Question Switched from AMD to Nvidia; GTX 1070 unable to handle heavy load (crashes, BSODs)

[frisbii]

computer specs​

• R5 3600​

• MSI GTX 1070 Gaming (formerly a Powercolor RX 570 Red Dragon)
• Teamgroup 16gb (2x8gb) 3000mhz cl16
• EVGA 650 BQ
• Asrock B450 Steel Legend

the issue!​

hi all! I bought an MSI GTX 1070 Gaming from a friend recently as he upgraded and didn't have a use for the old (quite used) card. I switched out my years old RX 570 for the new card today and everything seemed to work, but when being used in heavy tasks like games or stress tests, the program either crashes or windows BSODs after 30 to 60 seconds of running. I have tried to run:

• Valorant (crash)
• F1 22 (crash)
• FIFA 23 (BSOD)
• 3DMark's Time Spy (crash, DXGI call IDXGISwapChain:😛resent failed [0X887A0005])
• Unigine Superposition (crash)

In all of these cases, the game or program ran for a little bit at seemingly normal performance before stopping. There were no visual or graphical errors except in Unigine, where there were obvious visual glitches before the program crashed. According to Unigine's stats, the card ran at ~100fps, ~70degC. Using hwmon during Time Spy showed about the same, with ~150W power draw.

troubleshooting steps taken​

After installing the card:

• I uninstalled AMD Adrenalin from Control Panel and installed Nvidia's latest drivers. I tried to get into some games, but all ended up crashing after a bit as described.
• After looking into it, I used DDU to uninstall all AMD and Nvidia drivers, then reinstalled Nvidia drivers, to no avail.
• I then took the card out and dusted it off as there was a lot of buildup and temperatures were getting high during the brief periods of load that were sustained- still no change.
• I used DDU again and installed Nvidia's studio drivers with no luck.
• I then used Windows recovery to reinstall windows while keeping my data, then reinstalled AMD chipset drivers and Nvidia drivers, but the same result even after resetting Windows.

I'm not sure what to do next besides a fully clean install of Windows on my drive, which I would rather not have to go through again.

Any suggestions would be greatly appreciated. I do believe that the card worked fine for a long time for my friend, so I am hoping it is not a hardware fault.

 

[thestryker]

The board power for both are pretty close to the same, but I believe the 570 uses a 6 pin and the 1070 an 8 pin. If this is the case for your cards in question it's possible there's a cable issue. If your PSU has another PCIe power cable you can use try that first. Generally speaking if a GPU was working fine in another machine, but is crashing under load in a new one power delivery is where I look.

 

[Deleted member 2731765]

It might also be possible a faulty GPU. But like mentioned above, first try using a different PCIe cable. Was the card 100% working on your friend's PC under heavy load/stress/gaming ? Maybe try using a better PSU model ?

Which PSU was your friend using ? This also sounds more like a PSU issue to me, since the card only crashes under load/stress, IMO.

 

[beyondlogic]

computer specs​

• R5 3600​

• MSI GTX 1070 Gaming (formerly a Powercolor RX 570 Red Dragon)
• Teamgroup 16gb (2x8gb) 3000mhz cl16
• EVGA 650 BQ
• Asrock B450 Steel Legend

the issue!​

hi all! I bought an MSI GTX 1070 Gaming from a friend recently as he upgraded and didn't have a use for the old (quite used) card. I switched out my years old RX 570 for the new card today and everything seemed to work, but when being used in heavy tasks like games or stress tests, the program either crashes or windows BSODs after 30 to 60 seconds of running. I have tried to run:

• Valorant (crash)
• F1 22 (crash)
• FIFA 23 (BSOD)
• 3DMark's Time Spy (crash, DXGI call IDXGISwapChain:😛resent failed [0X887A0005])
• Unigine Superposition (crash)

In all of these cases, the game or program ran for a little bit at seemingly normal performance before stopping. There were no visual or graphical errors except in Unigine, where there were obvious visual glitches before the program crashed. According to Unigine's stats, the card ran at ~100fps, ~70degC. Using hwmon during Time Spy showed about the same, with ~150W power draw.

troubleshooting steps taken​

After installing the card:

• I uninstalled AMD Adrenalin from Control Panel and installed Nvidia's latest drivers. I tried to get into some games, but all ended up crashing after a bit as described.
• After looking into it, I used DDU to uninstall all AMD and Nvidia drivers, then reinstalled Nvidia drivers, to no avail.
• I then took the card out and dusted it off as there was a lot of buildup and temperatures were getting high during the brief periods of load that were sustained- still no change.
• I used DDU again and installed Nvidia's studio drivers with no luck.
• I then used Windows recovery to reinstall windows while keeping my data, then reinstalled AMD chipset drivers and Nvidia drivers, but the same result even after resetting Windows.

I'm not sure what to do next besides a fully clean install of Windows on my drive, which I would rather not have to go through again.

