Can a game kill a GPU?

[Purgatory]

It's interesting to see EVGA fanboys scrambling to defend this, it's worse than NVIDIA's.

Yes they are basically defending a company that sells them faulty and badly designed goods that they buy..

Some have a short memory and forget things like this have happened before, example here :-

EVGA GTX 1080s and 1070s allegedly exploding due to improper VRM cooling

https://www.extremetech.com/computi...-allegedly-exploding-due-improper-vrm-cooling

 

[Benass]

Here is a reasonable explanation of how hardware should protect itself, and software do not break hardware.

https://www.youtube.com/watch?v=dbCrNBpW-RQ&ab_channel=ActuallyHardcoreOverclocking

 

[Harlequin]

Software can exposes flaws in hardware protections....

 

[V F]

Here is a reasonable explanation of how hardware should protect itself, and software do not break hardware.

https://www.youtube.com/watch?v=dbCrNBpW-RQ&ab_channel=ActuallyHardcoreOverclocking

"people coming out of theyr gaming lair after the gpu burnt , looking out into the sunlight, "its a NEW WORLD"

I find it ironic it was the GTX 590 that blew up like this exactly 10 years ago and also had shunt resistors.

 

[Benass]

Software can exposes flaws in hardware protections....

EVGA got caught with their pants down

 

[dvinell]

Watched the video. I'm still 100% convinced software doesn't kill hardware.

It's pretty obvious to me that software can kill hardware.

For a Graphics Card, a design goal might be "We try very hard to design something that will cope with all the possible software out there, and mostly we get it right".

But if the hardware has a latent defect such as very high frame rates causing power draw that exceeds the limits for some reason and kills components, that's software killing hardware.

It shouldn't happen, but it can.

 

[Twinz]

Yes they are basically defending a company that sells them faulty and badly designed goods that they buy..

Some have a short memory and forget things like this have happened before, example here :-

EVGA GTX 1080s and 1070s allegedly exploding due to improper VRM cooling

https://www.extremetech.com/computi...-allegedly-exploding-due-improper-vrm-cooling

I bought my 1080Ti from EVGA because of how they handled the cooling issue on the 1070's. -Giving people the option to send it in, or EVGA would send the thermal pad kit to customers.

That being said, their lack of communication on the self-destructing 3090's is dissapointing.

 

[KompuKare]

Look like it's just as well EVGA have good CS as customers will often need it
Bit ofa mixed mind about that: if you expect that any manufacturer could have issues, then buying from one with good CS makes sense.
On the other hand, of they keep making design mistakes or have poor QA so you're more likely to need to use their CS? Not so good
I used to buy Crucial because their CS was okay and their RMA address is (was?) in Glasgow. However, I don't think too much of their QA as I actual had a few sticks of their go bad.

 

[Grim5]

Look like it's just as well EVGA have good CS as customers will often need it
Bit ofa mixed mind about that: if you expect that any manufacturer could have issues, then buying from one with good CS makes sense.
On the other hand, of they keep making design mistakes or have poor QA so you're more likely to need to use their CS? Not so good
I used to buy Crucial because their CS was okay and their RMA address is (was?) in Glasgow. However, I don't think too much of their QA as I actual had a few sticks of their go bad.

that's what I was just thinking, good CS sure but what's the point of the product they sells results in higher failure rate than the competing product. Having better CS costs money, it means Evga spends more on CS than others - but who pays for that? Paid for by all the corners cut in gpu production

 

[KompuKare]

that's what I was just thinking, good CS sure but what's the point of the product they sells results in higher failure rate than the competing product. Having better CS costs money, it means Evga spends more on CS than others - but who pays for that? Paid for by all the corners cut in gpu production

You'd almost wonder if they have competing managers in charge for CS, QA, and design bean counter'ing.
Bean counters cut corners.
QA doesn't spot the mistakes, or let's them pass.
CS then has to sort them out.
Maybe they should take inspiration from those Carlsberg commercials!
Back in the day, didn't BFG have excellent CS too, until too many failures killed the company?
I had one of their 8800GT but once it developed the solder bumb defect, BFG had folded. Not that there was much BFG could have done once Nvidia supplied them with chips with bad solder in the first placed.

