Amazon’s New World game is bricking GeForce RTX 3090 graphics cards

sharknice

2[H]4U
Joined
Nov 12, 2012
Messages
2,632
A more apt comparison would be the track installing software and your perfectly fine car's gas pedal would be stuck to the floor the entire time, and your engine redlines and blows. Comparing a 3090 to bringing a pos car to a racetrack is pretty poor anyway.

Yeah no it wouldn't. Imagine going to a track and not wanting to go fast. That's an analogy for how you think GPUs should work.
 

arnemetis

2[H]4U
Joined
Aug 2, 2004
Messages
3,545
Yeah no it wouldn't. Imagine going to a track and not wanting to go fast. That's an analogy for how you think GPUs should work.
Quite the difference between wanting to go fast and pushing your car to 100% redline the entire time. It appears you have no idea how cars, or gpus work.

Back on topic, this is only coming up now because Amazon's software was coded poorly to not implement frame caps at a menu. Something no other game has caused. Their fix of frame limiting clearly identifies the issue here.
 
Joined
Dec 7, 2010
Messages
985
Quite the difference between wanting to go fast and pushing your car to 100% redline the entire time. It appears you have no idea how cars, or gpus work.

Back on topic, this is only coming up now because Amazon's software was coded poorly to not implement frame caps at a menu. Something no other game has caused. Their fix of frame limiting clearly identifies the issue here.
Bullshit. Unless you modified the GPU BIOS or power management control firmware the GPU should never kill itself period. This is shitty hardware/firmware pure and simple.

Similar to using some of those larger SIMD/AVX instructions on intel CPUs - any software can try to use them all concurrently but to prevent dangerous current spikes the CPU throttles itself.

Nvidia/EVGA (and possibly AMD) didn’t properly stress test their own silicon. That’s the only news here. Thank you Amazon for finding a real design flaw/bug.

Blaming the game software shows that you in fact have no clue how GPUs work.
 

ElementDave

Limp Gawd
Joined
May 5, 2013
Messages
205
Higher end cards tend to be putting a lot more voltage into their chips and ram to run at higher frequencies.

It would be like taking 2 CPUs... clocking one 200mhz lower then its normal clock and then other 200mhz over. Which one is going to heat up the most ? If their is a voltage fault in the MOBO they are slotted into.... the one drawing more power is going to find the flaw first.

This seems to be something to do with voltage regulation or some such thing that is causing some 3090s to fry parts.
That's why I hate factory-overclocked parts. If manufacturers want to differentiate their cards by "OC" potential, maybe introduce something similar to XMP profiles. That way end-users don't have to fiddle around with individual OC settings unless they want to, but they do have to "opt-in" to the overclock.
 

arnemetis

2[H]4U
Joined
Aug 2, 2004
Messages
3,545
Bullshit. Unless you modified the GPU BIOS or power management control firmware the GPU should never kill itself period. This is shitty hardware/firmware pure and simple.

Similar to using some of those larger SIMD/AVX instructions on intel CPUs - any software can try to use them all concurrently but to prevent dangerous current spikes the CPU throttles itself.

Nvidia/EVGA (and possibly AMD) didn’t properly stress test their own silicon. That’s the only news here. Thank you Amazon for finding a real design flaw/bug.

Blaming the game software shows that you in fact have no clue how GPUs work.
So are you dismissing the fact that this is the only game doing this? I don't agree there should be a lawsuit or anything, but shitty coding is a piece of the issue here, especially considering they could patch it right away and now there is no issue. If this game had never existed, this story likely wouldn't have either. Let's also not forget that the issue was brought up during Alpha, and was never addressed. I don't understand how you can absolve the software entirely.

Sure let's say it was a hardware flaw. It's unreasonable to expect every single possible scenario to be tested, this was an unheard of scenario. Now moving forward they have to test a card running to produce 70,000 fps because someone might forget to limit a simple menu. Now we have to come up with every other potential scenario, that's a lot of R&D. That's how we get $250,000 gpus. In any product you cannot plan for every single scenario.

So please, do you have a suggestion for a resolution? The partners should do multi year rounds of testing? Consumers are only able to purchase gpus every few years (existing availability issues notwithstanding) and in the meantime they are paying thousands of engineers to test every possible piece of code?

Edit: To be clear, if this does come out to really be full on design flaw that can be exposed any other way, EVGA should do a recall and swap out 100% of 3090s sold. It will suck for them, but if they screwed the pooch so to speak and made a seriously flawed design (especially if this deviated from nvidia's design suggestions) then they should eat it. Yes I realize that will possibly lead to evga declaring bankruptcy and being dissolved.
 