Any suggestions would be greatly appreciated. I do believe that the card worked fine for a long time for my friend, so I am hoping it is not a hardware fault.

if memory serves correctly make sure both the 8 pin and 6 pin pci e connectors are connected make sure there properly seated


as that psu in question has 2 pci connectors both 8 pins that are 6 +2 make sure that the pin that makes up 8 pin is seated properly those have a bad habit of not being 100 percent in correctly.

also reseat the card.

lastly one other thing you can try is running it with your bios in default

basicly if you have any overclocks in your system. go into your bios and click the default to put it to normal. see if the problem persists.

also the rx 570 is a much weaker card then the 1070 you may have stability issues cause its pushing your pc harder and making it more unstable.

also make sure your bios in your motherboard is P2.30 or higher cause if your running a bios under p2.30 it may be the motherboard is unstable.

 

[frisbii]

Thanks all for your responses.

• The 570 used 8pin power and the 1070 now uses 8pin and an extra 6pin for power.
• I used both the hard-wired 8+6pin cable and the modular 8+6pin cable that came with the PSU, no change.
• According to him, the card was running fine at high loads about 6 months ago. He used an 800W PSU, which is more than mine, but 650W should still be enough for a 1070 and R5 3600. Unfortunately, I don't have a higher rated PSU on hand to test with.
• I've reseated everything multiple times but no change.
• I've reset BIOS to defaults, no change. My BIOS version is 4.60.

I'm thinking the card is just dead at this point, which is unfortunate since that RX 570 really shows its age these days. I'll do a clean Windows install before I give up though.

 

[beyondlogic]

Thanks all for your responses.

• The 570 used 8pin power and the 1070 now uses 8pin and an extra 6pin for power.
• I used both the hard-wired 8+6pin cable and the modular 8+6pin cable that came with the PSU, no change.
• According to him, the card was running fine at high loads about 6 months ago. He used an 800W PSU, which is more than mine, but 650W should still be enough for a 1070 and R5 3600. Unfortunately, I don't have a higher rated PSU on hand to test with.
• I've reseated everything multiple times but no change.
• I've reset BIOS to defaults, no change. My BIOS version is 4.60.

I'm thinking the card is just dead at this point, which is unfortunate since that RX 570 really shows its age these days. I'll do a clean Windows install before I give up thou

Thanks all for your responses.

• The 570 used 8pin power and the 1070 now uses 8pin and an extra 6pin for power.
• I used both the hard-wired 8+6pin cable and the modular 8+6pin cable that came with the PSU, no change.
• According to him, the card was running fine at high loads about 6 months ago. He used an 800W PSU, which is more than mine, but 650W should still be enough for a 1070 and R5 3600. Unfortunately, I don't have a higher rated PSU on hand to test with.
• I've reseated everything multiple times but no change.
• I've reset BIOS to defaults, no change. My BIOS version is 4.60.

I'm thinking the card is just dead at this point, which is unfortunate since that RX 570 really shows its age these days. I'll do a clean Windows install before I give up though.

one other thing to check on the card is see if hes opened it at some point if the sticker over the screws on the back of the card is missing he may have tampered with it. if so the 70c may be fine on the gpu core but if the memory chips arent making proper contact with there thermal pads that would cause instability.

that psu is fine you may wanna try a diffrent port on the psu since the cable is modular i think this model has 2 places for a gpu to be hooked up to the psu maybe issues with that port. or could be a bad cable.

 

[sitehostplus]

Thanks all for your responses.

• The 570 used 8pin power and the 1070 now uses 8pin and an extra 6pin for power.
• I used both the hard-wired 8+6pin cable and the modular 8+6pin cable that came with the PSU, no change.
• According to him, the card was running fine at high loads about 6 months ago. He used an 800W PSU, which is more than mine, but 650W should still be enough for a 1070 and R5 3600. Unfortunately, I don't have a higher rated PSU on hand to test with.
• I've reseated everything multiple times but no change.
• I've reset BIOS to defaults, no change. My BIOS version is 4.60.

I'm thinking the card is just dead at this point, which is unfortunate since that RX 570 really shows its age these days. I'll do a clean Windows install before I give up though.

Don't give up yet.

There is a utility on the market called 'who crashed' that can help. It reads the widows dump logs and translates them into plain english (with hyperlinks so you can do your own research). It's the first tool I call on when faced with a BSOD anymore.

Get it here.

 

[frisbii]

Final update on what ended up happening:

I ended up replacing my PSU with an RMA in warranty, citing both the weird noise it made and this issue. New one comes in, no fix. I don't know why I didn't think to do this earlier, but I decided to install MSI Afterburner and messed around with fan curve, clocks, and voltages. In the end, an underclock of -200 core, -200 vram led to stable operation and it's held up since. The card definitely has some sort of issue, but is still far faster than my old card even with the underclock. The heavy artifacting that was seen under some crashes makes me think it's dying VRAM, but I'm not sure. I'm going to roll with my current underclock and see how long it lasts; I expect that that'll need to be underclocked more as time goes on? I'm not sure.