• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Can a game kill a GPU?

Man of Honour
Joined
25 Oct 2002
Posts
31,737
Location
Hampshire
Would be interesting to know if these dead cards were running at stock or overclocked.

I think the point is that regardless of whether software should be able to kill a GPU or not, if it exposes a hardware flaw that results in death, then it's still a problem. I remember back in the day there were reports about Furmark damaging GPUs.
 
Soldato
Joined
1 May 2013
Posts
9,710
Location
M28
Would be interesting to know if these dead cards were running at stock or overclocked.

I think the point is that regardless of whether software should be able to kill a GPU or not, if it exposes a hardware flaw that results in death, then it's still a problem. I remember back in the day there were reports about Furmark damaging GPUs.

wouldn't that need to override the settings in the BIOS to limit temp and voltage?
 
Soldato
Joined
7 Dec 2010
Posts
8,249
Location
Leeds
Not possible. Software cannot kill hardware unless the hardware already has a design flaw (inadequate cooling etc).

Software can kill components. What you said is not true.


Drivers as an example have killed gpus before, furmark as an example kills cards that don't have protection for the tool in drivers or hardware and has killed many cards, google the subject you will find many times gpus and other components killed by software. How about viruses that rewrite/clear the bios on hardware and brick them or viruses that are designed to damage hardware by overheating it or making it fail by other means.

Good example of a well known worm Stuxnet that targets supervisory control and data acquisition systems and is responsible for damage to the nuclear program in Iran.
 
Last edited:
Soldato
Joined
7 Dec 2010
Posts
8,249
Location
Leeds
Seems to be mostly EVGA cards affected? Not a good look.

The 3000 series from EVGA have power balancing issues on their pcie slot and power pcie connections, EVGA cards have been the worst cards for issues with the 3000 series since day one, damaging motherboards by pulling too much power via the slot and the load balancing issue damaging the cards and causing the red lights of death issue.

So this tells me this game is acting like a power virus and the drivers are not detecting its behaviour as dangerous to the hardware. Many programs in the drivers are limited to stop such things eg Furmark, that is classed as a power virus and the driver stops it from using the power it requests and limits it.
 
Soldato
Joined
7 Mar 2008
Posts
2,614
Location
Kent
I'm not surprised its an EVGA(well I am actually I thought they would have fixed this by now) my GTX 1080 from them popped and fried itself. It can't really be the software it has to be a hardware problem that is exposed by the software.
 
Soldato
Joined
7 Dec 2010
Posts
8,249
Location
Leeds
Nvidia's unofficial retail arm and yes it's not a good look for them but by the time next generation of video cards roll around all will be forgotten/forgiven by most.

No chance, I was one to always buy EVGA Classified and Kingpin cards or ASUS cards before (Asus still make good top end cards, just not had one in a while), EVGA never again after all the issues with their gpus this generation and the poor quality componets used on a so called premium cards and same applies to their PSUS. The last card I had from them was the 2080ti kingpin, it went back same day it was that cheaply made and the coil whine was unreal on it, last good card I had from them was a 980 Ti Classified before it all went down hill for EVGA it seems.

Now I only buy cards after seeing reviews of them and seeing the boards taken apart so can see how they are built and with which components they have used.

EVGA this time could not even get the right sized thermal pads for their 3000 cards with the ftw3 cooler and plastered thermal putty all over the place where they couldn't get the right sized thermal pads, well because they designed the cooler wrong and didn't think about what thermal pads they could fit and buy when designing it. Then the shocking basic components on a premium board and all the power issues with them and failures due to load balancing or slot power that damages your motherboard. They are a bad joke now. EVGA never again..
 
Soldato
Joined
26 May 2014
Posts
2,953
No. Only inadequate cooling can kill a card (in this way), and running a game with no framerate cap is a way to heat it up as much as possible. That's exactly what's happening here, since New World runs uncapped by default. People will attribute blame in the wrong place of course, and blame the game for somehow magically frying their card, rather than it being because it was a poorly-designed piece of crap which fried itself.

Seems to be mostly EVGA cards affected? Not a good look.
A Twitch streamer I occasionally watch (LobosJr) had two EVGA 3090s die on him (months ago). The one he bought and the replacement for it. He was warning people away from them on his stream at the time. Seems they have major cooling problems. Though so do most 3090s since they're 350W+ cards, which is truly absurd, and have VRAM on the back getting very little cooling. No wonder they're cooking themselves to death when put under real stress.
 
Associate
Joined
9 Apr 2017
Posts
188
Location
Eve Online
I would assume most of the cards failing to be first-gen 3090 FTW3s that had faulty power management/voltage controllers and were prone to the fuses popping and the 'red light of death'. There are probably still enough of those in circulation that a popular game with conditions conducive (a high fps uncapped screen) can kill the last of them.

How easy is it to kill GPUs with software? It should be very difficult. That said, certain loads can create transient power spikes. When I first shunt-modded my 3090 FE, for example, there were certain conditions that could instantly cause it to trip OCP on a ROG 1250w psu (the analogue controller on a 1600 T2 was fine however). The issue was not triggered by furmark, 3dmark Extreme or workloads of that kind but rather very high uncapped fps scenes (certain menus, the instant of a 'loaded world' appearing in a game, etc.). These situations can create transient power spikes that trip things which sustained loads don't. And if a transient power spike can trip OCP on a ROG Thor, it stands to reason that the same spike could affect an imperfect pcb component too. In that case I solved the issue just by moving to a bigger, analogue PSU for the card - but I'd imagine something similar to be at play here, linked to the plethora of first gen FTW3 pcbs that are (or were) still in the wild.
 
Permabanned
Joined
22 Oct 2018
Posts
2,451
It's not exactly difficult to cap the framerate in the NVIDIA Control Panel and seriously given a high level 3000 series card people should be doing that just in case ~ some games run completely out of control without a cap.
 
Associate
Joined
18 Sep 2020
Posts
13
Framerate capping doesn't really help.
I had my FPS capped to 60 and still got a black screen + fans revving while walking around.
Luckily I was able to boot again but Im staying well clear of NW for now.
And yes - I have one of the first generation 3090 FTW3 Ultra with latest bios, drivers etc.
 
Back
Top Bottom