View Full Version : Modern Graphics Card Failure Rates
Ancient76
03-18-09, 06:25 PM
http://gpucafe.com/2009/03/modern-graphics-card-failure-rates/
http://i42.tinypic.com/r8dy6p.png
http://i41.tinypic.com/16arxxx.png
http://i43.tinypic.com/35j022p.png
Some vendors are missing.
=assassin=
03-18-09, 07:31 PM
Looks quite bad really, even 1% is too much, let alone rates many times that.
Cheers for posting!
BioHazZarD
03-18-09, 07:43 PM
so the most expensive cards has the highest rate ? something is wrong here..
so the most expensive cards has the highest rate ? something is wrong here..
3% is considered the target, or max, for modern mainstream electronics. However, high end cards are meant to be pushed to the limit, so there generally will be a higher failure rate.
Where's EVGA??
It was a European sample, they're not as big over there.
Personally I don't trust "un-named sources".
Too easy to be be mislead by FUD spread by competitors or other interested parties. (see any "article" by Charlie D. on The Inquirer for example)
AMD/ATi is literally on the edge of bankruptcy after losing almost $4b last year, I wouldn't put it past them (or someone who fears their demise) to plant this FUD.
Ancient76
03-19-09, 09:37 AM
I think it's FUD to.
How many people have had problems with GTX 280 on this forum? First examples were problematic, but not today.
wysiwyg
03-19-09, 10:08 AM
i thought the 8800gt had lots of failure rates when released
i thought the 8800gt had lots of failure rates when released
Only if you believe The Inquirer and their unnamed (FUD) sources.
Stakes are high, the game is getting interesting. Faced with the harsh realities and losing your job in these times, how many can say they wouldn't plant FUD like this?
gstanford
03-20-09, 01:07 AM
Most modern cards are highly reliable and resiliant (anything manufactured by Sapphire is just downright flimsy).
A while back I purchased a geforce 3 off of ebay for my collection. I'm in Australia, the clown of a seller was in the UK. Said clown shipped me the card in nothing more than a (very thin) padded postage bag.
http://img12.imageshack.us/img12/210/1dscf0676j.jpg
http://img8.imageshack.us/img8/4972/1dscf0681.jpg
The card still worked perfectly well (used in a system with a very good PSU installed), and I've since had it repaired.
3% is considered the target, or max, for modern mainstream electronics. However, high end cards are meant to be pushed to the limit, so there generally will be a higher failure rate.
So when I buy a mobo, psu, gfx card, hdd, cpu, video card, ram, optical drive, and monitor, that's 9 items.
1 - ((1 - 0.03) ^ 9) => 24.0% chance that one of the components needs an RMA! (And I did get 2 things that needed an RMA, the lcd monitor and the video card, go figure.)
Absolution
03-20-09, 01:19 AM
I've never seen a graphics card DOA, much unlike MB's and ram. I'm calling BS.
wysiwyg
03-20-09, 06:41 AM
Only if you believe The Inquirer and their unnamed (FUD) sources.
Stakes are high, the game is getting interesting. Faced with the harsh realities and losing your job in these times, how many can say they wouldn't plant FUD like this?
i had two dodgy 8800GT's when they were released
i had two dodgy 8800GT's when they were released
If the target failure rate for products like this is 3%, you could conceivably get two defective samples of any card.
A lot of people had no problems with their 8800GTs, or GTX280s.
LordJuanlo
03-20-09, 07:50 AM
As I posted in other thread, GPU cafe got this data from hardware.fr, and it's one of the most reputed French tech sites, but it may not be very famous outside France. They did a similar research to check hard drive failure rates, you can check it here (http://www.hardware.fr/news/lire/13-03-2009/), being Samsung the worst with a 2.8% and Western Digital the best with 1.2%. I consider those guys a reliable source.
Faulty GPUs cost nVidia $43.6 million (http://www.tomshardware.com/news/Nvidia-GPU-Damaged,7296.html), I think most of them are from the infamous notebook GPUs.
Certain notebook configurations of these MCP and GPU products are failing in the field at higher than normal rates. While we have not been able to determine a root cause for these failures, testing suggests a weak material set of die/package combination, system thermal management designs, and customer use patterns are contributing factors.
As I posted in other thread, GPU cafe got this data from hardware.fr, and it's one of the most reputed French tech sites, but it may not be very famous outside France. They did a similar research to check hard drive failure rates, you can check it here (http://www.hardware.fr/news/lire/13-03-2009/), being Samsung the worst with a 2.8% and Western Digital the best with 1.2%. I consider those guys a reliable source.
Faulty GPUs cost nVidia $43.6 million (http://www.tomshardware.com/news/Nvidia-GPU-Damaged,7296.html), I think most of them are from the infamous notebook GPUs.
