Info [PSA] Amazon's new game reportedly causing GPUs to brick

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Thibsie

Senior member
Apr 25, 2017
727
752
136
Well, we'll see but remember something from a slightly different point of view.
Amazon, years ago bought Dpreview. Dpreview was but is * a lot* more since then trying to mush new ***** for you to buy, good reasons or not.

Will Amazon deliberately not optimized too much, so people will buy new cards ?
Frankly, I wouldn't be surprised.
 

GodisanAtheist

Diamond Member
Nov 16, 2006
6,715
7,004
136

In case anyone gets into the beta, be careful. Looks like it is just 3090s right now but probably better safe than sorry even if you have a different card.


-SOUNDS LIKE THE BIGGEST SCALPERS OF ALL ARE GETTING DESPERATE! BEZOS MOON BUCKS AREN'T GOING TO PRINT THEMSELVES!!!

*Dances naked wearing only a tinfoil hat*
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
I gotta say a game should not kill a GPU at stock speeds, provided the case has even somewhat decent airflow and the ambient temperatures aren't too high (falling within the card's stated operating temperature).

If it does, then the manufacturer either didn't design it properly (or their is some kind of defective component they were unware about).

This story really brought back the issue EVGA had a few years ago where they didn't have thermal pads on some of the PCB components, and the temps got way too high.

It's just a symptom of the very top end cards pushing the limits to be fastest, if you get one of those then usually you're ok but they're much closer to the limit than lower end cards so high end games/mining/etc can possibly cause you issues.
 
  • Like
Reactions: Leeea

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
It's just a symptom of the very top end cards pushing the limits to be fastest, if you get one of those then usually you're ok but they're much closer to the limit than lower end cards so high end games/mining/etc can possibly cause you issues.
I understand mining (or Folding for that matter) as it pegs the GPU at very high levels 24/7.

But if playing a game at the video card's stock settings damages it, then the default speeds should be lower out of the box.

But who knows for sure at this point what exactly is causing the claimed issues the have suddenly popped up? It could be a hardware/design issue, or hell I guess it could even be something a disgruntled Amazon employee slipped into the game code (like a "power virus" OCCT type thing).

Hopefully they will figure it out quickly, and fix whatever the issue is. I personally believe if it was a true hardware defect issue, this sort of thing would have popped up all over the internet by now since the cards have been out for a while, and they definitely have been pushed hard by all the miners. It could end up being something like just some bad batches of cards that used problematic components (much like the bad capacitor issues on motherboards in the past).
 
  • Like
Reactions: lightmanek
Feb 4, 2009
34,494
15,729
136
Im having a hard time believing this story.

I am thinking idiots built the system wrong with improper cooling or used a POS power supply, or its a OEM of HP or something.

I do know the rear ram on the 3090 gets insanely hot.
Too hot sometimes ive seen it spike 80C before i loaded the rear backplate with heatsinks and have a dedicated fan to it now.

But i think a long time ago Starcraft 2 had the same issue...

Idiots who do not know how to properly build a PC should not be playing with 2400 dollar videocards.
At the very least, change the pads like miners do. If anything again, people do not know how to keep the cards cool because they just do not know how fix a overheating problem.

Also GPU's Throttle, so there is no way you wont notice things shuttering before it gets terminal, if its heat related.
So if they are dying, its the VRAM on the rear i highly suspect.



Get a waterblock for your 2400 dollar card....
Problem solved...

id really like to see you brick this card with temps like this unless you used a very poor powersupply, or your card was just faulty.
View attachment 47511

Shouldn’t a high end/Uber card be automated enough to not cook itself?
I mean someone is smart enough to build and assemble and load and over clock the card but they are not smart enough to keep it cool? Also shouldn’t a $2k plus card shut down before it cooks itself?
 
Last edited:

bryanW1995

Lifer
May 22, 2007
11,144
32
91
But who knows for sure at this point what exactly is causing the claimed issues the have suddenly popped up? It could be a hardware/design issue, or hell I guess it could even be something a disgruntled Amazon employee slipped into the game code (like a "power virus" OCCT type thing).
Amazon has no disgruntled employees, I mean Bezos just thanked them all (and all customers too) for paying for his joyride yesterday...they must all be so happy!

Shouldn’t a high end/Uber card be automated enough to not cook itself?
I mean someone is smart enough to build and assemble and load and over click the card but they are not smart enough to keep it cool? Also shouldn’t a $2k plus card shut down before it cooks itself?

It's just a few people having the issue now. Should it shut itself down before cooking? Yes ofc, but if it's running very close to the margin, as those very high end prestige cards usually are, then there's less margin for error and a few of them might just not shut off fast enough. I mean there aren't even that many 3090's out in the wild these days. If I had to bet I'd say that aigomorla is right and it's just some bad luck on a few cards that were too close to the limit combined with some poorly-designed rigs combining to push cards over the edge, just like the SC2 thing was.
 
