Will LTT do a piece on the crazy failure rates of 2080 ti video cards?

[Grimlakin]

Or perhaps the question really should be.  CAN LTT do a piece on the crazy failure rates of the 208 TI cards?

 

I suspect that NDA precludes any real investigation on LTT's part...  and really between them and the OCP site that is oh so Hard I would half expect to see some good points brought up.   You clearly have the staff and expertise if not on hand on call.  

 

So Linus and crew....  will there be a story or can you only publish what Nvidia allows in this matter?  (Or is it not something you feel is core to LTT's goals.)

[WereCat]

Not really their kind of content. 

Check GamersNexus for that stuff. 

[TheSLSAMG]

7 minutes ago, WereCat said:

Not really their kind of content. 

Check GamersNexus for that stuff. 

Adding to this, I can't imagine it would be in LTT's interest to speak in a largely speculative manner about what is an ongoing situation involving the graphics market's largest/most powerful actor.

[Mira Yurizaki]

Trying to get a company to say something about failure rates is like asking North Korea to back down on nukes. A better picture would be to get the cards that failed and analyze the cars to see what the problem really is.

 

However that sort of analysis requires a skillset that I don't feel the LMG staff has.

[Bcat00]

Gamers nexus already did a piece so go check them out

[Billy Pilgrim]

4 hours ago, TheSLSAMG said:

Adding to this, I can't imagine it would be in LTT's interest to speak in a largely speculative manner about what is an ongoing situation involving the graphics market's largest/most powerful actor.

Unfortunately. I feel like this is the case in many situations, either report and issue, or keep a sponsor.

[skunkmunkey]

It's also not really a case of crazy failure rates, GN and others have stated that most of the card suppliers have said it's still within their usual failure rate for new cards

[Mick Naughty]

Seems like normal rma rate. Doesn’t need to be covered. Half bad, half user error. 

[OttoVonBismarck]

1 hour ago, Mick Naughty said:

Seems like normal rma rate. Doesn’t need to be covered. Half bad, half user error. 

i feel like with the ease of the internet access these days that more and more people can jump on and say "wow look my stuff is broken" and it's really commonly painted pictures of issues being more widespread than they are. Not even related to GPUs or anything, its just something i've noticed in general

 

 

[LinusTech]

20 hours ago, Billy Pilgrim said:

Unfortunately. I feel like this is the case in many situations, either report and issue, or keep a sponsor.

Keeping a sponsor is nothing to do with it. I just haven't seen any evidence to suggest this is some kind of widespread issue. 

 

Retailers and partners would be the ones I'd be watching and so far not a peep. 

 

Bad units happen, and even if your RMA rate was 2-3% (typical for video cards from my experience), if you sell a million units, thats tens of thousands of returns. Doesn't mean there's anything unusual happening.. 

[Boomwebsearch]

On 11/23/2018 at 3:36 PM, Grimlakin said:

Or perhaps the question really should be.  CAN LTT do a piece on the crazy failure rates of the 208 TI cards?

 

I suspect that NDA precludes any real investigation on LTT's part...  and really between them and the OCP site that is oh so Hard I would half expect to see some good points brought up.   You clearly have the staff and expertise if not on hand on call.  

 

So Linus and crew....  will there be a story or can you only publish what Nvidia allows in this matter?  (Or is it not something you feel is core to LTT's goals.)

All cards have some failure rate which is associated with them because failure is possible although some factors of their manufacturing process can impact their rate of failure. I know that there have been some failed units of the RTX 2080 Ti although it would be hard to actually test and I do not think that Nvidia would want to sponsor that so, it would also be very costly to do. Usually, there is a warranty to some extent and protection plans, so if you are planning to buy one then you may want to consider factoring that in based on studies which actually show that certain models are proven to fail more although this is not an experiment which has been done on all GPUs.

[TahoeDust]

Watch Gamers Nexus video about it.  RMA rate is not abnormal.

 

Fake news.

[Billy Pilgrim]

53 minutes ago, LinusTech said:

Bad units happen, and even if your RMA rate was 2-3% (typical for video cards from my experience), if you sell a million units, thats tens of thousands of returns. Doesn't mean there's anything unusual happening..

