Most games crash as GPU usage reach to 100%

[Night]

Hello everyone,

so I have been experiencing a random crash when playing many high graphic games to list some: Gears of war 5, Destiny 2 and Nier Automata.

I used GPU-Z to monitor my computer and I found that nothing abnormal in the moment where the crash happens but that my GPU usage gets to 100% right after that the game crash immediately and usage drop down to 1% or so.

normally the error messages I get related to undetected GPU etc depends on the game this is the message Destiny 2 give me: https://i.imgur.com/wldJ81y.jpg

this is my specs: https://imgur.com/a/6EngolB
my GPU is not overclocked.


I start to think it may be related to my PSU I'm using EVGA Supernova 650 GS, 80+ Gold 650W not sure if it gives enough voltage to my GPU or not.

any suggestions or help will be appreciated

thanks

[Snow]

Ursprünglich geschrieben von Witski:

Ursprünglich geschrieben von Snow:

I didn't get the question. More complicated the picture is (graphical effects, resolution, antialiasing) - more GPU usage. Basically, you can raise usage to 90% in any game by simply increasing the resolution. But I don't see the point of doing that on purpose.

im asking because.. you know, is not stable because of my cpu 😂, in some games i can set 1440p and achieve stable 60fps most of the times if the cpu can handle well the game, but in most cases the game itself won't let me choose a higher res then my native 1080p 75hz monitor, still looking for a program that let me do that

Well, you can try what FeilDOW said. But trying to limit your CPU's performance with GPU running close to 100% won't help performance much. Depending on the game, you might or might not get rid of stutters. RTSS limiting is still the best choice to avoid hardwave limitations.

[Witski]

Ursprünglich geschrieben von Snow:

Ursprünglich geschrieben von Witski:

im asking because.. you know, is not stable because of my cpu 😂, in some games i can set 1440p and achieve stable 60fps most of the times if the cpu can handle well the game, but in most cases the game itself won't let me choose a higher res then my native 1080p 75hz monitor, still looking for a program that let me do that

Well, you can try what FeilDOW said. But trying to limit your CPU's performance with GPU running close to 100% won't help performance much. Depending on the game, you might or might not get rid of stutters. RTSS limiting is still the best choice to avoid hardwave limitations.

Well my game doesn't stutter at all even when the fps drops, i capped it to 60 and for other games to 40,i have 0 freezes or stutter, the only bad thing is when the fps goes under 30 and i start see the difference, by playing in 1440p or 4k the fls doesn't ever go under 40/60 (1440p) 35/40 (4k) so i don't notice nothing..
The only game that bothered me to death was state of decay 2,20 tp 30 fps, the game feels in slow motion.. can't do nothing the gpu goes tp 0% to 20% or 60 but for just some seconds(is not a bad monitoring, i used msi afterburner /the amd softwate in game and task manager, all of them shows that spikes up and down).
I even tried to oc the cpu like in the photo you send me in private, (amazing performance for like 2min of playtime before crash) (( only the game crash not the pc)), then i lower it a bit, 210 if inremember but still no much performance gain but unstable

[Night]

Thank you all for the help you guys are the best...

Ursprünglich geschrieben von notkennyS:

The PSU is perfectly fine, even a 2080ti build would be fine with that.

well thats good to know thank you =)

Ursprünglich geschrieben von notkennyS:

I see that you have a 4K monitor, do you use that for gaming? The 980ti is more of a 1080p card, not 4K.

you know what that might be actually the reason...I do play on that monitor and maybe thats why low demanding games never crash...I will start playing on the other monitor and see how that goes for me.

Ursprünglich geschrieben von Autumn:

Try reinstalling drivers.

Also, try running Furmark for a while, if it clears that fine, then it's not a PSU issue, because Furmark will push a GPU further than any game would.

I did so it never helped but I will try running Furmark to see how that goes.

Ursprünglich geschrieben von Snow:

If nothing helps - just use RTSS limiter to keep usage under 90%. That's basically how you want to play any game anyway.

