NVIDIA shows signs ... [2008 - 2017]

Status
Not open for further replies.
"People who have experienced a failure":
  • G84/G86 mobile - 38%
  • G84/G86 desktop - 26%
  • G92/G94 - 16%
  • multiple from this set - 17%
  • overall - 20%
Pity Tech Report didn't put G80 in there, its longevity plus good track record could have provided an interesting comparison.

Jawed
 
"People who have experienced a failure":
  • G84/G86 mobile - 38%
  • G84/G86 desktop - 26%
  • G92/G94 - 16%
  • multiple from this set - 17%
  • overall - 20%
Pity Tech Report didn't put G80 in there, its longevity plus good track record could have provided an interesting comparison.

Jawed

This math is a little dodgy, tho I'm willing to concede that Tech Report is at fault. It relies on ignoring the "multiple" catagories.
 
I'm pretty sure that 90% or more of those failures are from OCing or improper cooling. Their poll does not exclude either...
 
I'm pretty sure that 90% or more of those failures are from OCing or improper cooling. Their poll does not exclude either...

I've never seen a card fail from overclocking in all my years of gaming on the PC platform using 3d accelerators (since the mid-90's, and yes, I've been o/c'ing since then). Cards fail from manufacturing and component defects, as well as poorly-ventilated cases and/or clogged primary card cooling systems.

If you try to run an o/c that is greater than what the card can reliably handle, the driver will generally lockup, causing the system to reboot (XP and earlier).

I'm relatively sure that people smart enough to build their own machines and are part of TechReport's community probably know how to keep their systems well-maintained. Just a guess. It's not like we're talking about some comment section on a vendor site.

NV's current problems are a result of chip creep, a thermal issue. While the clockspeed of a chip certainly affects its heat output, even stock clocks can cause a chip to fail over time, especially when there is a defect in the packaging used which exacerbates chip creep, as is the case w/NV's failing chips. Case-in-point: look at all the non-overclocked mobile G84/G86 failures on bog-standard OEM systems.
 
Other known causes are Vmods and using after-market coolers that are either installed incorrectly or are inappropriate for the card.

Those with the latest AMD cards that changed coolers and failed to cool the voltage regulators learned the latter.
 
NV's current problems are a result of chip creep, a thermal issue. While the clockspeed of a chip certainly affects its heat output, even stock clocks can cause a chip to fail over time, especially when there is a defect in the packaging used which exacerbates chip creep, as is the case w/NV's failing chips. Case-in-point: look at all the non-overclocked mobile G84/G86 failures on bog-standard OEM systems.


I have atleast 1 card from every gen Nvidia has made and I've only ever had 1 card die, a GF256SE by Herules and it was a memory issue.

I agree with I.S.T on this one, Nvidia has fan crontrol set to way wierd settings for low and high fan speeds. Most users will OC their cards and not adjust these settings which only worsens the problem. Personnaly I keep my fan speed at 100% no matter if idle or gaming.
 
Other known causes are Vmods and using after-market coolers that are either installed incorrectly or are inappropriate for the card.

No doubt. Poorly-installed aftermarket coolers are not the fault of anyone other than the installer, however. Likewise, voltage modifications void warranties anyway, and anyone that is capable of performing one will already know this and have no one but themselves to blame for any failure.

Those with the latest AMD cards that changed coolers and failed to cool the voltage regulators learned the latter.

True, but I would also like to point out that a lot of these aftermarket coolers did indeed state support for some of these cards, right on the packaging often-times.

I have atleast 1 card from every gen Nvidia has made and I've only ever had 1 card die, a GF256SE by Herules and it was a memory issue.

I agree with I.S.T on this one, Nvidia has fan crontrol set to way wierd settings for low and high fan speeds. Most users will OC their cards and not adjust these settings which only worsens the problem. Personnaly I keep my fan speed at 100% no matter if idle or gaming.

"Most users" don't even know what overclocking is, so you'd be wrong in your assumption.
 