Last edited:

sharknice

2[H]4U
Joined
Nov 12, 2012
Messages
2,632
So are you dismissing the fact that this is the only game doing this? I don't agree there should be a lawsuit or anything, but shitty coding is a piece of the issue here, especially considering they could patch it right away and now there is no issue. If this game had never existed, this story likely wouldn't have either. Let's also not forget that the issue was brought up during Alpha, and was never addressed. I don't understand how you can absolve the software entirely.

Sure let's say it was a hardware flaw. It's unreasonable to expect every single possible scenario to be tested, this was an unheard of scenario. Now moving forward they have to test a card running to produce 70,000 fps because someone might forget to limit a simple menu. Now we have to come up with every other potential scenario, that's a lot of R&D. That's how we get $250,000 gpus. In any product you cannot plan for every single scenario.

So please, do you have a suggestion for a resolution? The partners should do multi year rounds of testing? Consumers are only able to purchase gpus every few years (existing availability issues notwithstanding) and in the meantime they are paying thousands of engineers to test every possible piece of code?

Edit: To be clear, if this does come out to really be full on design flaw that can be exposed any other way, EVGA should do a recall and swap out 100% of 3090s sold. It will suck for them, but if they screwed the pooch so to speak and made a seriously flawed design (especially if this deviated from nvidia's design suggestions) then they should eat it. Yes I realize that will possibly lead to evga declaring bankruptcy and being dissolved.

99.9% of cards handle this game perfectly fine just like they should. A few don't because they have a flaw. It's not the games fault.
 

SvenBent

2[H]4U
Joined
Sep 13, 2008
Messages
3,308
So are you dismissing the fact that this is the only game doing this? I

absolutely because that is not an intelligent way to logical evaluate anything at all

- hey I brought this 2 tonn load truck and it broke down only loaded with 1.5ton of stones

- yeah its not the truck faults it worked fine with everyone else that just used it for 50kkg....

nah right. a product is suppossed to be working for what it is sold for and it is sold to take 3d instruction and calculate them
and because it get utilize fully, which is a limit set by the manufactorer, and breaks down means something was wrong with it.

down let companies go away with shitty designs
 
Last edited:

Ebernanut

[H]ard|Gawd
Joined
Dec 15, 2010
Messages
1,544
The main blame and the blame for cards actually dying is on the hardware itself but the cause is something that has been a lesser issue in the past. Previously I've only been aware of it causing coil whine and overheating that led to cards throttling for a bit but I haven't noticed it in quite a while and my last card that I ran for eons had bad coil whine in games with that issue so it seems like most devs learned to avoid it.
 

arnemetis

2[H]4U
Joined
Aug 2, 2004
Messages
3,545
Alright so we're all confident it's a hardware design failure, even though no one has disassembled & diagnosed this. What's next? I'm keeping my 3090 far away from this game, but now what should I do?
 

TheSlySyl

[H]ard|Gawd
Joined
May 30, 2018
Messages
1,583
I don't think "waiting for either New World Developers or Nvidia" to make a patch for a game that's still in beta (alpha?) testing is that big of a demand.

Just... hold off playing it for a week or so? It's not even that great is my understanding.
 

Mchart

Supreme [H]ardness
Joined
Aug 7, 2004
Messages
4,678
Alright so we're all confident it's a hardware design failure, even though no one has disassembled & diagnosed this. What's next? I'm keeping my 3090 far away from this game, but now what should I do?
As long as you have a FPS cap set in NVCP it won't happen.

Personally, if I had a 3090, i'd let the damn thing blow if it were going to blow, because the AIB needs to fix their poorly tested hardware. I've got a 3080ti, and i'm not adjusting shit. If it blows it blows. It would be like buying a high-horsepower car like a Hellcat and never going WOT because you're afraid it's going to blow up. Rip that shit, and if it blows make FCA replace it under warranty.
 

Ebernanut

[H]ard|Gawd
Joined
Dec 15, 2010
Messages
1,544
Alright so we're all confident it's a hardware design failure, even though no one has disassembled & diagnosed this. What's next? I'm keeping my 3090 far away from this game, but now what should I do?
Early reports are suggesting it's a fan controller on the EVGA cards that's causing the issue but when I said hardware I was using the term loosely and including potential driver and firmware problems.
 

arnemetis

2[H]4U
Joined
Aug 2, 2004
Messages
3,545
As long as you have a FPS cap set in NVCP it won't happen.

