Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Stress your Nvidia GPUS Now!!
- Thread starter gangzoom
- Start date
- Sort by reaction score
So? This is a piece of computer hardware, not the Candyman! I can open and close Mail.app a 1000 times in a row and it will continue to work. I can even press the Get Mail button over and over and over again within worrying that my Airport card will die from exhaustion. It's unacceptable that a GPU can even think of showing problems after ONLY 25 cycles.
We can still game on our MBP's, but don't put the machines to sleep after doing some GPU intensive applications. Just allow the GPU to go back to normal operating temperatures so the system can dispel the heat with the fans on, etc.
PS: It isn't wrong to test to see if you have the faulty part, though doing it just to get a new machine is wrong.
PS: It isn't wrong to test to see if you have the faulty part, though doing it just to get a new machine is wrong.
This is just my opinion.
I must say that I agree with you, but stress testing an affected machine isn't going to do any good. What's the point? We already know the models which are bound to fail. This will only exacerbate the situation even more. The most Apple is going to do in the mean time is hand out affected replacements. I highly doubt they will even think about refunds - unless it's within the 14 day range of the return policy.
They need some time to assess the damage and figure a way out to solve the problem. The best thing we can do is wait until more info comes out from Apple/Nvidia or any other PC makers who are also in this dilemma.
Even though I have an affected machine (Feb '08 model), I'm going to try to stay clear of overheating my machine intentionally. I will do all my work no matter how GPU/CPU intensive it is, but I will not run a script to blow the GPU up. Yes, these are expensive machines and we shouldn't be standing here doing nothing. Instead, we should be rational and voice our distaste to Apple via phone, email, feedback forms or whatever. If that fails, we jump to the man himself - Steve Jobs. We can email him and let him know this isn't acceptable and that they should get Nvidia to pay for this.
Destroying the GPU at this point in time - it's not going to help anyone.
The "scratch-resistant" analogy doesn't make any sense here. CPUs/GPUs are not supposed to die during use. That means ANY kind of use, providing they've got the correct cooling and the correct voltage, and this is all predetermined by the manufacturer. If this had been possible before you would have seen viruses which would physically break computers by just running them fast then slow.
Actually it was about 25 times over 2-3 weeks, and not in a row cos i actually use my MBP for work!!! I simply wanted to see it was true that rapid changes in the temperature of the GPU can cause a serious problem over time...
My last Apple Laptop lasted me 8 years (Pismo G3), and I had planned on keeping my MBP for as long, so when i heard about possible problems around nVidia's GPUs i simply wanted to be sure that the product i just bought wasn't going to crap out on me in 2-3 years time..when my apple care warranty runs out!!...i was simply been prudent.
My family have bought over 13 macs over the last 20 years, i can safely say if Apple doesn't sort out this problem quickly its going to be a while before i consider buying another Apple product....not counting the iPhone
When he wasn't using the laptop != in a row.
Also, I encourage everyone to stress their GPU if they're under warranty. It's not against the agreement to use your computer in any way that you want so long as you're not altering the clock speeds and other 'warranty breaking' activities.
For those calling it fraud, what about it is fraud? The GPU should be able to preform in those circumstances, and it didn't. It's defective and should be replaced.
The fact is, if the GPU is defective and Apple knows about it - they haven't done anything about it. That's irresponsible and nearly makes me reconsider my future Macbook Pro purchase.
Also, I encourage everyone to stress their GPU if they're under warranty. It's not against the agreement to use your computer in any way that you want so long as you're not altering the clock speeds and other 'warranty breaking' activities.
For those calling it fraud, what about it is fraud? The GPU should be able to preform in those circumstances, and it didn't. It's defective and should be replaced.
The fact is, if the GPU is defective and Apple knows about it - they haven't done anything about it. That's irresponsible and nearly makes me reconsider my future Macbook Pro purchase.
I do believe you've misunderstood what I mean by 'responsible'. If your laptop has an issue, it goes back to the people who sold you it, of course. What I meant was defrauding Apple to claim on your insurance policy is a little extreme and ill-advised as your treating it as if it's Apple's fault. It isn't; the GPU issue is nVidia's. Contact Apple if you have an issue, and be patient.
its not fraud people. EVERY major company stress tests their machines for this. if someone wants to test their machine under tech specs, they cann do so how many times they like.......and it SHOULDNT break the machine as its being tested under tech specs
however, if i like took a hammer to the screen to break my mb and say that it was damaged under mail delivary, that would be fraud as the mb WASNT designed for that
see the difference?
how can it be fraud if the test is run under the supposable design of the machine? its like me constantly writing huge files to a hard drive, deleting them, and doing it again over and over. if the hd fails and under warranty, why shouldnt i get a new one? the hd company says they have a waranty on it
last i checked, companies cant enforce usage patterns on individuals for waranty purposes as long as the user is using the product within design specs
however, if i like took a hammer to the screen to break my mb and say that it was damaged under mail delivary, that would be fraud as the mb WASNT designed for that
see the difference?
how can it be fraud if the test is run under the supposable design of the machine? its like me constantly writing huge files to a hard drive, deleting them, and doing it again over and over. if the hd fails and under warranty, why shouldnt i get a new one? the hd company says they have a waranty on it
last i checked, companies cant enforce usage patterns on individuals for waranty purposes as long as the user is using the product within design specs
ok I'm sold. You are right. I thought you were booting into crysis on max settings, getting your GPU to max temperatures and slaming the lid shut and repeating. Sorry I doubted you. If your test holds true for most mbps, thats ******** and apple needs to address the problem. Another reason to wait out until the next revision.
all im trying to make out with that analogy is that there is a reasonable limit within which to test out hardware. repeating something 25 times means that he was expecting it... i mean if nothing happened on the 25th time im pretty sure he wouldve continued until something did come up.
and btw, even if the CPU/GPU did have proper cooling and there were no defects, repeated trials of a very stressful test would compound the heating from previous trials thus leading to overheating. u could try this with a lightbulb (altho i wouldn't advise u to do so cuz it's not really safe), turn it on and off 50+ times the filament will remain glowing even after u have switched it off.
