RTX 2080 Ti FE Escapes Testing by Dying After 8 Hours @ [H]

Well I am trying to beat the ever loving Jesus out of my card. Ran Haven for 18 hours straight , played about 8 hours on it, and ran Haven for another 24 hours earlier this week, if he is in there I'll find him.

Going to give the system a little rest, impressively the temp of the GPU still shows 72c after hours of use.
 
We may just see a market for 2k series heavily tested cards. Ran heaven benchmark for 2 weeks straight, no space invaders! Nvidia should start just shipping you double what you order, since there's a pretty good chance you'll need an rma in the first week anyway.
 
this is got to be costing Nvidia in sales right now worst possible time. No excuse.......
 
I wonder if this impacts their professional line as well?

Maybe, but you usually buy that type of hardware through vendors that include pretty substantial warranty service with it such as next-day on site replacement so it's a problem for nvidia/vendors but the professional is taken care of. It's part of the reason why the pro cards cost so much.
 
Maybe, but you usually buy that type of hardware through vendors that include pretty substantial warranty service with it such as next-day on site replacement so it's a problem for nvidia/vendors but the professional is taken care of. It's part of the reason why the pro cards cost so much.

But if they did, it would evidence something wrong with the underlying architecture of Turing.
 
But if they did, it would evidence something wrong with the underlying architecture of Turing.

Maybe, but I really doubt Nvidia screwed up with the GPU core itself. This just seems to be a manufacturing issue related to the PCB.
 
Eh, if I had a choice between failing within a month or a year I’d take a month.

At least now people who were waiting for Christmas have the chance to be informed and put off this large purchase in lieu of something like a new Diamond ring for the wife or significant other.
 
Maybe, but I really doubt Nvidia screwed up with the GPU core itself. This just seems to be a manufacturing issue related to the PCB.

Considering it has been impacting AIB vendors as well... What we need now is a list of vendors who's boards follow exactly Nvidia's design, and those that chose to do better or different. THEN we can work on a comparison of failure rates that makes sense.

If it is a board design issue the AIB vendors with custom boards should be largely avoiding it. IF they are custom boards and still having the same issue, as we now knew new memory isn't a preventative, that points to something on the DIE itself. That means it is going to be a VERY expensive problem for Nvidia to fix.

But as another has stated. Nvidia's value is easily 4x AMD's. They can absorb the hit with their consumer grade cards to do this right. The question is... will they?
 
Correct me if i'm wrong - But I thought it was still too early in the game and all the 2080ti's are still the nvidia reference PCB. There isn't a true custom board yet?
 
Correct me if i'm wrong - But I thought it was still too early in the game and all the 2080ti's are still the nvidia reference PCB. There isn't a true custom board yet?
I honestly don't know. Perhaps?

In some quick searching there are models that ARE using custom PCB designs.
 
After 2 RMAs on MSI RTX 2080 Duke card my wife's 3rd card now lives on its 2nd week. My own Gainward RTX 2080Ti I RMA'ed I got a full refund for, and I bought 2x GTX 1080Ti's again, they just work.
I am not buying more 1st generation RTX cards.
 
After 2 RMAs on MSI RTX 2080 Duke card my wife's 3rd card now lives on its 2nd week. My own Gainward RTX 2080Ti I RMA'ed I got a full refund for, and I bought 2x GTX 1080Ti's again, they just work.
I am not buying more 1st generation RTX cards.

A theory on another site in regards to partner cards was that the ones using the reference design from NV might also be susceptible. They did specifically mention that MSI and Gainward might have problems if this is true.
 
Quick question. Is this problem also apparent on the RTX 2070s?

Most of the issues I read about are regarding 2080tis. Would these issues be fixed for new models of cards just coming into stock or is that still an unknown?
 
Quick question. Is this problem also apparent on the RTX 2070s?

Most of the issues I read about are regarding 2080tis. Would these issues be fixed for new models of cards just coming into stock or is that still an unknown?

This will most likely depend on how the cards themselves are binned. If 2070's are just software disabled versions of 2080ti's then no. But IF they are different what is important to figuring out the cause here is how are they different.

Meaning if the 2080 ti shares base core function and design... and that isn't a problem on the 2070 then we have to look at what is different.

I don't have the time now to play that game of comparison but perhaps someone else will explore this rabbit hole?
 
Anyone try a 2080 Sea Hawk hybrid?

Nevermind. I looked at the backplate of the MSI 2080 Sea Hawk X and the screw pattern is identical to the Founders cards backplate. Must be reference design with a hybrid cooler. Damn.
 