Personally, if I had a 3090, i'd let the damn thing blow if it were going to blow, because the AIB needs to fix their poorly tested hardware. I've got a 3080ti, and i'm not adjusting shit. If it blows it blows. It would be like buying a high-horsepower car like a Hellcat and never going WOT because you're afraid it's going to blow up. Rip that shit, and if it blows make FCA replace it under warranty.
Heh, few reasons I won't be joining you in that endevour. First, it means you're without a card for an undetermined amount of time, which will be exasperated due to the shortage and most likely a lot of others having the same problem. Second, I highly doubt they will even fix the problem this generation, so you'll just keep blowing cards. Third, I have to imagine after enough times they will either find an excuse to deny your warranty or just keep taking so long that it becomes maddening for you. Forth is for me specifically, I water blocked my card and I REALLY do not want to disassemble my system unless I have to. It's kind of a bitch to change:
0704211248_HDR.jpg
 

Mchart

Supreme [H]ardness
Joined
Aug 7, 2004
Messages
4,678
Heh, few reasons I won't be joining you in that endevour. First, it means you're without a card for an undetermined amount of time, which will be exasperated due to the shortage and most likely a lot of others having the same problem. Second, I highly doubt they will even fix the problem this generation, so you'll just keep blowing cards. Third, I have to imagine after enough times they will either find an excuse to deny your warranty or just keep taking so long that it becomes maddening for you. Forth is for me specifically, I water blocked my card and I REALLY do not want to disassemble my system unless I have to. It's kind of a bitch to change:
View attachment 377583
Well, yes, in your case you are somewhat screwed because depending on the manufacturer, what you did invalidated the warranty anyways. This is part of the reason why I stopped doing that stuff, TBH. I buy the best that I can, and run it stock. If it shits the bed so be it, i'm not worried.

As for not fixing the problem, I don't agree there. This isn't the first time this has happened, and after enough money lost EVGA has always developed a fix. (Remember the ACX cooler problems & VRAM overheating?) However, they won't be bothered to do so if people aren't RMA'ing their broke ass cards.
 

notarat

2[H]4U
Joined
Mar 28, 2010
Messages
2,331
Yeah no it wouldn't. Imagine going to a track and not wanting to go fast. That's an analogy for how you think GPUs should work.

Why don't you imagine going to the track and revving your car to/past red line while waiting in the line to pull up to the starting line...
 

Mchart

Supreme [H]ardness
Joined
Aug 7, 2004
Messages
4,678
Why don't you imagine going to the track and revving your car to/past red line while waiting in the line to pull up to the starting line...
Auto manufacturers put specific terms in their warranty that negate the warranty if the damage is due to track usage, etc.

There is absolutely zero language like this in consumer GPU warranties. If I buy a normal retail 3090 and beat it to shit by mining 24/7 and it blows, any manufacturer is going to have to warranty it assuming I didn't mod the card. Further, a game isn't directly interfacing with the Nvidia driver API and causing the card the overclock itself, etc. The game isn't putting the foot on the pedal, so to speak. The game is like a person at a Mustang meet requesting the Mustang driver to go WOT in the parking lot. In this case, the idiot controlling the car is smashing into the crowd.
 
Last edited:

RazorWind

Supreme [H]ardness
Joined
Feb 11, 2001
Messages
4,107
This seems relevant.

I'd really like to get my hands one of these dead 3090s so I could do an autopsy (and possibly repair it?), but purely speculating, I think Buildzoid probably nailed it.

I've seen reports of a few cases where the person claimed that the whole system shut down, which implies a pretty catastrophic failure, but most of the cases sound like they're tripping the overcurrent protection on some part of the card. The ones where the card is totally dead afterward seem to be the ones where, as Buildzoid points out, the card has a fuse for each power rail. It wouldn't surprise me to find out that those fuses are a hair under spec'd, and can burn out due to a load that the card is otherwise capable of sustaining without damage.
 

undertaker2k8

[H]ard|Gawd
Joined
Jul 25, 2012
Messages
1,490
I wonder how many of the cards being bricked are running the 500w bioses? Why anyone would want to do that has always puzzled me, I'm happy with the 'default' 420W on the FTW3, I'd happier still if it could manage the same perf with 240W instead :)
 

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,810
https://forums.newworld.com/t/known-issue-nvidia-rtx-3090-series-100-gpu-usage/126068/28
It's interesting that most users seem angered at amazon. I think I would be more angry at Nvidia/AMD/EVGA/Gigabyte/ whoever that I don't even have enough control of my hardware to keep it from bricking from some arbitrary software load. Ofcourse poorly designed software is nothing to be happy about either, but my card BRICKING? I would place the blame of that on the hardware.

