News Amazon: Don't Blame New World for GPU Deaths, Blame Card Makers

[hotaru251]

It is the hardware designers job to produce a design that is not damaged by software running in user space.

the game
ignores
gpu power settigns.


how do hardare makers do what you say when the game devs say <Mod Edit> you we aint listening?



eli5 version: imagine you say u can have this gun but only hunt animals and give it to a serial killer. You made it for animals yet the user ignores ur rules.

 

[TerryLaze]

the game
ignores
gpu power settigns.

You can't ignore something if it's a physical limit, that's what this discussion is about.
Any hardware should have enough fail safes to not burn out, it should reduce power draw/clocks/whatever if it becomes too dangerous.

I mean even intel CPUs that go well above twice the rated TDP, by ignoring power limits and using some software, there is still zero of them that burned out because no matter what you do in software there is a physical limit that you can not break.

 

[btmedic04]

So much ignorance in this thread, oh my goodness...

This is a hardware/driver problem with nVidia and their AIBs. Period.

New World can be coded like crap, much like any school/uni project can. It should not be able to override or bypass HARDWARE protections. How is that hard to understand?

In Jay's video, he only found a bug with Afterburner and EVGA cards. He should've been using Precision instead. And no one here has any idea how the game itself was coded. And it shouldn't matter. Power Virus or not, it must not be able to kill a GPU so easily.

Regards.

If it were strictly a hardware/driver problem on behalf of nvidia or their AIBs, then why did games like cyberpunk 2077, watchdogs legion, or CoD black ops cold war not expose these issues? we all know cp2077 wasnt finished on launch, CoD is CoD and is always poorly written and watchdogs legion is just difficult for hardware

 

[hotaru.hino]

If it were strictly a hardware/driver problem on behalf of nvidia or their AIBs, then why did games like cyberpunk 2077, watchdogs legion, or CoD black ops cold war not expose these issues? we all know cp2077 wasnt finished on launch, CoD is CoD and is always poorly written and watchdogs legion is just difficult for hardware

Conversely if it is an issue with software, how come New World isn't frying less powerful cards? And no, the answer isn't "because they're less powerful, duh!", less powerful cards still have a similar power density as higher end ones.

(although as much as I've cared to follow this issue, I've only seen NVIDIA's 80-90 level being fried by this game)

 

[JamesJones44]

If the software is inadvertently exposing a hardware fault, then it's a problem of the hardware. If the software is deliberately exposing a hardware fault, then it's a problem of the hardware and the software.



The hardware running above given limits is literally what boosting does. The difference is that GPUs are designed to boost in short spurts, not be running in boost mode all the time.



Why should we believe that they didn't "hack the hardware drivers" when literally no other game on the market is behaving this way? Amazon is effectively banking its entire future as a game studio on this one title, so it's in their best interest to deny any and all guilt in PR nightmares of this magnitude.

Furthermore, there's a difference between aggressive optimization and telling the card to literally overclock itself to death.

There are a crap ton of assumptions in here.

Software should never be able to break hardware, ever, it's just that simple. It's bad hardware design to say, well let's let any random vendor out there override what our power and thermal limits are, because, what's the worst that could happen? Oh, crap, someone designed some malware or a virus targeting our GPUs and then... poof millions of our GPUs dead overnight and we have a warranty fiscal on our hands. No hardware maker in the right mind is going to leave their hardware venerable to software like that intentionally.

I understand people hate Amazon, but this is clearly a poor hardware design/manufacturing issue.

 

[hotaru.hino]

Also here's something to consider, Furmark was (perhaps to a minor degree still is today) infamous for blowing up cards. Or rather, blowing up the VRM due to stressing them out. AMD did something to combat this: provide better overcurrent protection on their video cards. NVIDIA added overpower protection instead. Let's also not forget there was a case of GeForce 10 cards exploding because of improper VRM cooling.

Though in the case of the former case, i.e., using Furmark, you can easily shift the blame to the user for running a "power virus." However, it's still on the onus of hardware manufacturers that if a user can run such software, then they need to protect their hardware from such.

