Contant crashing

[not so Important]

Hello, so I finally upgraded my graphics card from a GTX 1070 to an RTX 3060Ti and I wish I could say that it's been great. I've been experiencing nothing but constant crashes while gaming for what it seems no reason at all. For context, here is the full list of my PC parts:

AMD Ryzen 2600X

ZOTAC GeForce RTX 3060 Ti Twin Edge

Gigabyte Aorus B450 Elite V2

Corsair Vengeance 16GB DDr4 3333MHz (2x8 kit)

be quiet! Pure Power 12 M 650W

Samsung 980 NVME M.2 SSD

Seagate 1TB HDD

Seagate 500GB HDD

Western Digital 240GB Green SSD

And if it matters, 6 case RGB case fans.

 

Before the upgrade, I would rarely ever crash no matter how much strain I put on the PC, but now I get constant crashes (every 2-5 minutes) when playing games. The games I've tested are Cyberpunk 2077 with the Patch 2.0 and the Phantom Liberty DLC. I've also tried Forza Horizon 5. Strangely enough, the benchmark apps do not crash when I run them. The benchmarks I've tried running are Heaven, 3DMark and UserBenchmark. During gaming and doing benchmarks, I've noticed that the graphics card tops out at around 76C, while my CPU at around 75C. The only game that did not crash at all is Rocket League.

I also want to mention that most of the time only the game crashes, but sometimes I get BSOD.

 

I have tried literally everything to try to fix the crashes or at least diagnose the problem but to no avail. I have tried doing the following:

• I tried reinstalling Windows 10 (twice)
• I tried fixing any corrupt system files on Windows 10 (like 5 times)
• I tried installing the latest GPU drivers, as well as older versions.
• I've tried to uninstall the drivers using DDU
• I've tried updating all of the Runtimes, DirectX, PhysX, Geforce Experience, etc
• I've tried installing Windows 11

Yesterday, just before going to sleep, I did some digging on how I could possibly diagnose the problem, and I saw that someone suggested unplugging a few hard drives and SSDs. And if that fixed the problem, it was probably the PSU being too weak for the system and needed upgrading. Well, today, I tried doing that, and to no avail.

I had a theory that maybe the RT cores on the 3060TI are somehow faulty so I tried running both of the games without ray tracing but they kept on crashing regardless. Then, I tried running the games on the lowest possible settings - yet they still crashed.

Then my buddy had an idea that maybe it's the NVME SSD that I was using that was faulty, so I tried moving the games to the WD Green SSD yet they still crashed.

 

Unfortunately, I do not have any of my old parts (motherboard or graphics card).

 

I've never ever had a problem like this before and I have sadly ran out of things to try :(

[Tetras]

Can you put the graphics card in someone else's PC? It sounds like it might be broken.

 

You could try underclocking/power limiting the card, if it helps then that's contributory confirmation it doesn't work properly.

[FPS_Lego]

It is a damaged graphics card. Try putting it in a different pcie just to check and see if it makes a difference. But it definetly looks like a damaged gpu as you reinstalled the OS like 3 times and did a full DDU and it didnt make much of a difference

[Zuruox]

yeah try underclocking it I've had 1660 Super before that had similar situation like yours and underclocking does the trick. Also this might be unrelated but I have a RX 5700XT which crashed when playing games but strangely fine when running benchmarks, all I had to do was slightly underclocking the Core Clock and slightly Undervolting it and then turned off wallpaper engine in which seems to be the main issue for mine. After all that VOILA!
The GPU went smooth sailing no crashes at all.

[Jon-Slow]

If you did a proper DDU or a fresh install of the OS and things didn't get fixed then your card might be damaged.

[not so Important]

The reseller I bought it from said they can change the card to a different one (same model) if I come across any issues, is it worth bothering with that model of the card? Or should I just request a refund?

[not so Important]

For what it's worth, the cards they're reselling are ex-mining card. The reason why that didn't deter was because since I dabbled in crypto mining myself, I know that if the original guy who bought all the cards to mine with them was smart, not only made sure that the cards were not overheating but he probably also undervolted them.

[Tetras]

15 minutes ago, not so Important said:

The reseller I bought it from said they can change the card to a different one (same model) if I come across any issues, is it worth bothering with that model of the card? Or should I just request a refund?

