RTX 3070 crashing games, Event ID 13 "Nvlddmkm" and Event ID 4101 "Display"

[Ty2525]

So I help people on a pretty regular basis in gaming communities and online. I have helped hundreds of people with small, easy-to-fix issues, to people with major hard-to-fix issues. However, I have yet to be at a complete loss....until now. Got a guy with a Gigabyte 3070 Gaming OC with a crashing issue in games. It only crashes the game with the already mentioned Event ID 13 and 4101 Both GPU driver related in my experience. However, even with Clean installs, DDU, and even new Windows. It fails to run games at stock settings. His PC runs the old card just fine, mind you, being only a 1050 it might not be enough to say for sure it's the fault of the 3070, but nothing else supports anything else being the issue. Otherwise, his system specs are as follows: 

- i9 9900k

- Gigabyte 3070 gaming OC

-Aorus Ultra Pro z390

- 16gb 4000Mhz 

- BeQuiet 700 W gold 

 

Worth noting the issue also occurs at 2133-4000mhz ram speed.

 

With that being said, I had him run Heaven without issues or crashes at stock speed. Using EVGA Precision X1 had him set boost lock to see if maybe the card was crashing due to it's boosting behavior. That proved to be a dead-end as it ran it fine at both these settings, though with boost lock hitched randomly, never crashed nonetheless. I also monitored EVGA precision during these runs, showed no sign of abnormal behavior.

Furthermore, I looked into the PSU, being a red flag at only 700watts (comparatively) however, I can't see how the PSU would be at fault for these issues since it fails to trip any sort of power protections, and does it randomly. Never hard crashes the computer. Also, the Event logs suggest a driver issue, not a hardware issue, from my knowledge. And no other error/critical logs appear. 

 

With that being said, I am at a total loss. I am starting to think the GPU is defective and he can get a refund but I told him to wait since he can't actually get a replacement. So if anyone is familiar with these Event IDs or 3070 crashing games more so than myself, I would love to hear what you have to say. Thanks  

[CommanderAlex]

Welcome to the forums!!!

 

1 minute ago, Ty2525 said:

I have helped hundreds of people with small, easy-to-fix issues, to people with major hard-to-fix issues.

Strange with a new account...

2 minutes ago, Ty2525 said:

- BeQuiet 700 W gold 

What is the exact model number of the PSU? Also, how old is it? The fault may lie with the power supply. Anything under GPU-Z's Sensors tab with PerfCap reasons, also, are you running individual PCIe power cables to each connector on the graphics card?? Are the 12V PCIe rails greater than or equal to 12V??? This may show why your getting crashes. 

[Mark Kaine]

 

28 minutes ago, SpiderMan said:

What is the exact model number of the PSU

I agree with that as I also want to know - usually the 00w Bequiet PSUs are the lower models which doesn't necessarily mean they're bad though. 

 

28 minutes ago, SpiderMan said:

Strange with a new account...

I just think that's funny cause this ain't the only forum for pc issues - albeit probably one of the best... : D

 

33 minutes ago, Ty2525 said:

being a red flag at only 700watts (comparatively)

for what it's worth you may want to look at my sig... no issues whatsoever with a 500w Bequiet PSU and my GPU is as far I know the exact same model as the Gaming OC (with whopping 270w power limit) ... 

 

 

I'd be still curious what kind of errors he's getting though... as that may hint at the issue, hopefully. 

 

Wait... what game is he playing? 

 

is it just one specific game with these issues? 

 

EDIT: did you guys try lower the power limit, say to 90 and maybe core clock to - 25?

 

 

 

[Ty2525]

21 minutes ago, SpiderMan said:

Welcome to the forums!!!

 

Strange with a new account...

What is the exact model number of the PSU? Also, how old is it? The fault may lie with the power supply. Anything under GPU-Z's Sensors tab with PerfCap reasons, also, are you running individual PCIe power cables to each connector on the graphics card?? Are the 12V PCIe rails greater than or equal to 12V??? This may show why your getting crashes. 

Yes so I mostly help people on Discord as that's where most of these communities reside. I've also had bad experiences with people on toms hardware so I haven't been too inclined to use ltt forums. But I'm here haha. The psu is a pure power 11 700watt with two separate PCIE cables for its power. Thanks for the reply!

[Ty2525]

38 minutes ago, Mark Kaine said:

 

I agree with that as I also want to know - usually the 00w Bequiet PSUs are the lower models which doesn't necessarily mean they're bad though. 

