So Diablo 4 could be damaging Nvidia 3### catds

[TheoneandonlyMrK]

This game has almost melted my GPU twice!

I have a 3080 TI, with the latest drivers. Performance has been fine today, but it has not once, but twice randomly caused my computer to shut down in a panic while all the fans are spinning at max speed. This is not normal behaviour, the last time I heard about something like this was when...

us.forums.blizzard.com

Not alone either there's a readit thread and a few other sites with news pieces.

 

[Nuke Dukem]

3080 Ti owners advised to avoid Diablo 4 if their card has AOZ5312UQI power driver modules.

Just like with New World, where the 3090 FTW3's using this analog controller were dying, 3080Ti's using this same chip are experiencing black screen and then card failure when trying to run this game without a framecap. Owners can confirm the usage of these problematic chips via TPU high res...

www.techpowerup.com

 

[R-T-B]

It's just select cards with a certain cheap buck controller, seems like only very early 3090 ftw3s (ti not affected) from evga and some 3080s from gigabyte really.

 

[CrAsHnBuRnXp]

Makes me glad that I buy the higher quality Asus cards

 

[Jeelo]

i dont even have a pc lol

 

[TheDeeGee]

i dont even have a pc lol

I don't even have a Diablo.

 

[oxrufiioxo]

Wife played 10 ish hours with a 3080ti 0 issues. FTW3 model. 1440p/165hz

I really enjoyed the beta in general.

My buddy who also owns a 3080ti same model finished all the beta content and leveled 3 characters to 25 also without issues. On a 3440x1440 200hz monitor.

The fans on my 4090 didn't even come on the majority of time 4k120hz so it doesn't seem very demanding.

The one thing I did notice is that at least with the High resolution texture pack it was maxing out the Vram of whatever gpu was being used so 24GB with my 4090.

Makes me glad that I buy the higher quality Asus cards

Apparently some of the cards that died were asus cards.

 

[Sora]

lol, delete my thread which was the older of the two and actually had information on the problematic part and threads, but keep this one......

Diablo 4 isn't damaging anything, this thread title is disengenous, misleading and contributes to the continued ignorance of the tech and gaming community every time something like this happens.

Diablo 4 is provoking a flaw in cards with poor component choice/configuration to damage themselves.

 

[ThrashZone]

Hi,
Amazon export ?

Nobody ever said hell wasn't hot

 

[GerKNG]

Diablo doesn't damage anything.
dirt cheap trash PCBs with disabled OCP damages GPUs.

 

[oxrufiioxo]

Diablo doesn't damage anything.
dirt cheap trash PCBs with disabled OCP damages GPUs.

idk Asus Strix cards are typically well built and apparently at least one of the cards that died was a strix. Same with the TUF really which tend to have pretty decent PCB.
Also Nvidia has to approve whatever design AIB are using and they are a lot more strict than AMD from everything I've read/heard.

I do agree this is a card issue and not a game issue though and given how bad transients are on 30 series gpus even compared to 40 series I am not surprised.... Glad my 30 series card is from EVGA honestly their support is pretty fantastic.

That being said it does seem to primarily be Gigabyte Eagle/Vision cards to the point their RMA site crashed hopefully gigabyte gets out ahead of this one.

 

[TheoneandonlyMrK]

lol, delete my thread which was the older of the two and actually had information on the problematic part and threads, but keep this one......

Diablo 4 isn't damaging anything, this thread title is disengenous, misleading and contributes to the continued ignorance of the tech and gaming community every time something like this happens.

Diablo 4 is provoking a flaw in cards with poor component choice/configuration to damage themselves.

I get your point.

But the title despite having a typo says could be, not IS.

And as for your perspective of diablo 4 not damaging them, it wasn't any other games or load.

So anyone worried should avoid the example noted.

Of course ALL GPU should not die when a software is run on them, and no, , it wouldn't be the software's fault but which software is causing it is mentioned to inform others not allocate blame.


