• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

NVIDIA Confirms Issues Affecting Early Production Run of GeForce RTX 2080 Ti Graphics Cards

The lie was specifically that the 970 had the "same memory subsystem" as the 980.
 
People need to stop spinning the lie that Nvidia is somehow evil, and stop spamming every thread with it. Nvidia is no more evil or dishonest than AMD, and I could have given a long list of cases where AMD have been deceitful or outright lying, but I'm going to restrain myself and not do so, because it's off topic.

Back to topic;
There is nothing wrong with buyers expecting their product to work, but this doesn't mean that every failed sample is proof of a fundamentally defective product. Every mass-produced piece electronics have a failure rate >0, while the failure rate may vary between products, a typical DOA or short-time failure rate of 1-2% and 4-5% over three years, is completely normal for graphics cards.

The other day there was one guy on Youtube which experienced the same type of artifacting on a RTX 2070 as the famous defective RTX 2080 Tis, implying a bigger scope for the "Turing problems". But this looks like just a normal defective card, how come all these "tech people" know so little about the fact that some samples are just defective?
Here in the forums, one guy got the exact same symptom on a GTX 1080. Every series of graphics cards have a failure rate, even AMD's ones, that's why we have warranties to cover such problems. The existence of a few bad samples doesn't prove a bigger problem, but an abnormal high failure rate would do. So far, most complaints seem related to driver issues and relatively few cards have actually been returned.
 
Last edited:
People need to stop spinning the lie that Nvidia is somehow evil, and stop spamming every thread with it. Nvidia is no more evil or dishonest than AMD, and I could have given a long list of cases where AMD have been deceitful or outright lying, but I'm going to restrain myself and not do so, because it's off topic.

Back to topic;
There is nothing wrong with buyers expecting their product to work, but this doesn't mean that every failed sample is proof of a fundamentally defective product. Every mass-produced piece electronics have a failure rate >0, while the failure rate may vary between products, a typical DOA or short-time failure rate of 1-2% and 4-5% over three years, is completely normal for graphics cards.

The other day there was one guy on Youtube which experienced the same type of artifacting on a RTX 2070 as the famous defective RTX 2080 Tis, implying a bigger scope for the "Turing problems". But this looks like just a normal defective card, how come all these "tech people" know so little about the fact that some samples are just defective?
Here in the forums, one guy got the exact same symptom on a GTX 1080. Every series of graphics cards have a failure rate, even AMD's ones, that's why we have warranties to cover such problems. The existence of a few bad samples doesn't prove a bigger problem, but an abnormal high failure would do. So far, most complaints seem related to driver issues and relatively few cards have actually been returned.
QFT...because thanks isnt enough.

The problem, imo, is the overall knowledge level and maturity of some forum users. It's the same. ones. every. time. Even long standing members fall into this hole. While the issue is seen everywhere, people's opinions here seem to be more polarizing and extreme.
 
People need to stop spinning the lie that Nvidia is somehow evil, and stop spamming every thread with it. Nvidia is no more evil or dishonest than AMD, and I could have given a long list of cases where AMD have been deceitful or outright lying, but I'm going to restrain myself and not do so, because it's off topic.

Back to topic;
There is nothing wrong with buyers expecting their product to work, but this doesn't mean that every failed sample is proof of a fundamentally defective product. Every mass-produced piece electronics have a failure rate >0, while the failure rate may vary between products, a typical DOA or short-time failure rate of 1-2% and 4-5% over three years, is completely normal for graphics cards.

The other day there was one guy on Youtube which experienced the same type of artifacting on a RTX 2070 as the famous defective RTX 2080 Tis, implying a bigger scope for the "Turing problems". But this looks like just a normal defective card, how come all these "tech people" know so little about the fact that some samples are just defective?
Here in the forums, one guy got the exact same symptom on a GTX 1080. Every series of graphics cards have a failure rate, even AMD's ones, that's why we have warranties to cover such problems. The existence of a few bad samples doesn't prove a bigger problem, but an abnormal high failure would do. So far, most complaints seem related to driver issues and relatively few cards have actually been returned.
Man, we can't even rant about stuff on the Internet because all of a sudden you start talking sense :(

Here's a little exercise I like to pull up whenever various people start discussions like this (especially about electronics):
1. think about something dead-simple, like a paper airplane
2. start building 100 of them. identical
3. now realize you're asking someone else to arrange billions of transistors (plus all the components around them) in hundreds of thousands of products. flawlessly

Note 1: It can be done. But then no one would afford to buy the product, because it would be like a Rolls Royce.
Note 2: Yes, it still sucks every single time you end up drawing the short straw. Just don't let your anger get the best of you, keep your perspective.
 
Equipped with 13 SMX units and 1664 CUDA Cores, the GeForce GTX 970 also has the rendering horsepower to tackle next generation gaming. And with its 256-bit memory interface, 4GB frame buffer, and 7Gbps memory, the GeForce GTX 970 ships with the same memory subsystem as our flagship GeForce GTX 980, allowing gamers to crank up the settings and resolutions in graphics-intensive games like Assassin’s Creed: Unity and still enjoy fluid frame rates.

http://webcache.googleusercontent.c...&cd=12&hl=en&ct=clnk&gl=us&client=firefox-b-1

Just a clarification, no spin here.
 
You could have edited that in your other post, lol... but yeah, we know this (though NVIDIA called it a marketing snafu as IIRC, they showed their whitepapers then sent did NOT state that).

Either way, the polarizing and misinformed opinions here are tiresome.


EDIT: One thing to note, that wasn't a statement from NVIDIA you quoted... that was from some OCdrift website....
 
2014 is along time ago, I'll admit but that text was standard marketing material for the 970. Just happened to find it on a cached page of OCdrift via google. Back then, every press site was spewing the same info, before the "revision".
 