I'm not trying to be rude, but REALLY?! You encourage us, who use these machines for work, to actually "stress" the GPU? What's there for us to gain from this other than a possible replacement from Apple with the same GPU? Once you get the new one, are you going to stress it again? The cycle is just going to continue. We know the affected models, so there is no need to keep testing. Just voice your displeasure with Apple by either contacting them via email, feedback forms, the discussion forums, and/or telephone. Email Steve Jobs if you have to. I believe it's sjobs@mac or apple.com.
If there was nothing wrong with the GPU then nothing bad would happen. The only reason he was expecting something to go wrong is because of the rumours that there are problems. And it appears that (for him at least) they were correct.
This is an irrelevant analogy. Cars use incandescent bulbs for their blinking turn lights and these last thousands upon thousands of times being switched on and off.
It gives an example of a worst case usage pattern to failure... I now have a slightly better idea how thick my kids gloves need to be, so personally I find this info quite useful. Additionally my impression there are many on this forum who believe this is a sporadic issue and not endemic to a design flaw - it lends more credence if some random sample can be made to fail when stressed.
Agreed. Apple has no available fix and we can do little more than speculate on how things are going to carry out. And admittedly as a developer who regularly sees my CPU/GPU (sustained for 30 minute stretches)to over 70C during compilations I'm probably a bit more concerned than most.
lol... all im getting at is that the first issue at hand was that devising a way to purposely destroy a component (defective or non-defective) by stressing it multiple times to get a full monetary refund/credit IS considered fraud by law. i know that for sure. for similar reasons, there is the ISO 13406-2 standard for pixel damage in LCD screen manufacturers. Viewsonic and Samsung have switched from a zero-tolerance policy to the ISO 13406-2 for their screens due to this possible loophole
If there's a chance a laptop breaks under stress I'd rather find out about it sooner while it's still under warranty and when I'm NOT in the middle of critical work, than have it break randomly when both I and it are under stress during a tight deadline.
Presumably now that NVIDIA knows about the problem they're going to ship fixed chips. What people hope to gain from this is a non-defective laptop. A "weak die/packaging material set" replaced with "We have switched production to a more robust die/package material set". Why is this too much to ask for?
I guess if the OP has not done any overclocking (on osx anyhow) and stressed it to make it do what his mbp did.. then it shouldnt be his fault and by default should work fine.
But if he went and logged onto windows and majorly overclocked the card and stressed the hell out of it then that's not right.
But if he was legit about using the default settings to make the card go a bit high under osx and just put it to sleep then that's bad news for us...
I mean even though the temperature might run in the 70's or 80 degrees and you put your mbp to sleep instead of letting it cool down should be normal. People dont wait until the fans cool it off before putting the notebook to sleep, I know I sure as hell never did that.
When I'm done with using the mbp I just closed the lid and go.
But if he went and logged onto windows and majorly overclocked the card and stressed the hell out of it then that's not right.
But if he was legit about using the default settings to make the card go a bit high under osx and just put it to sleep then that's bad news for us...
I mean even though the temperature might run in the 70's or 80 degrees and you put your mbp to sleep instead of letting it cool down should be normal. People dont wait until the fans cool it off before putting the notebook to sleep, I know I sure as hell never did that.
When I'm done with using the mbp I just closed the lid and go.
I do get where you're going with the analogy but using a computer to its max has always been considered perfectly acceptable, both by us and by the manufacturers. This is why NVIDIA has to cough up the $200 million because this nonsense just isn't normal. People run all kinds of stress tests on their PC to make sure they're working correctly (applications like Super PI, 3D Mark, etc etc), things designed specifically to try and make the computer fail. It's standard stuff and fully approved by computer manufacturers.
The difference is overheating w/ failsafes kicking in to prevent damage vs. actually incurring damage.
Also doing this test back to back might actually be more benign than doing it randomly - allegedly the issue is caused by thermal expansion/contraction. 25 cycles starting from cold to running hard back to cold (spaced out over hours or days or whatever) vs. running back to back where there is some latent heat from each prior test, there is a greater temp differential in the former.
Thermal issues like this don't just go away if you ignore them. They'll just take longer to develop. You might split your 25 runs of Call of Duty over a period of 18 months instead of 3 weeks, but in the end the result is the same. After 18 months it's still going to be a defective GPU, except you'll be the one forking out the money for a new logic board instead of Apple who can at least claim back from NVIDIA.
Alternatively they'll just slow the 8600M down so that it doesn't get as hot. Having a slower laptop than the one I bought doesn't seem like a great option either.
Has anyone tried calling Apple Care, and leveling with the tech on the phone. Just ask them, seriously, should I be worried? what will apple do to make me feel confident that if and or when my machine dies, I'm not going to get the run around. I'm curious what their respone would be?
Dont forget though nvidia's gpu costs them something like $5-$10 instead of the $300 on the xbox 360.
Also microsoft was losing money to begin with, with the xbox 360 to make money off the games.
I had my xbox 360 repaired once 2 years ago and it still works fine today but still crossing fingers.