 

I just think that's funny cause this ain't the only forum for pc issues - albeit probably one of the best... : D

 

for what it's worth you may want to look at my sig... no issues whatsoever with a 500w Bequiet PSU and my GPU is as far I know the exact same model as the Gaming OC (with whopping 270w power limit) ... 

 

 

I'd be still curious what kind of errors he's getting though... as that may hint at the issue, hopefully. 

 

Wait... what game is he playing? 

 

is it just one specific game with these issues? 

 

EDIT: did you guys try lower the power limit, say to 90 and maybe core clock to - 25?

 

 

 

Thanks for the reply. I haven't touched unndercocking it since the issue happens on stock settings though would boost locking, then lowering it be the best way to go about this? Or do you think just lowering it without touching anything else is? As for which games? Well all of them. And there is no error. Usually just closes, I would imagine the basic error for certain games too. Like for Fortnite, the app hung. And for coldwar, the app encountered an error. Or whatever the messages say normally. But these crashes are not like "you ran out of memory" it's literally just soft crashes that don't tell you anything, and these crashes also happen for many reasons, they usually don't change. And this is the fault of the app, some will say gpu device removed, etc. (not in this specific case but that's just an example, games just in general are pretty useless at giving errors that are helpful) That's why I Included the Event logs. As it's the only thing that it gives us. The reason I mention the PSU is because I know 3000 cards tend to trip protections, though I was wondering if this was a symptom of PSU issues that I wasn't aware of. Thanks for the input!

[Mark Kaine]

1 hour ago, Ty2525 said:

Thanks for the reply. I haven't touched unndercocking it since the issue happens on stock settings though would boost locking, then lowering it be the best way to go about this? Or do you think just lowering it without touching anything else is? As for which games? Well all of them. And there is no error. Usually just closes, I would imagine the basic error for certain games too. Like for Fortnite, the app hung. And for coldwar, the app encountered an error. Or whatever the messages say normally. But these crashes are not like "you ran out of memory" it's literally just soft crashes that don't tell you anything, and these crashes also happen for many reasons, they usually don't change. And this is the fault of the app, some will say gpu device removed, etc. (not in this specific case but that's just an example, games just in general are pretty useless at giving errors that are helpful) That's why I Included the Event logs. As it's the only thing that it gives us. The reason I mention the PSU is because I know 3000 cards tend to trip protections, though I was wondering if this was a symptom of PSU issues that I wasn't aware of. Thanks for the input!

No problem, and I see. Yes like I said just lower power limit to 90% and set core clock to - 25... IF that or even a bit lower fixes it then yes, it's likely the PSU, as I've read that also a lot that basically down clocking and allowing a bit less power draw can fix those issues... 

 

 

Thats also btw why I did it, I just don't want to risk it with my PSU, although everything seems to be running fine, I also forgot to lower clocks / power limit once, played a demanding game and nothing happened lol... 

 

Still, I'm getting above average benchmarks with these settings so I don't feel like I'm losing out much if anything at all. 

 

I'd say give it a try... as it seems a likely fix. 

 

 

see, it's a really simple setup, also temp limit will be at 81C and I'd leave it there as it obviously also plays a role... 

[Ty2525]

11 hours ago, Mark Kaine said:

No problem, and I see. Yes like I said just lower power limit to 90% and set core clock to - 25... IF that or even a bit lower fixes it then yes, it's likely the PSU, as I've read that also a lot that basically down clocking and allowing a bit less power draw can fix those issues... 

 

 

Thats also btw why I did it, I just don't want to risk it with my PSU, although everything seems to be running fine, I also forgot to lower clocks / power limit once, played a demanding game and nothing happened lol... 

 

Still, I'm getting above average benchmarks with these settings so I don't feel like I'm losing out much if anything at all. 

 

I'd say give it a try... as it seems a likely fix. 

 

 

see, it's a really simple setup, also temp limit will be at 81C and I'd leave it there as it obviously also plays a role... 

So I had him do -100 on the core and power to 70% but it still crashed. So I had him do -1000 on the memory and -100 on the core and it appears to be working fine. I will update if anything changes, though it didn't crash in a few hours and as a problem that would consistently occur within a few minutes I would say we found the suspect. Though we don't actually know yet if it's result of the core of memory settings; I'll separate them into different tests and see if I get the same results. I will have him do like -300 and 60% power on the core just to ensure the -1000 memory clock didn't just pull the power consumption down enough to fix the issue. But if I had to take a guess I would imagine this sounds like the vram just can't run stock for whatever reason, power limitation/defect, driver issue, etc. Hopefully we'll find out today what the issue is specifically. Again thanks for everyone that has replied. Never even thought of underclocking!