 

[Blackjack Davy]

EVGA's Pascal cards were literally catching on fire at one point.

EVGA responded by sending out thermal pads to people, issuing a new BIOS with a more aggressive fan curve and eventually just plain replacing the affected models with FTW2 revisions. Personally, I've always thought that EVGA's coolers weren't that great. Most of the EVGA cards I've had have been pretty loud. The reason most people shill them so hard is because of their warranty service, but even that isn't so great for UK owners any more. They shut down their UK service centre, so now you have to ship the card all the way to Germany. Anecdotally, I've also found their email support to be pretty unhelpful and disinterested.

As for what's best right now, it's Asus without question. I really don't like Asus as a company either (and they're apparently bad for warranty service too), but they undisputably knocked it out of the park with their recent coolers. The Strix dominates everything in noise-normalised testing, but even the cheaper TUF models are better than almost anything else on the market.

EVGA, Colour me surprised. I had one of those 1080 ftw's still do infact I'm too embarressed to pass it on to anyone worst card I ever owned poor design horrific coil whine overheating VRM's that lead to it black screening all the time, then it turned out there was an issue with the cards. Never owned another EVGA since and they still havn't learned their lesson. The BIOS revision, I remember that I put that in and... the fans wouldn't turn anymore. It was hilarious. Lucky there was a 2nd BIOS you could switch to or the card would have to go in the bin...

Yes this absolutely can happen, it's not the fault of the game it's a fault with the video cards being designed and tested around what is commonly expected use, rather than the extremes.

Some apps can more perfectly load the GPU than others. Most games even though they might read say 99% GPU load aren't actually making efficient use of the GPU, complex sets of instructions that utilize different parts of the GPU can be inefficient if some parts of the GPU are waiting for other instructions to finish. If you more perfectly load the GPU pipeline with work then you can cause unusually high power draw and temps which can damage either the VRMs, the GPU or overload capacitors.

This happened in the past with Furmark and I believe someone at AMD a while back called this a "power virus" because it draw abnormally large amounts of power which is not typical of games. But just because things aren't typical of games doesn't mean they can't happen in the wild, which is specifically why benchmarking tools test the extremes. If I had to bet I'd say maybe its the menu doing this, if you have a simple 2D menu that has a very simple set of instructions to draw it, and its uncapped so you have thousands of frames per second then that's a good candidate for the problem. People have said it happens after you recall to a new area and that's right as the 2D loading screen pops up to load another zone. People have also complained about extremely high GPU temps in the menu and while queuing in game.

Sidenote, it's a very fun game, I've been playing the beta and love it.

Well yeah Furmark loaded up all the cores in a way that games don't so everything overheated I remember trying it once on a 4870x2 and watching the VRM's hitting 120c+ at which point it shut it down before they went pop the temp was still rising. Its the same issue the overheating VRM's on the 30xxFE cards were having during mining they constantly load the card in a way that games don't, they weren't designed for that kind of use, mining cards exist for a reason. Later manufacturers found ways of detecting Furmark and stopping it from doing its thing I remember early attempts you simply renamed the .exe and away it would go again, later driver revisions were more sophisticated but the principles the same its detecting and load limiting it rather than a more robust design of the hardware.

The EVGA though was bugging out in normal use not Furmark etc a regular gaming session would break it. According to rumour and hearsay the US built versions are better quality the euro cards are just rubbish.

 

[Grim5]

Uh oh, poor 3090 FTW3 - that's the fan controller on the pcb blowing up because it tried to run the fans at 2 million rpm

 

[Purgatory]

Uh oh, poor 3090 FTW3 - that's the fan controller on the pcb blowing up because it tried to run the fans at 2 million rpm

Like I said FTW3 issue and their silly chip for all the sensors and fan control. No other AIB or FE has the issue and no AMD card. Jays made up rubbish to protect EVGA as they supply him with neverending free cards and many of the same, of course he was not going to blow his up because as he said in the video he called EVGA earlier in the day and I'm sure they told him to set a fan limit or a fixed fan speed.