Which my OP avoids, no blame or finger pointed, ,, too early anyway.

Yours though in most ways your thread may have been better had it gone the other way I wouldn't be in your thread pissing on your points, I'm no mod I didn't do it.

 

[Sora]

Which my OP avoids, no blame or finger pointed, ,, too early anyway.

It isn't.

 

[oxrufiioxo]

Of course ALL GPU should not die when a software is run on them, and no, , it wouldn't be the software's fault but which software is causing it is mentioned to inform others not allocate blame.

I remember The Gears 5 beta causing my pc to shut down I had a 9900K/2080ti strix at the time psu was a prime 850w tx unit. No game before or after that beta not even the retail game caused my system to do that. Sometimes software is just weird but it was obviously doing something my system didn't like. That was over 3 years ago and I just finally retired my 2080ti never had any other issues with it.


My 3080ti didn't like very many 750w/850w psu I tried 5 different ones with it it shut them all down I don't really count that as an issue other than I had to swap in a 1000w psu. It's currently running in a system just fine with an 850w prime TX unit that is relatively new though.

 

[Sora]

I remember The Gears 5 beta causing my pc to shut down I had a 9900K/2080ti strix at the time psu was a prime 850w tx unit. No game before or after that beta not even the retail game caused my system to do that. Sometimes software is just weird but it was obviously doing something my system didn't like. That was over 3 years ago and I just finally retired my 2080ti never had any other issues with it.

Games cannot shut your pc down.
Period.

 

[oxrufiioxo]

Games cannot shut your pc down.
Period.

Not saying it was the game. I'm saying the beta was doing something my system that lasted over 3 years without issues running way more demanding games didn't like.

I even said in a previous comment this is a hardware issue not a software issues

stop being butthurt over your thread getting deleted life goes on.

 

[TheoneandonlyMrK]

It isn't.

What isn't wtaf.

And you will discover software absolutely can shut down a pc , are you new to this?!.

 

[oxrufiioxo]

What isn't wtaf.

And you will discover software absolutely can shut down a pc , are you new to this?!.

I still remember all the drama about Furmark over a decade ago and while I am not gonna sit here and say it killed gpu's that wouldn't have died otherwise the fact that both Nvidia/AMD put protections at a driver level to heavily throttle gpu's when running similar programs makes me think it was a problem.

 

[Sora]

And you will discover software absolutely can shut down a pc

Yes, it calls the shutdown API present on the machine, which is not what the poster there was talking about

Games/Benchmarks cannot cause an unsafe shutdown without thermal, electrical fault or overload presence.

are you new to this?!.

i certainly question anyone not new to this with such misguided faulty beliefs.

I still remember all the drama about Furmark over a decade ago

The GTX 400 and 500 series had a gimped power delivery which could not handle what nvidia coined "Power Virus" loads, the node was also at the time sensitive to electrolysis, the silicon would rapidly degrade the oxide barrier when all parts of the core were active at high temps.

In the case of Furmark, it is a shader intensive application which triggered the former scenario, the power delivery could not safely handle 100% shader loads and there was a missing protection mechanism that would have put the GPU into protect mode (Black screen, Fans maxed) which was added with the 600 series and remains since. Nvidia opted to mitigating the issue with Fermi by throttling the gpu based on executable matching.

The issue with Fermi and Furmark, the system would not have shutdown under those high loads, it would have run itself till the overtaxed components hit critical temperatures and melted, shorting the substrate and likely killing the gpu.

Kepler (and later Pascal) revised the VRM and components used so the issue that would cause failure on Fermi parts doesn't occur, you can run furmark, or kombustor on these cards without much concern.

Power Virus and severe variance should be required testing on any of these 350w+ cards.

 

[Mussels]

This would be the exact same GPU's that were dying from other games in the past, the ones that cheaped out on power controllers and had surge issues - not every user of those cards played the games that triggered the issues back then.