Right. We know there was an issue. The point was, it has 4GB. They didn't lie about the quantity, just did not mention/lied about the last 512MB being slower.
 
Agreed. Waiting 6 months or more for companies to put out a finished product and then buying it, it just sends the wrong signals.
While that can be true in many cases, hardware is a bit different. Like Intel's MMX CPU's. It was a tech that went unused for almost a year. RTX is no different, devs need time to learn and properly utilize it. With time it will become a very cool thing that everyone will want.

Note 1: It can be done. But then no one would afford to buy the product, because it would be like a Rolls Royce.
Note 2: Yes, it still sucks every single time you end up drawing the short straw. Just don't let your anger get the best of you, keep your perspective.
Note3: If you do draw the short straw, exchange it through the proper channels or get a refund and try again.
Note4: NVidia and it's AIB's are not like Microsoft with the XB360 RROD problem, they're going to set things right because they don't want a bad reputation.
 
While that can be true in many cases, hardware is a bit different. Like Intel's MMX CPU's. It was a tech that went unused for almost a year. RTX is no different, devs need time to learn and properly utilize it. With time it will become a very cool thing that everyone will want.
That's true. I only meant we shouldn't put up with unfinished products (see the discussion about some Ryzen mobile laptops getting nothing after the initial drivers and even those are supposedly rather problematic). Sure, not everything is optimized form the start. But it should still work.


Note3: If you do draw the short straw, exchange it through the proper channels or get a refund and try again.
Note4: NVidia and it's AIB's are not like Microsoft with the XB360 RROD problem, they're going to set things right because they don't want a bad reputation.
It still stings while you've parted with your cash and are waiting for the replacement.
 
an old thread i know but after just over two months use my own Palit gaming pro oc 2080ti has just died on me...

over a two to three day period it went from the odd game lock up to an instant crash to the desktop when something like the heaven benchmark is started.. this is with the memory speeds reduced from 7000 to 6000.. the card is now unusable..

the card worked fine for a couple of months its now dead for anything except browsing purposes.. no flames or anything dramatic.. it just dosnt work any more..

this isnt a "new" card not working.. just a rather expensive product failing after just a couple of months use.. i dont think this story is over yet..

trog
 
an old thread i know but after just over two months use my own Palit gaming pro oc 2080ti has just died on me...

over a two to three day period it went from the odd game lock up to an instant crash to the desktop when something like the heaven benchmark is started.. this is with the memory speeds reduced from 7000 to 6000.. the card is now unusable..

the card worked fine for a couple of months its now dead for anything except browsing purposes.. no flames or anything dramatic.. it just dosnt work any more..

this isnt a "new" card not working.. just a rather expensive product failing after just a couple of months use.. i dont think this story is over yet..

trog
Palit honors their warranties. Get a replacement.
 
Palit honors their warranties. Get a replacement.

in the UK we deal with the retailers who are obliged by UK consumer law to warranty stuff.. i dont think i will have problems getting a replacement card..

what is surprising me is just how well the borrowed single 1070 card i now have in my machine works.. :)

i am just waiting for an rma number before sending the dud 2080ti card off..

trog
 
in the UK we deal with the retailers who are obliged by UK consumer law to warranty stuff.. i dont think i will have problems getting a replacement card..

what is surprising me is just how well the borrowed single 1070 card i now have in my machine works.. :)

i am just waiting for an rma number before sending the dud 2080ti card off..

trog

Overkill is never notuceable and that is whaf youve been having most of the time ;)
 
in the UK we deal with the retailers who are obliged by UK consumer law to warranty stuff.. i dont think i will have problems getting a replacement card..
Did not know that. Very cool.

Overkill is never notuceable and that is whaf youve been having most of the time ;)
Either you were on a mobile device or you were very tired when you typed that out.. LOL!
 
i now have an rma arranged for my two month old palit gaming pro oc 2080ti card.. having spoken with an Ebuyer UK service engineer they are aware of issues with these cards and how they are failing he seemed apologetic and did say they were getting less palit cards back than others..

i confirmed a pretty standard UK warranty period of two years and this will start again from when i receive the replacement card..

having checked around the internet i cant find any comment regarding 2080ti failures later than mid November.. its just as if the problem has gone away but it quite clearly hasnt..

the problem seems to be a memory issue artifacting and lock ups or crashing to the desktop..

i assume sooner or later whatever is causing the problem will get fixed at source and we will know a little more..

my advice to anyone that has one of these cards is to check your warranty.. you may need it..

trog
 
i now have an rma arranged for my two month old palit gaming pro oc 2080ti card.. having spoken with an Ebuyer UK service engineer they are aware of issues with these cards and how they are failing he seemed apologetic and did say they were getting less palit cards back than others..

i confirmed a pretty standard UK warranty period of two years and this will start again from when i receive the replacement card..

having checked around the internet i cant find any comment regarding 2080ti failures later than mid November.. its just as if the problem has gone away but it quite clearly hasnt..

the problem seems to be a memory issue artifacting and lock ups or crashing to the desktop..

i assume sooner or later whatever is causing the problem will get fixed at source and we will know a little more..

my advice to anyone that has one of these cards is to check your warranty.. you may need it..

trog

Fingers crossed for a good clocker :) Its not always a bad thing to get a silicon lottery re-roll ;)
 
i shall compare the new one with the old one but after that it will probably get run at stock settings.. he he..

one thing i did notice.. down-clocking the memory - 1000 whilst watching the old card die on me didnt effect the performance that much.. in my case it did give me a few hours extra use before the card became totally unusable for 3D applications but it didnt knock the performance as much as i thought it would..

trog
 
Back
Top