All nVidia G84 and G86 cores are defective?!

I already posted this in the other thread from the front page, but I'm posting here in case no one is watching the old thread. TheInquirer is reporting that all of nVidia's chips using either the G84 or G86 cores are defective. Basically the chips have a flaw relating to how they handle heat (basically they can't). The affected chips are:

8300 GS
8400 GS
8500 GT
8600 GS
8600 GT
8600 GTS

Time to cancel my XPS 1530 purchase? I think so.

 

Why would the 8800 GTS be affected? It's built on the G92m core.

 

You're right KGann, the 8800 GTS is not affected.

 

Let's hope this isn't true, as this will not only destroy the company, (which I don't care too much about), but will destroy the Notebook world. If Nvidia has to recall these cards, they will most likely go into a financial deficit, which will only cause chaos in the Notebook world.

 

Is this just laptops or are desktop card as well.

I suppose it could explain the down-clocking that laptop users have reported!

 

Honestly, I doubt that nVidia will do a recall. There is already a BIOS update from HP and Dell the causes the fan to startup sooner, and run longer. The real problem for nVidia is how tough HP, Dell, and others are going to be when it comes to recouping the costs of replacing/repairing laptops.

Considering that nVidia is laying blame at feet of the OEMs I expect to see a huge increase in ATI graphics offerings. Too bad that is at least 6 months out.

@Tinderbox (UK)
From the article at TheInqurer.net:

 

Is that A03 revision the GF9 that are G84/86 ?

 

So does that mean free upgrades to 8800 for all G84 and G86 owners!

 

Damn. That sucks for nVidia and the users of those cards.

I wonder if we'll see a push of the ATI HD 2000 series cards, and eventually the 3000 series.

 

Most probably NOT. The affected ones are notebook GPUs. Majority and almost all notebooks that uses G84/G86 NVIDIA cores won't be able to handle the heat of 8800M.

 

What was the problem exactly?

 

ckh: I'm going out on a limb here... but I think he was joking when he said that...

 

There's a flaw somewhere in the silicon that makes the core sensible to temperature variations. When the laptop's off, it' about 25C, idle 70C and load 90C and they seem to have difficulties handling that.

 

Here's the full article:

 

heh this makes sense. now that ive hit summer months my laptop goes up to at least 80c while playing wow. Usually around 90 the card downclocks and destroys my framerate from liek 60 to 15-17 fps. Looks like I need to talk to dell soon maybe.

 

It does make sense, I was always wondering why the Asus G1s idled at 70-80c.

PS: Articles written like this is the reason I hate The Inquirer.

 

I like how mine's problem free until 100C, usually handle 104C though

 

I feel better knowing I don't overclock, yet to see GPU go over 65 degrees. I hope the problem doesn't affect me in the present or future either.

I agree with link1313, the article may be informative but the writing style is insufferably arrogant and egoistical.

 

Yea i don't kow I hadn't had any heat issues for 10 months my laptop stayed aounrd 60 even while gaming but recently temps have exploded only while gaming tho. I'm at 56 idle right now.

 

We should not take anything Charlie Demerjian says about NVIDIA as "news" or fact.

 

Finally an explanation why some users of the 8600M GT GDDR3 experience insane stuttering past 96C?

 

While it may very well be the case that something is wrong with your card, overheating in the summer is not necessarily an indicator of that. Computers are simply not designed to operate at summer temperatures unless specifically stated to do so. My old HP laptop with an ATI card would occasionally shut down for no apparent reason when I tried to run it in 35C/95F weather.

To be honest, I'm not really too worried. The G86/G84 cards have been around since May 2007 and the number of failures has not been enough to overwhelm customer support of the OEMs. Thus, even if they are flawed, these things can live for at least a year. Is it unfortunate to have this increased probability of failure? Yes. But worth worrying about? Probably not -- at least not until they actually do start going "south in inordinate numbers".

 

Looks like Over-clocking is about to start seeing its demise. Well, if you use common sense. I OC my 8400m GT, but I don't see myself doing that anymore.

 

While unfortunate, my G2S has been and will be fine. So since nVidia did nothing to ME, I won't touch them. I do have sympathy for those affected, however.

 

Now may be the time to extend my warranty if at all possible. Also, I get downclocking at about 60c.

 

"Defective" or not, not all G84/G86 GPU's perform bad. Assuming this isn't some "ticking time bomb". As long as you research the laptop you plan to purchase beforehand and/or ensure you have a warranty to cover abnormal performance or failure, there's probably not much to worry about.

I knew my G1S would have heat management problems prior to buying it, so I merely see this as a potential explanation instead of as a cause for concern. So I'll probably continue to overclock it.

 

And where is this bios update..........?

 

hmmm just wondering if there is anything owners of these cards can get replacements even if there are no problems right now? My Inspiron 1420 has warranty for 3 years, can I call them up and exaggerate my situation or they will not do anything about it unless something actually happens?

 

The Inquirer always posts sensationalists stories so I would take what they report with a grain of salt. While I agree that the problem is likely more widespread than the HP units that nVidia admits, I don't think it's as bad as as 2 entire chip families getting ready to blow up in people's faces.

I'm still not clear what exactly The Inq is pointing that is the source of the problem. They mention that the extra heat and early failures are due to the extra voltage nVidia needed to use on the A03 stepping to get it to run stable at the clocks they needed. (I don't believe the A02 stepping ever shipped). But do they mean that this is the root cause rather than poor packaging or is this in addition to packaging problems?

In any case, I wish we knew the max temperature that the GPUs are rated. Intel mobile CPUs are rated to operate at 100C, and I have a feeling that GPUs would be similar, meaning that 80C+ GPU temps may not be indicative of anything, although lower temps are obviously preferable.