Games/Benchmarks cannot cause an unsafe shutdown without thermal, electrical fault or overload presence.

This is correct. Other than the OS passing a legitimate shutdown request, you're talking malware or faulty hardware as the other options.

This is simply software exposing hardware faults, just like furmark did.

 

[KrazyT]

@Sora :
Reading your post I wonder why GC manufacturer did not implement a tiny speaker (like in the old days case) who will beep when things goes wrong.
At least you know when the card is overheating / malfunctionning

 

[ratirt]

@Sora :
Reading your post I wonder why GC manufacturer did not implement a tiny speaker (like in the old days case) who will beep when things goes wrong.
At least you know when the card is overheating / malfunctionning

They do not have those speakers since the cards are protected against overheating and damage. They will start throttling and crank the frequency and voltage down to mitigate the overheating problem and damaging the hardware. Why it does not work as it should with Diablo 4 with some cards is at least worrying. Especially, if you pay hard buck for it and RMA may not be applicable.

 

[TheoneandonlyMrK]

Games cannot shut your pc down.
Period.

Yes, it calls the shutdown API present on the machine, which is not what the poster there was talking about

Games/Benchmarks cannot cause an unsafe shutdown without thermal, electrical fault or overload presence.



i certainly question anyone not new to this with such misguided faulty beliefs.



The GTX 400 and 500 series had a gimped power delivery which could not handle what nvidia coined "Power Virus" loads, the node was also at the time sensitive to electrolysis, the silicon would rapidly degrade the oxide barrier when all parts of the core were active at high temps.

In the case of Furmark, it is a shader intensive application which triggered the former scenario, the power delivery could not safely handle 100% shader loads and there was a missing protection mechanism that would have put the GPU into protect mode (Black screen, Fans maxed) which was added with the 600 series and remains since. Nvidia opted to mitigating the issue with Fermi by throttling the gpu based on executable matching.

The issue with Fermi and Furmark, the system would not have shutdown under those high loads, it would have run itself till the overtaxed components hit critical temperatures and melted, shorting the substrate and likely killing the gpu.

Kepler (and later Pascal) revised the VRM and components used so the issue that would cause failure on Fermi parts doesn't occur, you can run furmark, or kombustor on these cards without much concern.

Power Virus and severe variance should be required testing on any of these 350w+ cards.

I've witnessed many an untimely shutdown while running software with various reasons, some of which were in no way a fault state.

Oc protection, for example power spikes can shutdown even the 1200watt PSU I'm using and HAVE, experience, not ideology is what I'm talking about.

 

[Vayra86]

Ampere, best Nvidia gen in years. What a shitshow. I'm absolutely certain now, I'm staying miles away from anything 3xxx series, even if dirt cheap. The last years have been confirmation upon further confirmation: the node sucks, power consumption through the roof (with all of its side effects on durability), nonsensical marketing-based feature set push and no compatibility with newer versions of it (FG/DLSS3), and far too expensive. Questionable quality from AIBs in especially the higher end segment...(where you'd expect differently most of all places in a stack) Its a big box of no.

Team green is rapidly losing its credibility, on every possible aspect.

I've witnessed many an untimely shutdown while running software with various reasons, some of which were in no way a fault state.

Oc protection, for example power spikes can shutdown even the 1200watt PSU I'm using and HAVE, experience, not ideology is what I'm talking about.

That's the point right. The untimely shutdowns are created not by software, but because the hardware is lacking its checks and balances, or runs into one.

New World was no different. The only thing these events point at, much like the Space Invaders and other 'batch related' issues we've seen before, is some problem in the hardware or the way its driven. And there's been far too many lately... We can probably also attribute part of those issues to pandemic/chip shortage/logistics problems and supply lines. It definitely inspired bait & switch with components on the PCB.

 

[lemonadesoda]

Just wait until mid summer heatwave