If you want an analogy, imagine if a car manufacturer does not put a rev-limiter to prevent the user from exceeding the engine's redline. They're just hoping the user isn't "dumb enough" to exceed it.

 

[bigdragon]

If it were strictly a hardware/driver problem on behalf of nvidia or their AIBs, then why did games like cyberpunk 2077, watchdogs legion, or CoD black ops cold war not expose these issues? we all know cp2077 wasnt finished on launch, CoD is CoD and is always poorly written and watchdogs legion is just difficult for hardware

Amazon is new to the games industry and has brought a new engine with them. They're doing things that Activision, EA, Ubisoft, Epic, CDPR, and other companies likely aren't doing. Amazon's inexperience and lack of strong relationships with GPU designers opens the door to some sort of unexpected software behavior that is exposing a latent design flaw in Nvidia's high-end products.

I think the fault is on the hardware side here. Inadequately spec'ed components and a failure to enforce hardware limits both appear to contribute. However, Amazon has to live in an ecosystem their game can damage. Nvidia and partners need to fix a flaw on the hardware side, and Amazon needs to take action on the software side to avoid harming their players.

 

[husker]

I'm not an engineer or anything close to it; just a guy at home eating Doritos surfin' the 'net. But if you want to pay me 6 figures to work on your next video card, I'll tell you to put something like a $5 bit of simple surge suppressor circuitry in your $2000 card. Okay, that one was no charge. Now, lets talk health care and bennies.

 

[-Fran-]

eli5 version: imagine you say u can have this gun but only hunt animals and give it to a serial killer. You made it for animals yet the user ignores ur rules.

Well, no. In your example the equivalency is the gun is exploding in your hand instead of actually firing and your defense is "YOU PUSHED THE TRIGGER WRONG!".

Does that ring a bell with a certain phone a good years back? It's quite a famous quote as well, haha.

Regards.

 

[-Fran-]

If it were strictly a hardware/driver problem on behalf of nvidia or their AIBs, then why did games like cyberpunk 2077, watchdogs legion, or CoD black ops cold war not expose these issues? we all know cp2077 wasnt finished on launch, CoD is CoD and is always poorly written and watchdogs legion is just difficult for hardware

As others have already, correctly I should say, pointed out the issue in your argument, I'll just say you need to really understand how hardware and software integration works at different levels in order to truly understand why this is an issue with nVidia and AIB's design/manufacturing (whatever is it).

Also, to everyone else, please do remember the RTX3K series launched with a VERY STUPID power issue that nVidia had to solve via drivers, so this IS NOT A NEW THING. I wouldn't be surprised cards died then, but it was kept low profile.

Ampere is cursed, lol.

Regards.

 

[hotaru.hino]

Also the news wires keep telling me that only NVIDIA cards are affected. Assuming this is true, if this game really was a GPU killer, why aren't similarly high performance AMD cards affected?

Again, Furmark was notorious for killing cards from both companies.

 

[-Fran-]

Very timely and this guy is one of the few on the internet you can 100% trust in what he says in regards to hardware:

View: https://www.youtube.com/watch?v=w-zSgiMEzkU


Regards.*

 

[Deleted member 1353997]

Also, to everyone else, please do remember the RTX3K series launched with a VERY STUPID power issue that nVidia had to solve via drivers, so this IS NOT A NEW THING. I wouldn't be surprised cards died then, but it was kept low profile.

I can confirm this.

As soon as I had upgraded my GPU to the RTX 3080, Windows 10 had been crashing randomly. It never crashed while playing games, but simply watching Netflix or YouTube, or opening a new tab on my browser could cause my screen to black out, which was occasionally followed by the PC rebooting. All crash dumps pointed towards at Nvidia, and a couple of driver updates later, my PC has been crash-free for the past half year.

If a mere video (or browser) could crash my GPU, imagine what a game can do...
Rather than blame Amazon or New World, we should count ourselves lucky that Cyberpunk and other games didn't accidentally fry our PC.