[Mark Kaine]

That's good news this method worked, and it looks to be indeed a power draw issue. 

Though that's the wrong conclusion :

1 hour ago, Ty2525 said:

But if I had to take a guess I would imagine this sounds like the vram just can't run stock for whatever reason, power limitation/defect, driver issue, etc. Hopefully we'll find out today what the issue is specifically. Again thanks for everyone that has replied. Never even thought of underclocking!

Thats what people found out yeah the ram is drawing a lot of power, but this isn't a fault of the card - what we did here was testing the power supply, and it failed gloriously 

 

 

so getting a good ~850w PSU should by all means fix the issue... 

[Ty2525]

3 hours ago, Mark Kaine said:

That's good news this method worked, and it looks to be indeed a power draw issue. 

Though that's the wrong conclusion :

Thats what people found out yeah the ram is drawing a lot of power, but this isn't a fault of the card - what we did here was testing the power supply, and it failed gloriously 

 

 

so getting a good ~850w PSU should by all means fix the issue... 

Well we're getting different results now. But what I meant at the time is that the ram appeared to be the issue, not the psu. The reason is because with -300 on the core and 60% power, it still crashes but with -1000 and otherwise stock settings it didn't. This would pull more power than the previous tests while having no issues. However, we're pretty much back at square one as currently it still crashes underclocked with the handful of different configs we've tried with core, memory, and power settings. But we have more to test so I will update when this happens. Can't rule out the PSU but also can't blame it for the issues without more consistent results. 

[Ty2525]

Just a quick update, -1000 and 60% crashed it. But -1000 and 100% power has not. I assume the 60% power was too low for the card to run? Maybe? Idk but it was crashing right away and with -1000 memory and 100% power its fine for now. Will update again if anything changes, but I would love to hear anyone's opinion on why -1000 memory would fix the card's stock crashing issue since its running within spec, since again, that's 100% power and max boost on the core. It doesn't appear to be directly related to power though I will leave that on the table since its still possible the PSU is just randomly broken. 

 

Update: The card ran way longer than normal before crashing at -1000 memory and 100% power. Trying -1000 and 70% power as it has showed promising results. Will update AGAIN what the results are. But again, if anyone knows why -1000 mem and 70%-100% power appears to run the card better than stock settings, let me know. Thanks in advance

[Mark Kaine]

I mean that is still the most likely scenario though memory is pulling too much power, psu can't handle it, that's the whole point of down clocking stuff in this case... and I actually forgot that people were downclocking the memory too... 

 

I mean it's just very unlikely the gpu is defective, more like power delivery is the issue.

 

I know (literally lol) it sucks having to buy a new PSU but that's still better than returning a - most likely - perfectly fine gpu, especially in the current supply situation...

 

 

Also... I could have said this earlier - but I also wanted to wait on the results first - it's actually recommended to use "studio drivers" because they just have less issues overall and are more stable apparently. You guys could also try that, obviously. 

 

Also what's the power supply exactly...? I still don't know lol... 

 

 

[Ty2525]

1 hour ago, Mark Kaine said:

I mean that is still the most likely scenario though memory is pulling too much power, psu can't handle it, that's the whole point of down clocking stuff in this case... and I actually forgot that people were downclocking the memory too... 

 

I mean it's just very unlikely the gpu is defective, more like power delivery is the issue.

 

I know (literally lol) it sucks having to buy a new PSU but that's still better than returning a - most likely - perfectly fine gpu, especially in the current supply situation...

 

 

Also... I could have said this earlier - but I also wanted to wait on the results first - it's actually recommended to use "studio drivers" because they just have less issues overall and are more stable apparently. You guys could also try that, obviously. 

 

Also what's the power supply exactly...? I still don't know lol... 

