Amazon's New World is bricking RTX 3090 graphics cards

[qubit]

I've never liked it when one of my cards starts screaming when it does a really high framerate, like it's freewheeling. Always thought that meant it was overstressed and this story proves it to be true.

 

[BSim500]

You can cap the max FPS using nVidia's control panel in 3D settings. Why not using this ? Especially when you know that a lot of games don't have a FPS cap feature and you can hear your card working for nothing in the menus of many games.

Whilst that's a good idea (including for the many other unmentioned games that have uncapped fps in the menu that resulted in thousands of fps, Example 1, Example 2, Example 3, etc), it's still more a workaround hack than a solution as the underlying EVGA specific hardware issue is still there and could rear its ugly head on yet another unforeseen power delivery issue in future. If I owned a £1,500 EVGA card that was affected by this I'd definitely want it completely fixed & replaced, not just wallpapered over, as no one would accept even a £59 Pentium CPU physically exploding on being maxed out by a simple video game due to a power regulation / delivery failure in its FIVR not being an obvious hardware fault, and this is the GPU equivalent of that x20 more expensive. (Not to mention that if that's all it took to kill an expensive GPU then expect "Power Virus" malware to be written to do that deliberately including possibly disabling any VSync / frame-limiter first in GPU driver / Rivatuner first).

 

[R-T-B]

Uncapped frame-rates in menus are undesirable but it's 100% a hardware problem if a GPU dies from it (let alone only EVGA GPU's). And this isn't EVGA's first time in that area...

It's not exclusively EVGA GPUs, but you do have a strong point all the same.

I think it's an issue with any vendor that cheaped our on VRAM protections on the GDDR6X of the 3090/3080/3070 ti

 

[The red spirit]

Didn't know RTX 3090s were so unreliable. It seems that RTX 2080 Ti was unreliable too. I wonder what is going on at nVidia that stupid QA problems like that aren't caught before releasing cards. And come to think about it, whole RX 5700/5600 range was ridden with black screen bugs, which could be fixed with more voltage, indicating poor engineering, not bad software. There haven't been so many problems in past so often. Feels like another capacitor plague, but this time it's engineering plague.

 

[TheoneandonlyMrK]

One shouldn't have to.

Indeed ,not after two to three grand.

 

[CrAsHnBuRnXp]

 

[DeathtoGnomes]

Software shouldn't be able to kill a card

yea? Furmark comes to mind.

 

[GerKNG]

One shouldn't have to.

i agree.
but if i am in this situation i'd run the fans at a much higher RPM and RMA the Card for being a POS.

 

[The red spirit]

yea? Furmark comes to mind.

Furmark doesn't kill properly made cards.

 

[eidairaman1]

Good question. I'd say both. Software shouldn't be able to kill a card and also Amazon for putting a card under such undue stress.

Furmark

 

[Shrek]

Furmark doesn't kill properly made cards.

I agree, a card should be able to handle anything except perhaps a malicious attack, and even then should protect itself.

 

[mrthanhnguyen]

If your 3090 does not get killed in Quake 2 RTX or Timespy GT2 or Metro Exodus enhanced edition, a chance of bricking your 3090 is slim.

 

[The red spirit]

I agree, a card should be able to handle anything except perhaps a malicious attack, and even then should protect itself.

I haven't heard of malicious attacks on graphics cards (Furmark isn't one), but any type of computer hardware shouldn't catch on fire or die under any load. I personally never had that poorly engineered card (except for Powercolor, who have no ears and no common sense). It would be nice if TPU started some poorly engineered card list, so that some people won't buy those cards used. So far RTX 2080 Ti, RTX 3090, RX 5700/5600 series should be on that list.

 

[mrthanhnguyen]

I haven't heard of malicious attacks on graphics cards (Furmark isn't one), but any type of computer hardware shouldn't catch on fire or die under any load. I personally never had that poorly engineered card (except for Powercolor, who have no ears and no common sense). It would be nice if TPU started some poorly engineered card list, so that some people won't buy those cards used. So far RTX 2080 Ti, RTX 3090, RX 5700/5600 series should be on that list.

So if I sell you my 3 x 3090 for 1 cents, u wont buy them too?

 

[Mussels]

I've never liked it when one of my cards starts screaming when it does a really high framerate, like it's freewheeling. Always thought that meant it was overstressed and this story proves it to be true.

good old coilr whine

and yeah, its an audible sign of stress from whatever is causing it (coils, caps, whatever)


sounds like EVGA made a bad decision and its killing their cards

 

[looniam]

 

[venturi]

So, taking it all in (as an owner of 3x 3090 RTX FE cards) -- is the message beware some apps or games can toast your card at factory settings, OR is the message beware some video cards are manufactured so that there are no built in limitations in the hardware for self preservation of the device?

I would hope that, the hardware would have been assembled/designed etc to not allow an app/game to destroy it or at the very least built in protections would engage and then disengage after the 'threat' or cooldown took place - without permanent harm to the hardware.

at least, one would think. Otherwise the scenarios would happen with much greater frequency.
It would be interesting to see if with granular data, we could find out if the failures are occurring in unique situations (hardware mods/ overclocking/ etc)?

Curious to see what shakes out from this.

 

[Shrek]

I haven't heard of malicious attacks on graphics cards

Wasn't there one long, long ago that would keep changing the resolution of the monitor, or something like that?

 

[The red spirit]

So if I sell you my 3 x 3090 for 1 cents, u wont buy them too?

At 1 cent, I will buy them all. Even 2 of them are frying poo poo, I may luck out and get 1 working one, but unfortunately I'm not a fan of electronic ovens and I would need to upgrade my power supply. Oh and I can't modify vBIOS of them, which is shame, because I would love to set TDP to half of what they are currently. But there are two questions. Have you tried to play that Amazon game? Will you sell them for 1 cent and find some sexy RX 580 for 400 Euros as a proper replacement?

 

[Mussels]

Wasn't there one long, long ago that would keep changing the resolution of the monitor, or something like that?

thats an attack on windows, not the video card

 

[wolf]

Oh EVGA, quite an expensive corner to cut.

 

[qubit]

sounds like EVGA made a bad decision and its killing their cards

At least they're honouring the warranty when it dies like this. I wonder if the cheaper brands would try to duck out of it.

 

[mrthanhnguyen]

At 1 cent, I will buy them all. Even 2 of them are frying poo poo, I may luck out and get 1 working one, but unfortunately I'm not a fan of electronic ovens and I would need to upgrade my power supply. Oh and I can't modify vBIOS of them, which is shame, because I would love to set TDP to half of what they are currently. But there are two questions. Have you tried to play that Amazon game? Will you sell them for 1 cent and find some sexy RX 580 for 400 Euros as a proper replacement?

Nah I dont play games. Im mining with them currently and benchmark occasionally.

 

[Count Shagula]

Im running my water cooled EVGA 3080 on a 360hz monitor atm +250 on the core and +1200 on mem with vsync off always on a 450w bios with all the sliders maxed. Its at 100% load all the time when gaming and bounces between 2150-2260ish core speed. I don't see how 100% load could ever be a problem unless its a firmware issue or something whack, if they get too hot they're meant to downclock

 

[Keullo-e]

I don't know how to link from twitter as I don't use it, but JayzTwoCents says there that some 6800/6900 Radeons have also died on that.