 

[Ryrynz]

Strong disagree. Even if New World is revealing manufacturing defects in certain cards, the fact remains that these issues are largely exclusive to New World - no other game, not even demanding ones, cause the GPU to run so hard. While the game is running, you can watch your GPU's power usage regularly spike to 15-20% above its set limits, and that kind of cycle for an extended period of time will wear down the hardware (just like how turning a light bulb on and off will cause it to wear out faster than just leaving it on). Even cards that don't get bricked outright will still see their lifespans reduced as a result of this aggressive repeated power spiking.

Anecdotally, the only time I've ever heard my computer's fans ramp up so hard for so long as when playing New World is when I'm running extended benchmarks, and games should not be running like benchmarks. I've since enabled the 60 FPS cap, and now my computer runs much more in line with its usual behavior when playing other games. (I've got a Ryzen 9 5950X and an ASUS RTX 2080 Ti.)

By this reasoning every benchmark that doesn't kill your card is simply a bad benchmark because it's clearly not running your card as fast as it could.. 😂 If your card isn't running at the brink of death that's extra performance you're missing out on.. Sue everyone involved, you're not getting what you paid for 😂

 

[jepeman]

Wonder if there is also something similar to the recent news of AMD motherboards misreporting power telemetry so they would boost the CPU more.
GPU manufacturers having cards/BIOS misreporting power telemetry so the cards run at higher speeds, combined with poor design/parts causing the issue.

 

[btmedic04]

Amazon is new to the games industry and has brought a new engine with them. They're doing things that Activision, EA, Ubisoft, Epic, CDPR, and other companies likely aren't doing. Amazon's inexperience and lack of strong relationships with GPU designers opens the door to some sort of unexpected software behavior that is exposing a latent design flaw in Nvidia's high-end products.

I think the fault is on the hardware side here. Inadequately spec'ed components and a failure to enforce hardware limits both appear to contribute. However, Amazon has to live in an ecosystem their game can damage. Nvidia and partners need to fix a flaw on the hardware side, and Amazon needs to take action on the software side to avoid harming their players.

As others have already, correctly I should say, pointed out the issue in your argument, I'll just say you need to really understand how hardware and software integration works at different levels in order to truly understand why this is an issue with nVidia and AIB's design/manufacturing (whatever is it).

Also, to everyone else, please do remember the RTX3K series launched with a VERY STUPID power issue that nVidia had to solve via drivers, so this IS NOT A NEW THING. I wouldn't be surprised cards died then, but it was kept low profile.

Ampere is cursed, lol.

Regards.

nowhere do i say that it is strictly a software issue. it is both a hardware and software issue and instead of pointing the finger solely at AIBs, amazon needs to own up to its role as the software developer and start working with AIB's to correct the issue. that is where my issue is with this whole situation.

 

[TerryLaze]

nowhere do i say that it is strictly a software issue. it is both a hardware and software issue and instead of pointing the finger solely at AIBs, amazon needs to own up to its role as the software developer and start working with AIB's to correct the issue. that is where my issue is with this whole situation.

Yeah, not really. For amazon it is a problem but only for publicity.
If this is an nvidia issue then nvidia had to release guidelines for what you should not do in software, if there is such a guideline and nobody found/mentioned it yet then that's a much more serious problem.
If this is a problem of manufacturers then it is 100% a hardware problem where neither nvidia or amazon are to blame, although they are the ones that get most of the bad press.

 

[hotaru251]

Any hardware should have enough fail safes to not burn out, it should reduce power draw/clocks/whatever if it becomes too dangerous.

they do. called they listen to the rules of their control program.

look at jayz video to ez example.
his gpu was NOT able to go over 102% power even with a hacked bios...it refused.

yet New World had it over 111%.

what part of the game ignores the gpu's limits don't you understand?

issue is with it taking much more power than ti can handle.

your analogy on cpu is apples to oranges comparison. similar but not same.
a cpu is MUCH better equipped to deal with sudden surges and peaks.

you cant "limit" power in sense you want to as its out of spec and no "failsafe" can stop sudden power spikes. (same way if u have a storm surge your powerlines and you arent usign a proper surge protector.)

 

[Mattzun]

I agree with the posters who said this isn't an Amazon problem

I'm old enough to remember when games did directly address hardware and could actually break hardware with bad coding.