Did underclocking/power limiting it help, or no?

 

I don't think it matters what model the card is, if it works.

[MadAnt250]

Like Tetras said; try the card in another pc.

 

Also, it's 2023. With the type of games you play you maybe should upgrade to 32GB of RAM.

[not so Important]

50 minutes ago, Tetras said:

Did underclocking/power limiting it help, or no?

 

I don't think it matters what model the card is, if it works.

I'll have to try that after work. Will post an update later.

[Sawa Takahashi]

1 hour ago, not so Important said:

For what it's worth, the cards they're reselling are ex-mining card. The reason why that didn't deter was because since I dabbled in crypto mining myself, I know that if the original guy who bought all the cards to mine with them was smart, not only made sure that the cards were not overheating but he probably also undervolted them.

Mining cards that were cared for are normally as stable as new ones. Especially if they were undervolted and power limited while mining. That is a common way of increasing mining efficiency _and_ it keeps the cards healthy. In my opinion, you can trust mining cards but even then you may encounter a bad one from time to time.

 

As for the PSU, it should have more than enough power to run your system. And yes, trying to leave only the boot drive plugged is a good practice to isolate faults.

 

Do you know GPU-z ? This is software that monitor closely your GPU. Under the tab sensors, you can monitor a lot of things and even log the graphs into a log file. Maybe you can find something with this data.

Have a nice day !

[not so Important]

23 minutes ago, Sawa Takahashi said:

Mining cards that were cared for are normally as stable as new ones. Especially if they were undervolted and power limited while mining. That is a common way of increasing mining efficiency _and_ it keeps the cards healthy. In my opinion, you can trust mining cards but even then you may encounter a bad one from time to time.

 

As for the PSU, it should have more than enough power to run your system. And yes, trying to leave only the boot drive plugged is a good practice to isolate faults.

 

Do you know GPU-z ? This is software that monitor closely your GPU. Under the tab sensors, you can monitor a lot of things and even log the graphs into a log file. Maybe you can find something with this data.

Have a nice day !

I've certainly have heard of GPU-z, I guess I could also take a peek there once I'm able. However, unless something is really, really wrong, I don't think it will be of much use to me since I am really not technologically inclined

[Sawa Takahashi]

1 minute ago, not so Important said:

I've certainly have heard of GPU-z, I guess I could also take a peek there once I'm able. However, unless something is really, really wrong, I don't think it will be of much use to me since I am really not technologically inclined

You can always post a screenshot of your results here and we'll see if something stands out

[not so Important]

UPDATE: during my lunch break, I underclocked the card by turning the power limit from 100% to 50%, and I also took of 502MHz from both the core and the memory clocks and ran Cyberpunk 2077. Maybe it's a little early to draw conclusions, but the game did not crash and I played for about 25 minutes, which was mind mindbogglingly long since the game usually crashes after a few minutes. However, the game did seem like it WANTED to crash at a few point, I'm sure there is a term for what I was experiencing, but the best I could describe it was at a few points with a bit more going on, the audio went a little funky (if you've had your game or your system crash you know what I'm talking about). It didn't crash once, but I think I should do more testing before concluding that it is the graphics card's fault? Or is that enough data to base off a solution off of?

[not so Important]

Also, one thing still doesn't make sense to me. Why do my games crash when I try to play them but the benchmarks I have ran worked completely fine? I feel like I need some sort of an answer cuz the place I got the graphics card from, only lets you run benchmarks and not games, so if I want at least a chance to either get my card replaced or get my money back, I will have to have some sort of explanation for that

[Tetras]

8 minutes ago, not so Important said:

Also, one thing still doesn't make sense to me. Why do my games crash when I try to play them but the benchmarks I have ran worked completely fine? I feel like I need some sort of an answer cuz the place I got the graphics card from, only lets you run benchmarks and not games, so if I want at least a chance to either get my card replaced or get my money back, I will have to have some sort of explanation for that

It might be the benchmark just isn't demanding enough, and/or is too old (e.g. I believe Heaven is over a decade old). Cyberpunk is one of the most demanding games available right now.