  • Like
Reactions: Leeea

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Maybe NVidia will be in for another class-action, for designing/building/selling video cards with GDDR6X memory running at nearly 100C under normal load. Sounds ridiculous to me.
 
  • Like
Reactions: bryanW1995

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
This seems like an isolated case with a few cards. The card will just normally throttle if it gets too hot.

its the rear VRAM on the 3090.
Im 90% sure it has to be that, as i did have issues with it, as well many many others.
The solution to it was replace the stock pads, put a fan.

That kept the rear vrams from skyrockting to 90C + and keeping them nice and cool at around 70C.
I can see the VRAM's dying under long term load at 90C+... a lot of factors can contribute to this which is also lack of ventilation in the case to allow proper heat exchange...

Again again, the VRAM's are a problem and have been discussed MANY MANY times, as well as a fix.

I might try adding some sinks on mine. I'm not sure if the AIB cards have the same issue but the backplate does get very hot during games. I have a Corsair temp sensor on it and it goes over 80C at times. Maybe it means the thermal interface is working though.
 

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
This is really odd to say the least. I almost wonder if it has absolutely nothing to do with the game itself and it simply comes down to a large number of people all playing this at once (because it's new) and failures that would have normally happened in this time period almost regardless of game are being noticed as a pattern that doesn't actually exist.

If it isn't something like that which is leading us to draw incorrect inclusions we really have to ask what the game could be doing to cause something like this that shouldn't be possible. There are applications like Furmark that are intentionally designed to stress the hell out of a GPU and modern GPUs are a lot better designed to throttle in these situations so they don't become expensive paperweights.
 
  • Like
Reactions: Leeea

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
20,839
3,171
126
This seems like an isolated case with a few cards. The card will just normally throttle if it gets too hot.

Thats the thing... but the VRAM i don't know... it should artifact and bluescreen unless extreme damage was done on those rear ram chips because you never addressed the overheating issue from the start.

There are applications like Furmark that are intentionally designed to stress the hell out of a GPU and modern GPUs are a lot better designed to throttle in these situations so they don't become expensive paperweights.

Which is also called Tuning... trying to see what the temps are and seeing if it will throttle, but posters here are going after me for saying you shouldn't have to tune especially on a 2400 dollar card..
:rolleyes:

@Shmee if you haven't done it, heat sink your back plate before you also join this crowd.
Trust me on it, seeing how you have the same card i do.
I like Enzo branded heat sinks, they look a lot blingier then the ones meant for R-Pi.

 
Last edited:

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
Which is called Tuning... but 1/3rd of the posters here are flaming me for saying you shouldn't have to tune especially on a 2400 dollar card..

I'm asking what's this game is doing that a program that's practically designed to push a card to the point where it's close to bricking isn't? That's why it doesn't make sense to me. I'll admit that maybe it's possible that an intern copy-pasted something from the Necronomicon into the program code or something like that, but I think it's more likely to be a series of unrelated events that individually have small probability but appear related because a lot of people are trying this game out at the same time enough of them occurred to get people talking.

Really you shouldn't have to tune any GPU out of the box. I don't really see what price has to do with it because someone could just make your same argument for a $1,200, a $600, or even a $300 card. However, I'm not sure why you'd buy a 3090 and not want to play around with it just to see what it's fully capable of, or what kind of performance you can get at some particular fan settings, or just because it's fun to dink around with!
 
  • Like
Reactions: Leeea

bryanW1995

Lifer
May 22, 2007
11,144
32
91
I'm asking what's this game is doing that a program that's practically designed to push a card to the point where it's close to bricking isn't? That's why it doesn't make sense to me. I'll admit that maybe it's possible that an intern copy-pasted something from the Necronomicon into the program code or something like that, but I think it's more likely to be a series of unrelated events that individually have small probability but appear related because a lot of people are trying this game out at the same time enough of them occurred to get people talking.

Really you shouldn't have to tune any GPU out of the box. I don't really see what price has to do with it because someone could just make your same argument for a $1,200, a $600, or even a $300 card. However, I'm not sure why you'd buy a 3090 and not want to play around with it just to see what it's fully capable of, or what kind of performance you can get at some particular fan settings, or just because it's fun to dink around with!
It's the same reason the guy with ferrari doesn't give a shit about its 1/4 mile time, or brand of tires, or hp, or the perfect shift points, etc etc...I mean, sometimes people who buy the highest of the high end do it out of enthusiasm, but usually it's just b/c it's the best and they want it to accomplish something. Like high fps on a 4k screen maybe for a 3090, or perhaps making up for some sort of lack elsewhere in the case of the ferrari owner.
 
  • Like
Reactions: Leeea

coercitiv

Diamond Member
Jan 24, 2014
6,150
11,670
136
Which is also called Tuning... trying to see what the temps are and seeing if it will throttle, but posters here are going after me for saying you shouldn't have to tune especially on a 2400 dollar card..
Sometimes patience is such a virtue: here's a reddit thread full of links describing what users called "the red light of death", which was apparently a hardware problem with the first revision of the FTW3 cards.