What do you consider hardware.fr a "reliable source" based on Lord Juanlo? Have you been reading them many years and found what they said to be true most of the time? Or does agreeing with them serve your pro-ATi agenda?
As far as the Tom's article goes, we've all known for some time that some laptop chips had a higher failure rate and that NVIDIA and their partners did the right thing and offered extended warranties. That article just tells us that that only 1/5 of the money they set aside was used last year, suggesting the problem may be for less common than even NVIDIA thought.
The problem with hardware.fr's article is that is uses an "unnamed source" so we have no way of verifying it. It boils down to "some guy said". What if the guy applied for a job at NVIDIA, got told to get lost, went back to his job at the computer store shipping dept. and sent these cooked statistics on company letterhead for revenge?
What if the returns numbers are valid, but the GTX280 has the highest number because people returned them due to the highest cost?
Point being, data from one unnamed retailer in a small foreign country isn't exactly what I'm going to assume is the worldwide trend.
wysiwyg
03-20-09, 08:45 AM
http://tbn1.google.com/images?q=tbn:zgIJlgiy3h0QOM:http://lh3.ggpht.com/_jrwWiSVZBk0/RmGsyLsBdnI/AAAAAAAAAqc/g-3E-vNIn3I/Star%2BWars%2BCap%2B33.jpg
Ninja Prime
03-20-09, 10:47 AM
Good to see the fanboys out in force, questioning the source rather than the fact that nvidia made a lot of sh!tty cards.
LordJuanlo
03-20-09, 11:47 AM
Or does agreeing with them serve your pro-ATi agenda?
Anyone on this forum with half a brain would see that my agenda is not pro-ATI, but pro-truth. As I have posted a couple of times before, today I'm recommending people to buy nVidia cards if they don't have money problems. Today I have an ATI card, because it was the best, the fastest, when I bought it. My previous cards were a 8800GTX, a 7950GT and a 6800. And for god's sake, I will buy whatever card will be faster when the next generation is released, god damnit.
I'm not pro-ATI or pro-nVidia, I'm pro-ME, I buy what I consider the best and I couldn't care less about the brand. And if someone speaks bs about a product and I have something to say, I will say it, because I don't like people to get confused by lies from "brand fans".
Geez
john19055
03-20-09, 02:21 PM
I guess I have been lucky with graphic cards ,(Knock on wood)I have only had one ever go bad on me and it was a TI-4600 ,but all of my other cards have been good including ATI,Radeon 64DDR, 9500pro,9800pro,X1800XT,X1950XT and a HD-2900XT.and my last nvidia cards 2-7800GT,2-7900GTX,2-8800GTS,640mb version,2-9800XT,2-BFG GTX280 OCX.But I usally buy EVGA or BFG and they have always been good to me.
MaxBlade
03-20-09, 05:11 PM
Good to see the fanboys out in force, questioning the source rather than the fact that nvidia made a lot of sh!tty cards.
Hmm lol yeah thats it.. Well I can only speak from personal experience and I have yet to RMA one Nvidia card. Granted they have in the past made some.. questionable cards yet those have not failed. Yet with ATI I have only had to RMA one. So I would question the source and yet I am a fanboy lol (what a way to try to bate others) of none. I've been blessed.
So Evga, Asus, BFG blah blah have no accountability? Yet we all know be it graphics, soundcard, motherboard, CPU, anything like that all brake down. And please the odds that "oh I dont overclock anything it just stopped" you really believe that lmao? Now add those thousands if not millions to the source..then we start to see a different picture.
Ninja Prime
03-20-09, 08:32 PM
Hmm lol yeah thats it.. Well I can only speak from personal experience and I have yet to RMA one Nvidia card. Granted they have in the past made some.. questionable cards yet those have not failed. Yet with ATI I have only had to RMA one. So I would question the source and yet I am a fanboy lol (what a way to try to bate others) of none. I've been blessed.
So Evga, Asus, BFG blah blah have no accountability? Yet we all know be it graphics, soundcard, motherboard, CPU, anything like that all brake down. And please the odds that "oh I dont overclock anything it just stopped" you really believe that lmao? Now add those thousands if not millions to the source..then we start to see a different picture.
Take a stroll up about 5 threads and read the "overheating and other issues" thread for the 280, then come back.
Good to see the fanboys out in force, questioning the source rather than the fact that nvidia made a lot of sh!tty cards.
Problem with this is you say it's a "fact" NVIDIA made a bunch of ****ty cards based on a small sample from a unnamed source on a website I've never heard of in 10 years on the forums.
I'd say to me that means it's a "fact" you're a fanATic latching onto whatever negative info you can find about NVIDIA so you can chime in "It must be true as it's on teh intarweb!" in a lame attempt to help ATi avoid debtors prison another month. (lee)
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.