Almost all games crashing with new RTX 3070

[Figloalds]

I don't usually have hardware/driver issue, but this one is bugging me a lot and I couldn't get help from NVIDIA forums at all.
I bought this graphics card recently, previously I had a GTX 1060 installed, so I just opened my PC, replaced the card and went to Windows. Then I got to play for about 30~40 minutes total doing tests on various games and it was good, the graphics card itself wasn't stressing even 50~80% with the settings I was using, the CPU by the other hand was stressed, mostly I tested games on 1080p and 1440p (thus realizing that 1440p will be sweetspot for my build). Then I updated the drivers at GeForce experience to 461.09, restarted the PC a few times and now I get instant crash in almost all games, but consistently at specific moments:

- Horizon Zero Dawn: Menu works fine, game loads, in the first 10 frames of being in gameplay it crashes. Previously it was running fine 60 fps locked under vsync, everything max and 1440p.

- Quake II RTX: Sometimes it gives an error before the menu, when it doesn't then when opening the game past the initial cutscene there are black square artifacts and the game crashes 10~20 frames later

- Genshin Impact: The game goes through the logos and warning screens and crashes right at the first frame of the login screen.

- Astroneer: Seems to work fine, no crashes at all. Haven't tested to run it longer than 2 or 3 minutes thoguh 
- Black Desert Online, same as Astroneer

- Control: Same as Horizon Zero Dawn: Menu works fine, crashes at first few frames of gameplay
- Red Dead Online (Vulkan Renderer) Same as Horizon Zero Dawn too.
- Satisfactory: The crash happens directly at the menu screen
I tried rolling back driver on devmgmt.msc, uninstalling the driver with DDU on Safe Mode and then reinstalling fresh downloaded from both NVIDIA and Gainward websites, clean install even, I made double sure I'm installing 64-bit, I tried underclocking 500mhz on MSI Afterburner, i tried disabling vsync on nvidia control panel and I tried restoring defaults, the results are consistently the same.
Temps are consistently 44~49 C, it doesn't start enough work to heat above that
I tried changing TdrLevel config on Regedit, the only change is that game freezes and remains frozen. 

From game logs I could find it seems like the device "turns off" temporarily, but my psu is 850w gold, and it doesn't work even if I use the oveeclocking tools to reduce clocks/power limits
What type of problem can this possibly be?

 

Edit: I did a new test setting H:ZD graphics to lowest configurations and then it doesn't crash immediately, the game runs fine as expected up until the GPU hits 60C, at this point the game crashes, is there any config I can set that "crashing point" to 80 or something, is this a thing?

[Fasauceome]

2 minutes ago, Figloalds said:

my psu is 850w gold,

do you know the model?

[Figloalds]

22 minutes ago, Fasauceome said:

do you know the model?

Yes, it's Corsair RM850X, modular, there's a cable that's 8-pin on PSU side, (6+2) x2 on the other side, that's how the GPU is being powered.
Edit: I added a second of these 8-pin "Type 4" to PCI-e 6+2 x2 cables, connecting each 6+2 of the GPU to a different 8-pin "Type 4" of the PSU, nothing changed either

[OwayBaway]

16 minutes ago, Figloalds said:

I don't usually have hardware/driver issue, but this one is bugging me a lot and I couldn't get help from NVIDIA forums at all.
I bought this graphics card recently, previously I had a GTX 1060 installed, so I just opened my PC, replaced the card and went to Windows. Then I got to play for about 30~40 minutes total doing tests on various games and it was good, the graphics card itself wasn't stressing even 50~80% with the settings I was using, the CPU by the other hand was stressed, mostly I tested games on 1080p and 1440p (thus realizing that 1440p will be sweetspot for my build). Then I updated the drivers at GeForce experience to 461.09, restarted the PC a few times and now I get instant crash in almost all games, but consistently at specific moments:

- Horizon Zero Dawn: Menu works fine, game loads, in the first 10 frames of being in gameplay it crashes. Previously it was running fine 60 fps locked under vsync, everything max and 1440p.

- Quake II RTX: Sometimes it gives an error before the menu, when it doesn't then when opening the game past the initial cutscene there are black square artifacts and the game crashes 10~20 frames later

- Genshin Impact: The game goes through the logos and warning screens and crashes right at the first frame of the login screen.

- Astroneer: Seems to work fine, no crashes at all. Haven't tested to run it longer than 2 or 3 minutes thoguh 
- Black Desert Online, same as Astroneer

- Control: Same as Horizon Zero Dawn: Menu works fine, crashes at first few frames of gameplay
- Red Dead Online (Vulkan Renderer) Same as Horizon Zero Dawn too.
- Satisfactory: The crash happens directly at the menu screen
I tried rolling back driver on devmgmt.msc, uninstalling the driver with DDU on Safe Mode and then reinstalling fresh downloaded from both NVIDIA and Gainward websites, clean install even, I made double sure I'm installing 64-bit, I tried underclocking 500mhz on MSI Afterburner, i tried disabling vsync on nvidia control panel and I tried restoring defaults, the results are consistently the same.
Temps are consistently 44~49 C, it doesn't start enough work to heat above that
I tried changing TdrLevel config on Regedit, the only change is that game freezes and remains frozen. 

From game logs I could find it seems like the device "turns off" temporarily, but my psu is 850w gold, and it doesn't work even if I use the oveeclocking tools to reduce clocks/power limits
What type of problem can this possibly be?

you might just have to wait for a driver update, is ur gpu running hot?

[Figloalds]

Just now, OwayBaway said:

you might just have to wait for a driver update, is ur gpu running hot?