Now, EVGA did say publicly that the issue is NOT related to heat or temperature, and that it was extremely unlikely to experience. Since then EVGA introduced a new revision that supposedly fixed the issue, but obviously many users still have the initial batch running in their computers.

Going back to cards failing in New World, which are not all EVGA anyway, it is still very possible that the issue may not be related with the one above, or that we may have multiple points of weakness exposed in various systems. We can't even rule out software, as the last time 3080 cards crashed in high numbers, many pointed out the capacitor configuration on the PCB... yet Nvidia fixed power consumption in the driver.

TL;DR Have some patience, don't administer verdicts based solely on hunch and previous experience. We don't even know if there's a real issue yet, or just a few weak cards (relative to the entire user base).
 
  • Like
Reactions: Leeea

coercitiv

Diamond Member
Jan 24, 2014
6,150
11,670
136
Igor's Lab has a strange theory about EVGA cards: that the fan control chips use in FTW3 gets damaged, even burns up. Problem is crowdsourced "evidence" points to more than EVGA cards, in fact JayzTwoCents argues this issue may affect some AMD cards as well.

Meanwhile Buildzoid offers an interesting opinion, speculating that the way cards are described to fail indicates the cards are pulling way too much current through some of their power stages:
  • many described their first incident is the card locks-up, and the system requires a power cycle to resume operation: this is consistent with the way OCP works on modern voltage controllers. (they lock the VRM and release only after a power cycle)
  • those with ultimately dead cards also described a loud pop noise associated with the event, which can be explained by one of the fuses blowing up when Over Current Protection failed to accomplish it's goal of shutting down the VRM in time
Buildzoid also speculates as to why the card's power limits fail to do their job properly: this could happen if the AIB did not set proper (as in low enough) limits for OCP in the voltage controller and instead relied more on the card's power limit to avoid damaging the VRM. With a specific load such as the New World menu, it may be possible that part of the GPU is stressed more than the rest, leading to some of the VRM stages getting overwhelmed while overall power limit is far from being reached.

TL;DR It's possible this may be related to improper use of OCP on voltage controllers in various cards. Obvioulsy nothing is certain until pictures of failed PCBs become available for inspection.
 

Karnak

Senior member
Jan 5, 2017
399
767
136
Igor's Lab has a strange theory about EVGA cards: that the fan control chips use in FTW3 gets damaged, even burns up. Problem is crowdsourced "evidence" points to more than EVGA cards, in fact JayzTwoCents argues this issue may affect some AMD cards as well.
ofc Jayz doesnt want that to be an EVGA only thing. They are like one of his absolute top premium sponsors.

But he needs to differentiate between a black screen/gpu just shutting down (which can happen at like >1000fps, I know that myself) or a "physical damage" to the card which would be that fan control chip on the FTW3 if that's true. And that's a pretty big difference because then the latter would be literally a design flaw.
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
Looks like Amazon's game does have some kind of negative impact on the cards:
https://www.tomshardware.com/news/new-world-rtx-3090-death-patch-fix
As it turns out, the issue stemmed from the game's menu system not having a frame rate limiter, and allowing the GPUs to run at thousands upon thousands of frames per second. It's likely that the game was somehow "over-utilizing" the graphics card in a way that overloaded the VRM and power delivery system.
 
  • Like
Reactions: Leeea and Mopetar

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
Looks like Amazon's game does have some kind of negative impact on the cards:
https://www.tomshardware.com/news/new-world-rtx-3090-death-patch-fix

Yes, the uncapped menu system was a 'suspect' from the beginning (still unconfirmed). It still shouldn't cause a card to die, except of course there was an existing flaw or dysfunction in the card that this situation just exposed leading to the card's failure. It could also be that running the cards through furmark or other heavy load situations would not expose the flaw and that the extremely high fps caused high frequency transients which is the situation of failure for the power delivery, thus prior efforts to test stability or failure would have missed the issue.
 
  • Like
Reactions: Leeea

thilanliyan

Lifer
Jun 21, 2005
11,848
2,051
126
Your missing my statement.
I agree a user should not have to MOD anything to make it work out of the box, hence why the vendor will replace it.
But a user should realize the limits of the card and attempt to safe guard it within acceptable measures.
Playing a game while at stock should be within the limits of the card. If it can't do that then the card should not have been released.

I agree people can tune for temps (I usually do with w/c and extra heatsinks on vram/VRMs) but I'm willing to bet the majority of people do not tune their cards to get better temps. And they have a right to expect the card to work out of the box as-is. Yes they can RMA it, and hopefully there is some fix for the new card.
 
  • Like
Reactions: Leeea

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
If it's within warranty, then yes.
Interesting take.

I don't agree with it, of course.

If a person could do anything they wanted to hardware and still get it warranted, things like heavily overclocking CPUs with dry ice and plotting chia on SSDs would also be covered. You'd hate to own the company that had to cover these kind of losses (kind of like the old "lifetime" electronic rent-a-center abuse Costco dealt with before changing it to 90 days)
 
  • Like
Reactions: Leeea