These days, the problems is either Microsoft (something odd in Direct X causing failure - unlikely), a bad driver (could be Nvidia or an overly aggressive OEM optimization), bad firmware code or an actual hardware problem (i.e. bad soldering)

Unless Amazon's code is overriding power limits (no one has claimed this), there is no way that what was shown in Jayz video should occur.
Given that one of manufacturer 3090 he tested was MUCH closer to the defined power limit, it seems like a problem with certain OEM's breaking the power limits (either bad coding in driver or an intentional performance optimization)

Breaking the power limits could also explain the previous issues with power supplies failing when the load was theoretically within limits (i.e Seasonic issues)

Even the Furmark issues in the past were actually driver or firmware issues.
The manufacturers were just really bad at designing stress tests for their cards and finding issues that were possible through the API but unlikely in most games.

Furmark and New World exposed problems that don't normally occur in the warranty period.
They normally break cards sometime AFTER the warranty expires (the bean counters love this)

 

[TerryLaze]

look at jayz video to ez example.
his gpu was NOT able to go over 102% power even with a hacked bios...it refused.

yet New World had it over 111%.

what part of the game ignores the gpu's limits don't you understand?

issue is with it taking much more power than ti can handle.

So software can do something that even the bios doesn't allow you to do and you think only the software is to blame?
Do think the same about spectre and meltdown?
If hardware allows software to do something that the company doesn't want the hardware to do then this is a issue with the hardware.
Just imagine if hackers figure out what part of the code is responsible for burning GPUs, it would be real bad.

 

[SyCoREAPER]

Let's just all agree on this:

New World is coded like $hit by a multi-billion dollar company.
3090's getting fried is faulty hardware (and probably bad drivers which should detect overcurrent and shut the card down).

The fact that these two have any relation is proven.

Beyond that everything else is speculation.

If you have a 30xx and kept playing/pushing your luck after all this news, YOU are to blame.

Until we have a technical analysis on what exactly causes this there is no point arguing about it. Jayztwocents was a good start at finding some underlying issues. There is some undetermined link between the game and the hardware fault of the game.

 

[watzupken]

I do believe that there are flaws with the hardware design that contributed to the issue. And since it seems to affect primarily high end Nvidia cards, it is difficult for Nvidia and board partners to shrug the issue off. In addition, the EVGA card as shown by JayZ behaves even more erratic, which is likely by bad design/ component choice/ quality of build.

I don't believe New World as a software can get away with the issue as well since cards I predominately failing when running this particular title. There are very taxing game titles out there, but none to the point of killing the GPU whether it is a halo product or an entry level card.

The game definitely exposed some the weaknesses with the GPU design by being excessively taxing on the GPU for whatever reasons. Something that most games before don't have, even in a taxing title like CyberPunk that is really hard on the GPU.

 

[Colif]

Game shouldn't be able to run in 4k in windows with DSR off. Yes, the power figures get all the attention but game shouldn't let you run 4k on a monitor that doesn't support it with the features that are designed to let you run , turned off.

The 4k weirdness is tied to the power usage spikes from what I saw in that video of Jay. I don't know what it is asking GPU to do in that situation. Probably running as if monitor is 4k.

 

[TerryLaze]

Game shouldn't be able to run in 4k in windows with DSR off. Yes, the power figures get all the attention but game shouldn't let you run 4k on a monitor that doesn't support it with the features that are designed to let you run , turned off.

The 4k weirdness is tied to the power usage spikes from what I saw in that video of Jay. I don't know what it is asking GPU to do in that situation. Probably running as if monitor is 4k.

There are tons of games that allow you to render the game at a different resolution than it is displayed in, battlefield does it for years. (it's called resolution scale mode)

Higher resolutions will use more of the GPU render units if they aren't all used at lower resolutions already, so it will make a card use more power potentially but a card should still keep its power limits no matter what the resolution or how much of the card is being used.

 

[jasonkaler]

Strong disagree... While the game is running, you can watch your GPU's power usage regularly spike to 15-20% above its set limits,

It's not the software's fault if the hardware is going beyond its limits. If Hardware doesn't stop at its limits, it's a hardware fault.