Unusual high failure rates for GeForce RTX 2080 Ti?

NVIDIA is confirming the issues with Turing based Cards....More info soon!!!

 

i hope it doesnt affect the 2070 what i ordered

 

maybe it only works for 10hrs i joke ofc lol

 

Nvidia's response has been posted online i will just let Hilbert break the news..

 

Technically speaking, they are not gimping older series. All they need to do is reduce amount of game specific optimizations for newly introduced games. And that's what did happen before many times. Then you had wave of angry owners all over internet and "magical" driver patch suddenly delivers optimizations.

So, no, no gimping of already released code. But as they invest less and less into each game optimizations for older series, they fall behind. And it can be seen today on cards like GTX 780Ti.

AMD on other hand gains very little from game specific optimizations. They do general architecture improvements which deliver gains across the board and therefore losses from AMD ignoring older cards are much smaller.
(read FineWine tm) Paradox is that it is not as much AMD who delivers great improvements to older cards, it is nVidia who stops doing new optimizations.

AMD had few notable driver sets which brought very noticeable jumps in performance, but even then their cards were not really much better than nVidia's once price was considered. So, when you see that nVidia's card falls behind AMD's which it was previously outperforming in everything, there is really no reason to praise AMD. Blame nVidia.

 

Eh - I think the issue with Kepler was mostly due to a shift in game development towards GCN's strengths. AMD's presence in nearly every console required game devs to squeeze every ounce of performance out of that architecture. So if you're writing a game and you say "well GCN has a bunch of compute capability, let's leverage that" and Kepler's compute sucks (it does) then the result is eventually all games coming out will play to AMD's strengths (FineWine) and into Kepler's weaknesses (Downgrade Debacle).

I don't really recall any magical driver uplifts either? There was definitely specific issues in games that Nvidia fixed after outcry but there was never any "downgrade debacle catch-all" driver.

Either way we haven't seen similar results with Maxwell/Pascal, despite most saying we would, mostly because Nvidia's architecture optimizations started following suit with AMDs.

 

That's what I wrote about. No global gimping. Just instances where nVidia "forgot" to optimize newly released game for older architecture.

 

They added a game to the card.
2080ti users can play Asteroids for free

 

For all users who have currently defect-free RTX cards...: Did you tried to run fixed math benchmark like GPUPI to see if there are some hidden defects?
Maybe run some test for heating card at same time.

 

Just Post it please.

 

while this is highly concerning, i'm heartened by the proud owners of the new cards.
bravo - you're braver than me.

imho, i suspect since these are massive chips with the concomitant low yield, that the binning process is flawed.

i seriously doubt that Nvidia screwed the micro-architecture.

 

Current statistics for this thread on reported users with 2080ti failed or not.
Total 2080ti self reported 26 cards, of the total 6 have failed for a 23% failure rate from this thread
Of the 26, total 2080ti FE edition cards reported 14, of which 5 have failed for a 35.7% failure rate for the FE model
of the 26, total non FE cards (other manufacture) 12 cards, of which on 1 has failed, never heard of brand and questionable explanation for a 8.3% failure rate if believed otherwise 0%.

Two posts SamuelL421 and asturur were duplicate posts different user. SJCWall and Promachos also duplicate post same user. Both FE version one saying good the other bad.

Based on this I would say most issues are related to the FE model with only a few from other manufacturers and likely related to OC'eing, Power or other hardware configuration issue.

 

Very funny!...

 

Good post. I was going to suggest a look at your PSU, but since the first card failed after you exited your games and were just sitting on the desktop, that would tend to rule out an over-stressed psu, I think. The only other thing I might have advised was to take the second card and plug it into your primary pcie slot--some motherboards won't even boot with only one GPU running in the secondary slot. But the symptoms you describe on the second GPU failure are exactly what happens when a GPU is overclocked (not by you in this case, but the factory, possibly) *and* the cooling of the GPU is insufficient--a matter of the card's component design/functioning. Good luck! Hopefully the replacements will run as advertised.

 

Artifacting can also be caused by faulty vram on the card, or when the vram is overclocked. I've dealt with myself in that past. I remember years ago buying a $2k cpu accelerator (MC68060-based) for my Amiga 4k (or was it for my 3k, can't recall), and I had to spend days on the phone walking that bad boy back through various departments before I hit satisfaction. Nothing is more disappointing than laying out a wad of cash for an expensive item--after reading many reviews and researching--only to find the problem was real but never mentioned anywhere else! Hope you get it sorted soon!

 

Indicative of bad vram--sure sounds like it.

 

More and more reports are starting to appear online , even reviewers like hardware unboxed have their RTX 2080Ti die on them as well , not looking good at the moment. If you own a RTX 2080Ti please test and re-test your video card , stress test it just in case to be sure is running alright would be my suggestion to avoid problems later on.

reviewer dead RTX 2080Ti and some important info:

 

Wow.

Can it be that due to NVidia's pricing, partners now have such low margins that they use crap components?

Edit: Nope. Strike that. It seems to happen with NVidia boards too.

 

All this makes me glad that I decided to boycott the 20-series. It's bad enough with the price hike, no RTX titles and mediocre performance improvements, but having a $1200 piece of hardware dying is the proverbial icing on the cake. Nvidia certainly does not deserve my hard-earned money for Turing, and I wouldn't be surprised if there was a recall on these cards.

 

It's only the FE cards that seem to be having issues. Even then, not every FE card. I'm sure those who have theirs die will simply get a replacement. This seems to be getting stretched out into a much bigger deal than it really is.