edit: on the other hand that is the newworld thread, so I guess it would have more complaints at Amazon. Maybe I should find the EVGA thread.
I like how we're given the narrative that this Amazon's games fault and not Nvidia's. Don't even question that it's possibly Nvidia's fault.
A more apt comparison would be the track installing software and your perfectly fine car's gas pedal would be stuck to the floor the entire time, and your engine redlines and blows. Comparing a 3090 to bringing a pos car to a racetrack is pretty poor anyway.
The redline in an engine exists for a reason. That's the maximum safe RPM the engine can run without blowing up. If your engine blew up at redline then something else has gone terribly wrong. Either the engine got too hot and that blew it up or there's oil starvation. Most modern engines are smart enough to put the engine in limp mode when an engine gets too hot. Most engines built in the 80's and 90's don't care about oil pressure or temperature and will happily blow up the engine while redlining it for that long. Also, this is a graphics card and not a car engine so a lot of the analogy breaks down because running your hardware at max should not blow it up.

Remember when Doom 3 was released and was causing perfectly happy graphic cards to start crashing and artifacting because Doom 3 put a very different load on graphic cards compared to any other game that came before it? This is kinda the same situation except dead graphic cards.

https://arstechnica.com/civis/viewtopic.php?f=6&t=493127
 
Last edited:

Krenum

Fully [H]
Joined
Apr 29, 2005
Messages
18,310
Sounds like poorly designed hardware / coding or a mixture of both.

This is what happens when Nvidia comes up with a radical new designed PCB & doesn't give AIB manufactures enough time to design and test.
 
Last edited:

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,810
Sounds like poorly designed hardware / coding or a mixture of both.

This is what happens when Nvidia comes up with a radical new designed PCB & doesn't give AIB manufactures enough time to design and test.
So far it seems it's mostly EVGA so that might indicate something's wrong with EVGA cards. After watching JayzTwoCents video I'm certain the issue is that the game New World is pushing graphic cards very differently. When majority of games just reuse graphic engines like Unreal or whatever they've been using for decades then graphic cards will be designed around them. The game engine used in New World is called Amazon Lumberyard and probably built from scratch. New engine, new game, new way of putting a different kind of load on the hardware.
 

Marees

Gawd
Joined
Sep 28, 2018
Messages
994
Alright so we're all confident it's a hardware design failure, even though no one has disassembled & diagnosed this. What's next? I'm keeping my 3090 far away from this game, but now what should I do?

Is no one blaming Microsoft for allowing this to happen ?
 

Hallyday

Limp Gawd
Joined
May 20, 2021
Messages
154
Is no one blaming Microsoft for allowing this to happen ?

Are you just throwing the blame on MS without any logic behind or do you have an actual reason to? It's not like you can't overload a GPU outside of Windows environment. This is complete rubbish.
 

Marees

Gawd
Joined
Sep 28, 2018
Messages
994
Are you just throwing the blame on MS without any logic behind or do you have an actual reason to? It's not like you can't overload a GPU outside of Windows environment. This is complete rubbish.

True. You can overload GPU outside of windows APIs too.

I thought Microsoft would have a vested interest as their APIs or the interface between game & hardware.

(Something like a regulator role in the race tracks to make dangerous bends safer)
 

Hallyday

Limp Gawd
Joined
May 20, 2021
Messages
154
I mean there were games such as Quake 3 Linux, which was written for Unix based OS. At least of that time, not sure how functional it is now. If this happened there, would you know where to go after you card dies? Actually, let's not go that far because again, this 5000 FPS thing is nothing new. I know I've played older games that either rendered the scene too fast or just didn't have an FPS limit. Painkiller on top off my head is one. I recall getting 4-5k FPS in that game, so it's not far fetched to say this scenario could have happened in a different title. For sober fact, Nvidia added frame limit so late. It happened in one of the drivers of the last couple years.
 

notarat

2[H]4U
Joined
Mar 28, 2010
Messages
2,331
At this point, with different cards from different mfgrs affected, I'm wondering if the cause can be traced to a particular component shared amongst the different cards which was of borderline quality (so it passed normal QC procedures) and then it fails because of the different type of loads the game puts onto the cards...
 

sharknice

2[H]4U
Joined
Nov 12, 2012
Messages
2,632
So far it seems it's mostly EVGA so that might indicate something's wrong with EVGA cards. After watching JayzTwoCents video I'm certain the issue is that the game New World is pushing graphic cards very differently. When majority of games just reuse graphic engines like Unreal or whatever they've been using for decades then graphic cards will be designed around them. The game engine used in New World is called Amazon Lumberyard and probably built from scratch. New engine, new game, new way of putting a different kind of load on the hardware.