 

 

Now currently it still crashes though far less often than at stock settings at -1000 mem 80% power. And the PSU is Purepower 11 700watt from BeQuiet. Here is a link if you need more info: https://www.amazon.com/dp/B07PK1XDX7?ref=ppx_pop_mob_ap_share

Hopefully links are not automatically deleted lol

 

But I still don't see how the PSU is to blame. Again, I am leaving that option open for discussion, but nothing about how it behaves looks like a power issue imo. Power delivery looks way more concerning than PSU. The core pulls more power so when its lowered -300 and the power limit is set to 60% but still has the same issue, but that issue is fixed with -1000 memory clock. It looks more like a driver/firmware/memory problem to me. Keep in mind that the core is actually set to its max boost when the memory is set to -1000 so its pulling more power in this configuration than it would be in others that have the same issue. With that being said, PSU issues are hard to diagnose if it isn't blatantly obvious it's not working. With that being said, the PSU isn't something that is hard to test. I am sure a microcenter will swap the unit with a known working one to test if that's the problem. I just think that's a lot of work without direct evidence its the problem. 

 

As for the driver, would a studio driver work in games enough to test the card? Also, do you think the Vbios could be bad and how would one find a compatible one to flash to the card, I am not too familiar with Vbios databases lol. 

 

Anyways. Lets address the biggest concern. Which is how would one determine if the memory is to blame. Really, all we have to work with is that the card's issues still persist, but less often with -1000 memory clock. Everything else is a dead-end. I would love to here from more people what is another easy-to-test method of determining the problem. Because at the moment, the only other option is really to put the card in another computer/change psus and see if that makes a difference. Both of which are not the easiest to do in the current world. 

 

But again, we still have no idea for sure what's wrong with the card if anything, so I am open to testing anything that will give us more context to what the card is struggling with, and if the card is to blame for those struggles. Thanks

[Maluke]

On 1/17/2021 at 9:16 PM, Ty2525 said:

Now currently it still crashes though far less often than at stock settings at -1000 mem 80% power. And the PSU is Purepower 11 700watt from BeQuiet. Here is a link if you need more info: https://www.amazon.com/dp/B07PK1XDX7?ref=ppx_pop_mob_ap_share

Hopefully links are not automatically deleted lol

 

But I still don't see how the PSU is to blame. Again, I am leaving that option open for discussion, but nothing about how it behaves looks like a power issue imo. Power delivery looks way more concerning than PSU. The core pulls more power so when its lowered -300 and the power limit is set to 60% but still has the same issue, but that issue is fixed with -1000 memory clock. It looks more like a driver/firmware/memory problem to me. Keep in mind that the core is actually set to its max boost when the memory is set to -1000 so its pulling more power in this configuration than it would be in others that have the same issue. With that being said, PSU issues are hard to diagnose if it isn't blatantly obvious it's not working. With that being said, the PSU isn't something that is hard to test. I am sure a microcenter will swap the unit with a known working one to test if that's the problem. I just think that's a lot of work without direct evidence its the problem. 

 

As for the driver, would a studio driver work in games enough to test the card? Also, do you think the Vbios could be bad and how would one find a compatible one to flash to the card, I am not too familiar with Vbios databases lol. 

 

Anyways. Lets address the biggest concern. Which is how would one determine if the memory is to blame. Really, all we have to work with is that the card's issues still persist, but less often with -1000 memory clock. Everything else is a dead-end. I would love to here from more people what is another easy-to-test method of determining the problem. Because at the moment, the only other option is really to put the card in another computer/change psus and see if that makes a difference. Both of which are not the easiest to do in the current world. 

 

But again, we still have no idea for sure what's wrong with the card if anything, so I am open to testing anything that will give us more context to what the card is struggling with, and if the card is to blame for those struggles. Thanks

Hey Ty, I have the exact same issue as your friend or whoever you're helping. 3070 gigabyte OC gpu.
Did you figure out what the problem was? Any solution?
If you don't mind i would really appreciate if you could DM me on discord. Maluke#1322

[Mark Kaine]

57 minutes ago, Maluke said:

Hey Ty, I have the exact same issue as your friend or whoever you're helping. 3070 gigabyte OC gpu.
Did you figure out what the problem was? Any solution?
If you don't mind i would really appreciate if you could DM me on discord. Maluke#1322

this is a 'known" psu issue. 

 

maybe you should make a new thread though, and post full specs in order for people being able to help you. 

 

 

[Maluke]

On 3/4/2021 at 4:26 PM, Mark Kaine said:

this is a 'known" psu issue. 

 

maybe you should make a new thread though, and post full specs in order for people being able to help you. 

 

 

Known psu issue? How known, because i can't find anything on any forums that says it's a psu issue.
I have a corsair 750w (rm 750x)
3070 gigabyte OC
i7 8700k
Vipersteel 16 gb ram