It would be interesting to see how Marseille mCable on a RTX might work up scaling from 720p and comparing it to 1080p along with comparing the relative FPS between the two. The results might be rather surprising potentially those up scaling cables work wonders at lower resolutions so I doubt in the case of RTX ray traced graphics that would change much it should still be very noticeable on the other hand the performance difference between 720p and 1080p for ray trace graphics I'm sure is substantial enough. It would sort of be like testing and pitting one piece of promising yet a bit admittedly gimmicky piece of tech against the other one in a $$$ cage match winner takes all.
 
FWIW, I was at my local Micro Center yesterday and I saw LARGE stacks of various brands of GTX2080 and Ti versions, all in return status.

I inquired with one of the managers about it, and they are overrun with exchanges and returns on the cards, for the exact same reasons as stated in this article. The manager said most customers will try one exchange and then be back in a couple of days for a full refund once they figure out the replacement card manifests the same issue. I did not get a chance to ask about 2070 being affected...she was psycho busy with Black Friday. She did estimate that the return rate exceeds 25%, and that return rate translated into almost an exact 25% refund rate as well.

I asked if I could take picture of the veritable mountain of cards, but was denied as they don't want to get into trouble with their corporate office.

The problem is much much more serious than even folks here have imagined thus far.

I am so glad I was turned off by the high prices and decided to keep my 1080Ti.
 
FWIW, I was at my local Micro Center yesterday and I saw LARGE stacks of various brands of GTX2080 and Ti versions, all in return status.

I inquired with one of the managers about it, and they are overrun with exchanges and returns on the cards, for the exact same reasons as stated in this article. The manager said most customers will try one exchange and then be back in a couple of days for a full refund once they figure out the replacement card manifests the same issue. I did not get a chance to ask about 2070 being affected...she was psycho busy with Black Friday. She did estimate that the return rate exceeds 25%, and that return rate translated into almost an exact 25% refund rate as well.

I asked if I could take picture of the veritable mountain of cards, but was denied as they don't want to get into trouble with their corporate office.

The problem is much much more serious than even folks here have imagined thus far.

I am so glad I was turned off by the high prices and decided to keep my 1080Ti.

You asked to take the photo? Just take it next time.
 
Meh, people complaining about poor Nvidia doing their best, just because new cards are failing... If this was Apple they would just declare "you're holding it wrong", and customers would be happy.
 
To the people who say that we can't gauge the true return rate on anecdotal evidence, I agree. However, if you look at even the anecdotes here from a statistical standpoint, the return rate has to be higher than 1 percent like nVidia says. There are multiple people on this site who have received more than one bad card. If the 1 percent rate was true, itshould be statistically impossible for almost anyone to get more than one bad card. Granted, I'm not a statistician, but I did have some statistics coursework in college.
 
Yeah, it would be interesting to see a rant from him on this issue if he was into GPUs.

I don’t think it matters, so long as he has the schematics and layout I’m sure he could figure it out.

He might do it for free because this is a huge news story atm and a cross promotion between him and Kyle would increase his traffic from a different market.
 
I had thought about getting one for (relatively cheap) access to tensor cores, but looks like it's not worth the headaches.
 
The pricing of the card is beginning to make a lot more sense; they had to factor in a rather large buffer for RMA'd products...SO glad the [H] didn't sign that NDA!
 
A 2080ti dying live on twitch last night while a streamer was playing Farm Simulator 2019 (starts about 6-7s early):

https://www.twitch.tv/videos/339878940?t=06h41m41s

Space Invaders attacking Farm Sim '19. This is not looking good Nvidia. Not good at all.
Ken Lawson having the Way It Meant To Be Played experience there - I cannot think of any PC gamer wanting to have an unexpected change from gaming to hardware gymnastics with Windows 10. If the rate of failure for 2080 Ti's are this frequent, over time it will most likely get worst at a faster rate before it lowers. Not sure what threshold it will take before Nvidia pulls the plug and offers full rebates, these do not appear to be driver related but hardware. If we just had more exact information what exactly is failing, what causes it and if fixable or not or at what it will take to fix would help us consumers make a better decision. As of right now I can't think of any reason to recommend a 2080 Ti until this issue are clearly explain, extent, fix etc. Maybe Nvidia should go back to making Pascals in the mean time.
 
I'm thinking the early jump to GDDR 6 isn't working out so far. Maybe the memory controller or the GDDR itself?

My FE is kicking ass since 5 October, but it's sad to see so many here having failures.
 
Was about to pull the trigger on one but just couldn't finish the checkout. It's such a bad deal and can't be justified. $1350 for a video card was just too much for something that might blow up after a few hours of use.
 
I'm thinking the early jump to GDDR 6 isn't working out so far. Maybe the memory controller or the GDDR itself?

My FE is kicking ass since 5 October, but it's sad to see so many here having failures.
Only a matter of time. ;)
 
Back
Top