Amazon bought the Crytek engine a few years ago and turned it into Lumberyard. But that's basically what's happening. It's stressing GPUs in ways they haven't been stressed before and exposing flaws in them. It happens with every big new game. It just typically causes crashes instead of a bricked card. But in both cases it is clearly the fault of the hardware.

People blame the games because they don't want there to be a problem with their hardware. It's a pain in the ass to deal with hardware problems, and much easier to just download a patch or dismiss the game as broken and play other games. Likely getting the occasional crash in other games and blaming that on software too, when in reality their hardware is just unstable.
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
5,391
https://www.thefpsreview.com/2021/0...-lovelace-gpus-in-performance-and-efficiency/

This same situation played out with the current gen cards, the 3090 was a last minute attempt to stay on top. Hopefully the lesson has been learned and the 4090 won't have the same flaw.
Or better yet Nvidia (and AMD) won't release a cherry picked top 1% of the silicon card that has to be run 10% over the power envelope it was designed for to win a benchmark marketing battle king of the hill bragory.

And or if they do hopefully people won't be stupid enough to buy in. If they release a 4090 and a 7900.... the best thing the gaming world could do would be to ignore them. (I know never happen but it would be for the best) These types of cards have always been prone to burning out and in some cases up. lol
 

njenabnit

Limp Gawd
Joined
Jul 24, 2007
Messages
132
Heh, few reasons I won't be joining you in that endevour. First, it means you're without a card for an undetermined amount of time, which will be exasperated due to the shortage and most likely a lot of others having the same problem. Second, I highly doubt they will even fix the problem this generation, so you'll just keep blowing cards. Third, I have to imagine after enough times they will either find an excuse to deny your warranty or just keep taking so long that it becomes maddening for you. Forth is for me specifically, I water blocked my card and I REALLY do not want to disassemble my system unless I have to. It's kind of a bitch to change:
View attachment 377583
thats a CLEAN loop my dude. which block did you end up with? I put the new bykski front and back block on mine and have been happy with it.
 

cjcox

[H]ard|Gawd
Joined
Jun 7, 2004
Messages
2,024
Customer, reading the fine print: LIN cooling required for high performance.
 

defaultluser

[H]F Junkie
Joined
Jan 14, 2006
Messages
14,324
You know we have too many Clickbait Headlines when you use the word "Bricked " to describe a software fault .

Bricked has always referred to unrecoverable Firmware corruption / bootloops; just because the machine BSODed so hard it had to be restarted doesn't mean it suddenly cant reboot.

If these cards can't handle temp spikes to the point that they break permanently , then that is a recall-level fault, but otherwise this is a Nothingburger "Card Fans are being just as whiny and loud as most of the pretentious owners of these pverpriced cards"
 
Last edited:

Falkentyne

[H]ard|Gawd
Joined
Jul 19, 2000
Messages
1,812
In Jay's second video, he said that the eVGA card was bypassing its TDP Limit set in MSI Afterburner by as much as 10% (e.g. 107% set was going up to 117%), then he switched to a MSI 3090 and set the slider to 102% and it stayed around 102-103%.
That really stood out.

Too bad I don't have this game....
 

arnemetis

2[H]4U
Joined
Aug 2, 2004
Messages
3,545
thats a CLEAN loop my dude. which block did you end up with? I put the new bykski front and back block on mine and have been happy with it.
Not to get too off topic, but its the ek block and regular ek backplate. The actively cooled backplate won't be available for months so I went with this. The bottom fans cool the backplate well enough, even during mining and +1300 on the memory vram temp stays at 96c. Gotta get around to making a build log.

I will wait this out, and hopefully never need an rma. I did get the extended warranty, but man what a pain in the ass it would be to swap this out.
 

Digital Viper-X-

[H]F Junkie
Joined
Dec 9, 2000
Messages
14,619
Not to get too off topic, but its the ek block and regular ek backplate. The actively cooled backplate won't be available for months so I went with this. The bottom fans cool the backplate well enough, even during mining and +1300 on the memory vram temp stays at 96c. Gotta get around to making a build log.

I will wait this out, and hopefully never need an rma. I did get the extended warranty, but man what a pain in the ass it would be to swap this out.
That's why I added this.. :) ram won't break 90 without gpu hitting 50+
16270579657328131615323840184402.jpg
 
Top