 

1 hour ago, not so Important said:

UPDATE: during my lunch break, I underclocked the card by turning the power limit from 100% to 50%, and I also took of 502MHz from both the core and the memory clocks and ran Cyberpunk 2077. Maybe it's a little early to draw conclusions, but the game did not crash and I played for about 25 minutes, which was mind mindbogglingly long since the game usually crashes after a few minutes. However, the game did seem like it WANTED to crash at a few point, I'm sure there is a term for what I was experiencing, but the best I could describe it was at a few points with a bit more going on, the audio went a little funky (if you've had your game or your system crash you know what I'm talking about).

Interesting, it does seem like you may have a dud.

[not so Important]

36 minutes ago, Tetras said:

It might be the benchmark just isn't demanding enough, and/or is too old (e.g. I believe Heaven is over a decade old). Cyberpunk is one of the most demanding games available right now.

 

Interesting, it does seem like you may have a dud.

Alright, thank you! I will post another update here tomorrow on whether or not I was able to get either a replacement card, or at the very least a refund. Hoping for the best

[not so Important]

Quick update: before going to bed, I decided to continue doing some testing. I have the same underclock and in game graphical settings, and the game still crashes. Does that mean that it is still the graphics card's fault?

[Sawa Takahashi]

6 hours ago, not so Important said:

Quick update: before going to bed, I decided to continue doing some testing. I have the same underclock and in game graphical settings, and the game still crashes. Does that mean that it is still the graphics card's fault?

Unfortunately, there is little you can do to prove that it is the graphics card's fault. You can only rule out all other components until only the graphics card is left.

As for underclocking, 50% power limit is way to aggressive. At that point, you should buy a lower tier card. Try around 85% power limit and leave core and memory at base level. If it crashes at that point, the card (or other things in your system) is not usable for its intended purpose.

Do you have the choice the benchmark ? Furmark is a good one for stress rendering. Doom Eternal RT is one other good test for stability.

[Mark Kaine]

its the psu, had the same one , unable to power my 3070 properly (its a shit psu, the "tier list" be damned lol)

 

get a corsair rmx or superflower leadex III,  650w minimum! 

[Mark Kaine]

17 hours ago, not so Important said:

The reseller I bought it from said they can change the card to a different one (same model) if I come across any issues, is it worth bothering with that model of the card? Or should I just request a refund?

ok, its worth trying since its a used cards, but as said above its likely the psu....

 

when i switched out mine for a corsair rmi 650 any and all crashes stopped immediately, is all i can tell you!

[Mark Kaine]

1 hour ago, Sawa Takahashi said:

Furmark is a good one for stress rendering.

no, it isn't.  all nvidia and amd gpus reduce power when they detect fumark running,  its an entirely and completely useless test therefore  

 

 

personally there are two good stress tests (well 3 actually) 

 

Firestrike 

 

Rise Of The Tomb Raider (but not their benchmark,  game needs to be played normally for about 5 minutes) 

 

Prime 95 max settings ~

[FPS_Lego]

10 hours ago, not so Important said:

Quick update: before going to bed, I decided to continue doing some testing. I have the same underclock and in game graphical settings, and the game still crashes. Does that mean that it is still the graphics card's fault?

If you have another system by chance, use the gpu there and see if it crashes. But overall it looks like a defective gpu, i dont  recommend underclocking as you will lose some performance (not saying you should overclock either). But not to face any issues in the future, get the card replaced.

If you can also get a more powerful PSU like a 750w psu, that will also be good. But again its not necessary and that can be done in the future.

[Mark Kaine]

3 hours ago, FPS_Lego said:

But not to face any issues in the future, get the card replaced.

If you can also get a more powerful PSU like a 750w psu, that will also be good. But again its not necessary and that can be done in the future.

except a lot of psus can't handle 3000 series cards. this hasn't anything to do with "not enough power" but faulty / inadequate protections (ie transient power spike handling)

 

 

op needs to get a better psu before anything else therefore,  as i already mentioned. 

 

[Tetras]

1 hour ago, Mark Kaine said:

except a lot of psus can't handle 3000 series cards. this hasn't anything to do with "not enough power" but faulty / inadequate protections (ie transient power spike handling)

 

op needs to get a better psu before anything else therefore,  as i already mentioned. 

The Be quiet 12 M is a very recently released ATX 3.0/PCIE5 PSU, which is supposed to be designed to deal with the spikes.

 

Was your PSU definitely the same model as what the OP has?