No doubt. Poorly-installed aftermarket coolers are not the fault of anyone other than the installer, however. Likewise, voltage modifications void warranties anyway, and anyone that is capable of performing one will already know this and have no one but themselves to blame for any failure.
Sure, they'd be to blame, but that doesn't necessarily stop them from voting in a web poll that their cards failed.
 
I dont know what to think of those techreport polls. Its hardly a controlled polling enviroment. In my conversations with Nvidia. They vehemently deny any desktop problems with the G92. I have yet to ever see one fail. And I have 3 9800GTX, 2 8800GT, and 2 9800GTX + ((G92B may not be relevant)). The cards I recieve are no different than any editor hardware. I have lost 2 Nvidia cards in my lifetime. a Geforce Ti 4200, and a Geforce 6800NU Modded to 16/6. Both were massively overclocked. My problem with the poll is there are no ways to prove proof of ownership or failure. Your taking alot of anonymous data to heart from it.


*Edit* Heh I just tested it. I was able to vote 5x. And probably could do more if I felt compelled too. Pretty easy thing to manipulate.
 
Last edited by a moderator:
Boo for flawed polling! I also have yet to see any G92/G94 hardware failed in the field, and have quite a few friends using O/C'd 8800 GT/GTS's. My own 8800 GT runs @ 700/1688/1980 (albeit with aftermarket cooling - TT DuOrb) without issue.
 
I dont know what to think of those techreport polls. Its hardly a controlled polling enviroment. In my conversations with Nvidia. They vehemently deny any desktop problems with the G92. I have yet to ever see one fail. And I have 3 9800GTX, 2 8800GT, and 2 9800GTX + ((G92B may not be relevant)). The cards I recieve are no different than any editor hardware. I have lost 2 Nvidia cards in my lifetime. a Geforce Ti 4200, and a Geforce 6800NU Modded to 16/6. Both were massively overclocked. My problem with the poll is there are no ways to prove proof of ownership or failure. Your taking alot of anonymous data to heart from it.


*Edit* Heh I just tested it. I was able to vote 5x. And probably could do more if I felt compelled too. Pretty easy thing to manipulate.

I voted 6 times just by switching between IE and Firefox.
 
I was just about to post a link to that.

The only response I have to it is this:
rofl.gif
rofl.gif
 
Oddly enough I've had 4 friends with Nvidia hardware have their cards fail with no overclocking. 2x9800 GTXs, 1x8800 768 GTX, and I can't remember what the last guy had. Wasn't really a problem though they were all replaced under warranty and are happily running them right now.

Noone I know is using an 8500/8600 series GPU though, so can't comment on that. Most of us with laptops don't use them for gaming so most of us have a Intel IGPs in our laptops. [Edit] Whoops, one friend has an Apple notebook with an Nvidia graphics chip. He recently had to send it in for repairs/replacement when the graphics stopped working.

I have to say though that it's quite possible the desktop parts have the exact same problem as the notebook parts, but it's far less likely a desktop system is going to fail if it is indeed due to the issue of excessive heat/cool cycles and the extreme heat possible in a laptop compared to a desktop.

Regards,
SB
 
Quite many articles were released yesterday about the GPU failure.

Nvidia's Chip Troubles Worsen as Dell Extends Laptop Warranties
Nvidia, Dell, and HP respond to complaints of defective chips
Dell extends warranty on defective NVIDIA GPUs
Dell extends laptop warranties to cover duff Nvidia GPUs
Chip Problems Haunt Nvidia, PC Makers

From Wall Street Journal:

Nvidia hasn't recalled the affected chips or identified which models have problems. Dell Inc. and Hewlett-Packard Co., the two customers that have so far announced plans for coping with the problem, said they won't repair affected laptops until they fail.
The PC makers instead recommend updating internal software -- known as BIOS, for basic input-output system -- to adjust the speed of a laptop's cooling fan. That change doesn't guarantee the systems won't fail but is expected to reduce the likelihood of failure. In the meantime, the companies have extended their warranties; Dell on Monday added 12 months to its standard warranty -- usually one year for consumers and small businesses -- to systems affected by the problem. H-P previously had promised to fix affected computers for 24 months.
 
Status
Not open for further replies.
Back
Top