No, the games are crashing before the GPU can even get to 50 C. The only games that are not crashing right away so far are Astroneer and Black Desert Online, other varied games running on DX 11, DX 12, Vulkan are all crashing consistently in less than 20'ish frames of gameplay

[Mister Woof]

Is your CPU overclocked?

[Figloalds]

 

4 minutes ago, Mister Woof said:

Is your CPU overclocked?

No, all bios settings are set to defaults, I don't have the smarts or courage to fiddle with overclocking

[Mister Woof]

3 minutes ago, Figloalds said:

 

No, all bios settings are set to defaults, I don't have the smarts or courage to fiddle with overclocking

1. I know it's in the same ecosystem, but did you DDU before install?

2. If the 3070 has dual 8-pin (or 8 pin + 6 pin), are you using one PCIE cable from the PSU and using the daisy chained connectors, or two separate PCIE cables from the PSU?

[Figloalds]

8 minutes ago, Mister Woof said:

1. I know it's in the same ecosystem, but did you DDU before install?

2. If the 3070 has dual 8-pin (or 8 pin + 6 pin), are you using one PCIE cable from the PSU and using the daisy chained connectors, or two separate PCIE cables from the PSU?

1. Yes.
2. Yes I was using 1 8-pin from PSU to double 6+2 PCIe on the GPU, i added a second identical cable, now I have 2 8-pin to 8-pin cables going from PSU to the graphics card, and nothing changed

[LilPach]

3 hours ago, Figloalds said:

I don't usually have hardware/driver issue, but this one is bugging me a lot and I couldn't get help from NVIDIA forums at all.
I bought this graphics card recently, previously I had a GTX 1060 installed, so I just opened my PC, replaced the card and went to Windows. Then I got to play for about 30~40 minutes total doing tests on various games and it was good, the graphics card itself wasn't stressing even 50~80% with the settings I was using, the CPU by the other hand was stressed, mostly I tested games on 1080p and 1440p (thus realizing that 1440p will be sweetspot for my build). Then I updated the drivers at GeForce experience to 461.09, restarted the PC a few times and now I get instant crash in almost all games, but consistently at specific moments:

- Horizon Zero Dawn: Menu works fine, game loads, in the first 10 frames of being in gameplay it crashes. Previously it was running fine 60 fps locked under vsync, everything max and 1440p.

- Quake II RTX: Sometimes it gives an error before the menu, when it doesn't then when opening the game past the initial cutscene there are black square artifacts and the game crashes 10~20 frames later

- Genshin Impact: The game goes through the logos and warning screens and crashes right at the first frame of the login screen.

- Astroneer: Seems to work fine, no crashes at all. Haven't tested to run it longer than 2 or 3 minutes thoguh 
- Black Desert Online, same as Astroneer

- Control: Same as Horizon Zero Dawn: Menu works fine, crashes at first few frames of gameplay
- Red Dead Online (Vulkan Renderer) Same as Horizon Zero Dawn too.
- Satisfactory: The crash happens directly at the menu screen
I tried rolling back driver on devmgmt.msc, uninstalling the driver with DDU on Safe Mode and then reinstalling fresh downloaded from both NVIDIA and Gainward websites, clean install even, I made double sure I'm installing 64-bit, I tried underclocking 500mhz on MSI Afterburner, i tried disabling vsync on nvidia control panel and I tried restoring defaults, the results are consistently the same.
Temps are consistently 44~49 C, it doesn't start enough work to heat above that
I tried changing TdrLevel config on Regedit, the only change is that game freezes and remains frozen. 

From game logs I could find it seems like the device "turns off" temporarily, but my psu is 850w gold, and it doesn't work even if I use the oveeclocking tools to reduce clocks/power limits
What type of problem can this possibly be?

 

Edit: I did a new test setting H:ZD graphics to lowest configurations and then it doesn't crash immediately, the game runs fine as expected up until the GPU hits 60C, at this point the game crashes, is there any config I can set that "crashing point" to 80 or something, is this a thing?

I had some issues when I first got my 3070 also and it seems to be a driver issue. To fix this issue I downloaded the newest studio driver rather than the game ready one and for the most part the issues are gone. I would recommend giving it a try and using that until the next game ready driver is ready. If this doesn’t work I would try and rma the gpu since you may have gotten a defective one.

[shaz2sxy]

are you running a riser cable?

 

[Figloalds]

2 minutes ago, shaz2sxy said:

are you running a riser cable?

 

It's connected directly to the motherboard PCI-e slot. The slot is a PCI-e 3.0 though, I did read prior to purchasing the card that it would work

[shaz2sxy]

Just now, Figloalds said:

It's connected directly to the motherboard PCI-e slot. The slot is a PCI-e 3.0 though, I did read prior to purchasing the card that it would work

only asking as i had issues and it was my PCIE Riser cable as my CPU and GPU are gen 4 but the riser was gen 3

[Figloalds]

Ok, I did something that changed things.
I went to powercfg.cpl and configured CPU min and max performance to 5%, with that I was able to run HZD significantly longer than just reducing it's graphics, artifacts were also significantly reduced or absent; I was also able to get past Login screen in Genshin Impact, seen artifacts tho. Quake RTX still crashes at the same speed and gives the same artifacts. 
I got overall much higer GPU usage, temperature, and total frames rendered before crash than previously, with this I come to the possibility that the GPU is fine (which is my biggest fear honestly, don't want the hassle of sending it back and waiting for another one), but something related to CPU clock, voltage, temps or PSU power might be causing the driver to crash. The GPU itself is stock, brand new, default settings, default everything, running on PCI-e 3.0, the whole setup worked fine as expected for a while (it was nice to run HZD all maxed on 1440p) and only some time later I started having issues. I'll try to reapply thermal paste on processor and see if anything changes. I strongly doubt it's the PSU itself either, she's a little beast.