I will consider that thank you =)

Ursprünglich geschrieben von chiefputsilao✖️ping:

https://www.bungie.net/en/Help/Article/46853
https://www.bungie.net/en/Forums/Post/244958763?page=0
https://www.reddit.com/r/DestinyTechSupport/comments/7iuac2/destiny_2_gpu_crash_curse_of_osiris/
---
summary:
install latest driver or revert to previous one that works
increase gpu and chassis fan speed
(ask bungie; lots of people with brocolli error using nvidia and amd gpu)

I did install latest drvier it never helped thank you for the links buddy =)

Ursprünglich geschrieben von FeilDOW:

If you have tried multiple drivers and it happening in many different types of games your 980 ti could be on the way out. Zero reason for a GPU to crash at 100% load with no OC.

If the 980 ti has a factory oc try down clocking down your MEM and core clocks and see if it helps.

well thats very possible espcially that I read that my card which is: ZOTAC GeForce GTX 980 Ti AMP! Extreme comes overclocked from factory...I will try to underclock it and see what happen...

Ursprünglich geschrieben von SavageSoutherner:

Ursprünglich geschrieben von Fresco Vivir:

The factory overclock( I'm pursuming it isn't founders edition card )might be unstable by the sounds of it , install MSI after burner and reduce core clock /memory clock intill stable. I say this as gears 5 and destiny 2 are very sensitive to even the slightest GPU instability which causes them to crash .

I actually start to think that this is the case mine is overclocked from factory so I will try your suggestion thank you my friend =)

Ursprünglich geschrieben von ☥ - CJ - 8/6/19 - Meja 8/30/19:

2 Things
Did you OC your GPU any? If so cut it back some as that could be the cause..

The PSU
Its possible the PSU could actually be at fault
I built my son a PC a year or 2 ago with a 500 or 550w EVGA PSU
From the start he'd have random crashes in games among other issues, i tried everything from updating the BIOS, considering the RAM to be the problem, etc.

a few months ago i switched out the PSU with a different one that i had in the house, after which all his problems went away and havent had any crashing or blue screens since.

So while the Wattage of the PSU is generally plenty for that system, its possible the PSU itself is faulty.

I didn't overclocked my GPU.

well I have another PSU at home I could also try doing so thank you for the suggestion.

[Snow]

Ursprünglich geschrieben von Witski:

Ursprünglich geschrieben von Snow:

Well, you can try what FeilDOW said. But trying to limit your CPU's performance with GPU running close to 100% won't help performance much. Depending on the game, you might or might not get rid of stutters. RTSS limiting is still the best choice to avoid hardwave limitations.

Well my game doesn't stutter at all even when the fps drops, i capped it to 60 and for other games to 40,i have 0 freezes or stutter, the only bad thing is when the fps goes under 30 and i start see the difference, by playing in 1440p or 4k the fls doesn't ever go under 40/60 (1440p) 35/40 (4k) so i don't notice nothing..
The only game that bothered me to death was state of decay 2,20 tp 30 fps, the game feels in slow motion.. can't do nothing the gpu goes tp 0% to 20% or 60 but for just some seconds(is not a bad monitoring, i used msi afterburner /the amd softwate in game and task manager, all of them shows that spikes up and down).
I even tried to oc the cpu like in the photo you send me in private, (amazing performance for like 2min of playtime before crash) (( only the game crash not the pc)), then i lower it a bit, 210 if inremember but still no much performance gain but unstable

But your problem is not GPU usage, CPU usage is. If CPU can't deliver enough frames - of course GPU usage will go down to 0%.
Now, about overclocking - I've had memory set to 1800MHz on that photo I sent you, Phenom is known to handle anything above 1666MHz badly, so instead of reducing bus clock try just reducing the memory multiplier, it might solve your crashes. Also I'm not sure how your exact mobo handles the voltage, the one I used did just fine with "auto", you might want to try manual overvolting for CPU-NB and/or RAM. Generally - just look up a few threads on how people OC Phen 965. Or let it be, iirc you're going to get new CPU soon anyway.