 

My 8600m GT GDDR3 immediately downclocks at 110°C, so I would guess they don't them to reach 120°C.

 

YAY! 9800gtx on the desktop, and 8800gtx in the laptop. Thank god I missed it. This is one of the times where getting the cutting edge (of the time) doesn't hurt . lol.

 

what are you playing to get such high temps? The highest my g1s has gone is about 92C after about an hour of GRID, which seems to fun even hotter than Bioshock. Luckily I haven't experienced any downclocking after doing the sleep fix.

 

I too appear to have dodged the bullet.

G1 has go7700, desktop 8800gt

 

Hmm.. since my G84 (8700m GT) does seem to have the flaw and doesn't ever rise above 69C, odds are I have the cooling to handle an 8800m pretty easily.

I'll be more than happy to accept!
(yes, I know that if it doesn't ever really heat up, I never actually face the issue)

 

There isn't a chance in the underworld they will replace the G84 or G86 cores with 8800m's.


But if you petition...

 

It was actually the ATITool 3D view that would get it there (while overclocked). Also depending on the laptops positioning and the room temperature (85-97°F), games such as Crysis, COD4, FFOW, and Turok would also cause it to reach 110°C (while overclocked). However, since I applied Arctic Silver 5 throughout the G1S temperatures rarely reach 90°C.

 

In the M15x the 8800M GTX runs cooler than the 8700M GT from what I gathered here so Isn't the X205 SLI-capable anyway? It should have a very good cooling system then

 

Hmm, I thought the M15x had issues with cooling right up until they added a patch to make the fans hit high speed faster.

The x205 stays cool as a cucumber and the fans generally never kick into high... I think they planned to cool a much hotter GPU in this system or simply overengineered it not knowing the stats of the up-and-coming mobile GPUs and chose to err on the side of insane cooling.

With the new x305s shipping I am thinking talking to the right person may actually allow us x205/x200 users to have a GPU hardware upgrade component which might actually physically fit in our systems.... the problem will be video BIOS. (the GPU card is indeed a separate piece, but the video BIOS is part of the system BIOS)

I have no illusions about any "upgrade" and am not expecting a warranty or recall system to kick in. Truth is, I don't really need it... that doesn't mean I wouldn't like an opportunity to upgrade to 8800m or higher... yes, even if I have to pay for it and yes even if it isn't covered under warranty...

 

I'm idling around 45 degrees on stock clocks and 169.09.

 

I'm running with an 8600m GT in my Macbook Pro 2.4 model. So far I haven't had any issues, though the GPU temps do hit between 78 and 88 degrees Celsius while gaming. I am wondering if the EXTREMELY proprietary nature of Apple hardware might have saved me from these recent failures. The GPU is soldered directly into the Apple motherboard. I think these first generation models may also have been underclocked a tiny bit due to heat constraints.

It's been a year since I first got this laptop, hopefully it'll still be problem free another year from now.

 

8600GS/GT - G84M
9500GS - G84-50

The 9500MGS is a 'better' version of the 8600GT. Same performance @ lower power consumption & heat.
Does anybody know if they could fall under the threat as well?

 

And here I was all happy that my card wasn't listed..........now you go and ruin it.

Whatever, I'll continue to OC the card, when or if it dies, I'll buy a new laptop......it's not the end of the world and it's certainly not the first time a company has had issues of this nature.......recalls happen all of the time and if that happens here too, Nvidia will survive most likely just like GM, Ford, Chrysler and every other auto manufacturer has in the past.

 

It shipped. My 8600M GT has Revision A2 according to GPU-Z. I think this is a separate matter that they throw out there to highlight Nvidia's wrongdoings. Their basic point is that some material in the card is bad; voltage would have very little to do with it.

 

It's the simple of question how hot a cpu or gpu or any parts run.
Gpu wasn't made to run at 120C just because the display in the control panel goes that far.

Anything over 80C will start baking the chip.

Max temp is 73C in my 8600m gt and I even find that too hot. (g84m a2)
There is a reason for desktop components cooling size. And notebook max get a small fan or heat pipe.

 

GPU-Z also reports mine as being A2, but the memory is underclocked by about 15%, so that might be a reason why mine is still working well. The core is also underclocked a tiny amount.

 

hmmz i dont use my notebook that often for gaming but lately when i fire up dell media direct it goes crazy with ****ed up colours. If this is the beginning of a failure I'm going to play cod4 a lot om my notebook and hope it dies in a week so i can claim my money back >.<

guess AMD/ATI couldnt be hapier at the moment they are breathing in nvidea's neck and nvidea just keeps making major mistakes

my gpu at idle is only 50C though and when i game its mostly arround 65 never seen it hotter then 68 and thats oc'd at 600/900

 

Oh man, out of all the notebooks, HP just had to be the "worst victim". I just hope my laptop blows up, so I can get an even better replacement

 

8600M GS is a G86 btw

Mine's been baking for a year and no problem. I noticed that the downclockings (which actually are driver crash) were caused by the drivers. No problem on 167.xx and 177.xx

 

As noted earlier by Itcommander, anything from the Inq should be taken with a grain of salt. They are blaming the problem on incompetent engineering without any credible evidence other than pointing to a voltage bump between revisions A02 and A03. I'm no computer science expert/engineer, but the chips needed some more power. So what. Now if Nvidia didn't tell its clients that the chips were going to run as hot as they are, then yes, that's a problem.

Nvidia seriously messed up with the defective GPU shipments, that much is true. They also messed up with the GTX 200 series pricing. However to suggest all the GPUs are bad . . . I think not.

Disregarding this article, I've lost my confidence/enthusiasm in this company and will not be purchasing another GPU made by them again unless I see serious improvements in the problems areas I have been having with my desktop's Nvidia card.