, ethernet and firewire. So the CPU is the only place to get 4 lanes but it looks like Intel intends for Thunderbolt to use the PCH when it has more PCIe support. Which won't be an issue with desktops.

Chullo wrote:
So, In summary I recommend everyone download gfxcardstatus and keep the machine in integrated mode until a fix is found. I am now doing things that would certainly kill the machine on discrete. Hopefully that isn't too far off.

As I was saying since 10 pages back - I told you guys that integrated, when forced, stopped the issue. On my machine the gfxCardStatus should have other letters - w - for working, f - for freeze, instead of i and d, because it's basically the same thing.. a kill switch

You were also wondering if the performance isn't too good with the Intel gpu. I was testing few heavy load tests suggested here to try crash it, but also I tried playing few games, CoD or some snowboarding or whatever - with Intel gpu it just isn't the same performance. No anti aliasing is really an option and I had to lower some quality to make it playable. But still - it's far more powerful than my good old white iMac just next on the table.

Another thing is the Windows thing. Mine doesn't work good with the ATI card anymore. And the most surprising part was it all happened in the SAME moment. I had my first crash - restarted and booted into Windows.. and it all was broken there, gpu driver failed and is still not usable

anyone knows a way to force the integrated gpu on win? would be nice to test that out too

I had to come reply here after reading through the thread. This is software related for most people. How do I know? I crashed my mac over and over using the smalluxgpu. I went over to the windows side and burned this computer up using burnin and other program with no freezes.

UNINSTALL smcfancontrol. I ran it with smc fan control on and it freezes in an instant. Uninstall and it is fine. I will report as I find out more, but so far so good. The dynamic gpucontrol is fine for me. It is smc fancontrol.


Message was edited by: operaman1

I think Horniasty brings up a good point. Folks that are just now joining the discussion would probably benefit from reading through this thread. Many of the things people are discovering have already been found and discussed at length.

I know this can be difficult. For example somehow I missed Horniasty's post 🙂

I have read through this thread, and the likelihood of Apple releasing all of these laptops with known faults is bordering on zero. I have proven it from my end. I bought THE first batch available and had mine from America to me here in Germany with in days of release. I too could crash my system with the smallluxgpu within minutes.

Listen, I am not some Mac newbie or even PC newbie. I look to tear apart these things for the fun of it. I am an avid over clocker, and I love to push things to the limit. In the vast majority of cases this is a software issue. Mine was solved by removing both istat and smcfancontrol. I have tried to crash my machine since the first lockups. I have rebooted several times, and I will bet the house this a software-fIrmware issue at worst. I have pushed this laptop since removing smcfancontrol with insane amounts of loads and it is not crashing. This is tested both in windows and OSX. I first figured this was a software issue when I pushed the CPU and GPU to their limits last night in Windows after freaking out from getting the freeze replicating the smallluxgpu test from this thread.

I could not replicate the crash. I ran ATITool. I ran Orthus with 8 threads while running GPU burnin. While my MBP sounded like a jet it never ever crashed. For me this kills it being a bad GPU. I was also monitoring under Windows and there was no throttling to be seen. So I am feeling certain than in most cases this will be fixed with an update. I have a vested interest like everyone here in my MBP having dropped over 3k for mine.

Message was edited by: operaman1

That's very interesting....since I was able to reliably reproduce the issue on apples stock machines in the apple store. I'm now in agreement that it should bead river issue, but it varies directly with heat. Maybe an error in the driver, could be something as simple as a memory leak....running the machine @ high stress and generating just 4kb of leaks in x-cycles will bring the GPU to a halt in fair order. The GPU ram itself could be bad, could still be an overheating issue....of the GPU not the CPU (since we've been running stresses since last night on integrated w/o failure).

If you remember the 2009 SATA problem in detail, you will remember that Apple NEVER fixed the issue. They replaced my machine 4 times before finally giving me a fully upgraded 2010 model 10 months ago. No issues with SATA now, but I do have a similar sounding hard freeze issue.