[Witski]

Ursprünglich geschrieben von Snow:

Ursprünglich geschrieben von Witski:

Well my game doesn't stutter at all even when the fps drops, i capped it to 60 and for other games to 40,i have 0 freezes or stutter, the only bad thing is when the fps goes under 30 and i start see the difference, by playing in 1440p or 4k the fls doesn't ever go under 40/60 (1440p) 35/40 (4k) so i don't notice nothing..
The only game that bothered me to death was state of decay 2,20 tp 30 fps, the game feels in slow motion.. can't do nothing the gpu goes tp 0% to 20% or 60 but for just some seconds(is not a bad monitoring, i used msi afterburner /the amd softwate in game and task manager, all of them shows that spikes up and down).
I even tried to oc the cpu like in the photo you send me in private, (amazing performance for like 2min of playtime before crash) (( only the game crash not the pc)), then i lower it a bit, 210 if inremember but still no much performance gain but unstable

But your problem is not GPU usage, CPU usage is. If CPU can't deliver enough frames - of course GPU usage will go down to 0%.
Now, about overclocking - I've had memory set to 1800MHz on that photo I sent you, Phenom is known to handle anything above 1666MHz badly, so instead of reducing bus clock try just reducing the memory multiplier, it might solve your crashes. Also I'm not sure how your exact mobo handles the voltage, the one I used did just fine with "auto", you might want to try manual overvolting for CPU-NB and/or RAM. Generally - just look up a few threads on how people OC Phen 965. Or let it be, iirc you're going to get new CPU soon anyway.

Yeah ill let it bee everything and enjoy it until i get the new pieces.. i had enough for now

[[☥] - CJ -]

Okay so let me get this right...

980 Ti
2 Monitors
1 of which is 4K
Have you tried using the PC with 1 monitor or the other and got the same results?

[Night]

An update

I started following the suggestions you guys left for me easiest first. I used to play on my 1080p monitor, then I got my new 4k monitor and started playing on it for a while...I guess; lately, I played a lot of less demanding games, so I got no errors. However, with high demand games, I start getting this error, so I decided to lower the resolution to 1920x1080, and with that, I managed to play for two hours yesterday and two hours today without having this error. Now I'm not sure if its completely solved, but for now, I got no errors at least in destiny 2.

I plan to get a new card during black Friday sale, probably going to be either RTX 2080 super or RTX 2070 super depending on their prices at that time.


thank you all for the great help you guys are a great community.

[Night]

Ursprünglich geschrieben von ☥ - CJ - 8/6/19 - Meja 8/30/19:

Okay so let me get this right...

980 Ti
2 Monitors
1 of which is 4K
Have you tried using the PC with 1 monitor or the other and got the same results?

true and I normally play on my 4k Monitor however I switch sometimes from one monitor to another when I downgraded resolution to 1080p I didn't face this error so it is probably that.

[SeriousCCIE]

If you're replacing the card... maybe you can try fixing your 980ti. I am guessing it's been in your possessionf or a long time and you've used it a lot.

I have found that as a card gets older (like 4 or 5 years of regular use), the thermal interface material starts to lose performance. That is the stuff between the GPU itself and whatever is acting as the heatsink.It can dry out.

It might take a few hours to take it all apart and re-assemble again, but it could be worth investigating while you wait for the new card to show up. You also might get some extra performance out of it if you manage to get it running cooler than before.

If you can spare $7 for a tube of something like MX-5 if you don't have anything handy from a previous build (that also hasn't sat unused in a tube for 5+ years--it can go bad in there, too, but in different ways...) you might want to try taking apart the card, cleaning off all the dust and crust and whatnot, cleaning the GPU chip itself with a q-tip and/or lint free cloth or rag, and then applying new thermal material.

I tend to hold onto hardware for a long time; often I replace something, but don't get rid of the older thing--I just use it in something else. Sometimes I find that what I replaced still had a lot of life left in it, but because it was actively used in a primary (or at least, heavily utilized) system, there wasn't much time to take things apart and look at it.

(But I agree that... sometimes holiday sales are the time to upgrade. You'll find prices you can't get things at otherwise... so... maybe consider cleaning plan B if the new hardware price is right!)