It might be useful to make a video about this, not just about RTX, but about the fact that bad units happen etc.

[Mick Naughty]

Should be implied. If you are old enough to purchase stuff, you should understand a little how the world works.

[xg32]

it's not a wide spread issue, but since it's the first time we have a 1200 Ti card and it happens to run into some problems, people feel entitled to not have a brick. It was impressive how it garnered so much attention, and if anyone believed it was a major issue, they got fake news'd , though i'd be paranoid too if i had a 0818 FE card.

 

If you don't wanna look it up, take this with a grain of salt. t came down to a bad first batch of nvidia pcbs that mostly affected FE cards and a small portion of partner cards that used said pcb. The problem was narrowed down to a likely ram-related defect that's now likely fixed. Issues that did not involve space invaders appearing on screen were most likely one and done.

 

Some cards that were tested were sent in for driver issues (basically false positives), the 2 main driver issues are still present to-date, flickering on multi-monitor, and bsods when running g-sync with specific monitors.

 

If we take into account that a decent number of people likely sent in cards for rma for the multi-monitor and gsync issues also, the actual number of space-invader cards have to be exceptionally low, and mostly limited to FE.

[Grimlakin]

On 11/24/2018 at 3:13 PM, LinusTech said:

Keeping a sponsor is nothing to do with it. I just haven't seen any evidence to suggest this is some kind of widespread issue. 

 

Retailers and partners would be the ones I'd be watching and so far not a peep. 

 

Bad units happen, and even if your RMA rate was 2-3% (typical for video cards from my experience), if you sell a million units, thats tens of thousands of returns. Doesn't mean there's anything unusual happening.. 

I know this may seem anecdotal, and it is in large part.  But over on a different forum and site the site owner has had two fail back to back while one remains good.  Many are reporting they are seeing multiple failures within days/weeks of getting RMA's, and many are reporting RMA's as well.  (of course forums are the outliers.)  But retail outlets are being told to keep quiet about the return rates.   One shared experience is that a Microcenter store had stacks of returned 20xx series cards for being faulty behind a counter.  Guest asked about them and learned from a manager they are seeing failure/swap/refund rates easily over 20% for the 20XX series of cards.

 

I find that interesting, while it is anecdotal I udnerstand that you would need to have some other form of proof to act.    Question though.... have your team been actively using the 20xx series cards in gaming and in a always on system?   Any failures or is all well?   Just curious. 

[Grimlakin]

On 11/24/2018 at 4:30 PM, xg32 said:

it's not a wide spread issue, but since it's the first time we have a 1200 Ti card and it happens to run into some problems, people feel entitled to not have a brick. It was impressive how it garnered so much attention, and if anyone believed it was a major issue, they got fake news'd , though i'd be paranoid too if i had a 0818 FE card.

 

If you don't wanna look it up, take this with a grain of salt. t came down to a bad first batch of nvidia pcbs that mostly affected FE cards and a small portion of partner cards that used said pcb. The problem was narrowed down to a likely ram-related defect that's now likely fixed. Issues that did not involve space invaders appearing on screen were most likely one and done.

 

Some cards that were tested were sent in for driver issues (basically false positives), the 2 main driver issues are still present to-date, flickering on multi-monitor, and bsods when running g-sync with specific monitors.

 

If we take into account that a decent number of people likely sent in cards for rma for the multi-monitor and gsync issues also, the actual number of space-invader cards have to be exceptionally low, and mostly limited to FE.

Actually Kyle over at the other site already did a return, first card was brand X of member, new card was Samsung.  Samsung died in 8 hours before any real torture testing could be done.  Other card (3rd 2080ti in this case) is still good.  His personal experience is 66% failure rate.   I find that interesting.   Nvidia's own posting about it stated a .01% failure rate of cards that snuck through.

 

For Kyle to get two back to back... not to mention several other members there claiming the same..  (lets say 50% of them are true to be fair.)   That's still a CRAZY high number of failures to be generated.   But I appreciate your outlook I just think it is worse than Nvidia is claiming.  Hopefully I'm wrong...  but I'll be skipping this gen anyway.