Now people will believe it's a badly designed system and they use that as their protection instead of the VRM controllers safety features that are built in and designed for 30 series cards the VRM controllers this gen are even designed for the 30 series, if you google the VRM controllers part numbers and read the specs they state all the protections and how to enable them and set the limits on them. Another EVGA fail and reason I am really annoyed at them is they never own up to a design fault or recall the faulty cards and send out new designs, they pretend to do but really same cards with new bios limits to basically make the cards not explode or overheat as has happened before and then sending thermal pads to users ... This is not a thermal pad fix, this needs a recall and new designed cards without their silly chips and using the design Nvidia gave them without them adding stuff that can cause such unknown issues to happen.

 

[4k4uNuNu]

Jayztwocents seemed adament in his video(s) that the game was solely responsible, to the point he accuses Amazon devs of patching out the problem stealthily in a cover up.

 

[jonneymendoza]

Jayztwocents seemed adament in his video(s) that the game was solely responsible, to the point he accuses Amazon devs of patching out the problem stealthily in a cover up.

He tested it on a open bench ….

 

[Purgatory]

Jayztwocents seemed adament in his video(s) that the game was solely responsible, to the point he accuses Amazon devs of patching out the problem stealthily in a cover up.

Yes he's full of.... even tried to add AMD cards to the mix and other AIBS to deflect from EVGA being the only ones with the fault.. with zero proof.. Now the proof is right there and as expected the silly chip EVGA added to the FTW3 cards and the silly sensors that were meant to protect the card but clearly was not designed right and made the controller turn up the fans to unreal levels and fried itself.

 

[Nexus18]

Well it was rather funny, they release a note saying, "not our fault".... but happen to release a patch anyway which might just so happen to stop any further issues....

If the game wasn't to blame then why the need to release a patch???? If this was a fault specific to the card then why hasn't this issue presented itself far earlier? Metro and cyberpunk absolutely hammer the gpus when ray tracing fully used at high res. (especially metro, even with a heavily undervolted 3080 FE, that game makes the fan work at a ridiculous speed compared to any other games)

Also, there have been posts on reddit etc. from people who have got amd cards and other brands of 3090s and even 3080s too, of course you could say they were made up.....

 

[Purgatory]

He tested it on a open bench ….

He also called EVGA earlier in the day he said, they would have told him to set fan limits and not allow the fans to run out of control, you know the cause of the issue that they know. He rigged the testing knowing the fault, this is why he wouldn't live stream the testing when asked on twitter and replied in an arrogant way well it's boring and you will see the video in the end anyways.. ohh wait because he was filming a movie that he can edit later if the cards blew up and use one of his spares at end to show all is well.

 

[Purgatory]

Well it was rather funny, they release a note saying, "not our fault".... but happen to release a patch anyway which might just so happen to stop any further issues....

If the game wasn't to blame then why the need to release a patch???? If this was a fault specific to the card then why hasn't this issue presented itself far earlier? Metro and cyberpunk absolutely hammer the gpus when ray tracing fully used at high res.

Also, there have been posts on reddit etc. from people who have got amd cards and other brands of 3090s and even 3080s too, of course you could say they were made up.....

To protect themselves from lawyers and bad press because of one AIB that can't make cards right, reality is most peoples cards died in the game and not just on the menu, so the limits have not been changed in the game where people had failing cards. It's nothing to do with the game, the only thing to blame on the game devs is they should put limits on menus as many game devs do but again many don't too, many games with menus that shoot to silly fps too. The issue is the cards protection was being done by their "EVGA" chip they added for all the silly sensors they added on the card as a fake selling point and then turned into the point of failure.


Untill these people show proof of other cards going bang then it never happened, there is zero proof so far of any other card failing AMD or FE or any other AIB. The EVGA cards the proof is all over their own forums of this issue since day one of release of these cards with the FTW3 setup, 3080s and 3090s etc.

 

[GloriousMess]

Gamers nexus is about to release a video of the exploding gigabyte power supply. I can't remember the model name, it's their entry 650w or something - it doesn't have over current protection.... so when you exceed 650w from the wall the psu does one of two things: you hear a bang sound as it blows up or you smell smoke because it's on fire

As a side note I regret buying an X570 motherboard, or in fact anything at all, from Gigabyte. Their product quality is sub-standard and has been for years, based on purchases by my partner and I.