Question 90% sure GFX card is faulty but it never fails stress tests

[massop]

Hi,

I have a GTX 3090 and over the last few months it has been randomly crashing in Overwatch and Warcraft.

Initially I sent it back for RMA but it was rejected as it passed stress testing.

When I got it back the problem seemed to have gone so I though it was just a reseating issue but recently it happens everyday and reseating no longer works,

I have done all the stress testing for GPU, CPU & memory for 24 hours and it never fails.

If I play Overwatch it will probably happen 2-3 times a day

The reason I am 90% sure its the GFX card is when I had to use a 2080Ti as a replacement for the RMA (about 2 weeks) there was not a single crash and I was playing the same games

What can I do to get a repeatable crash so that I can get it RMAed?

PC SPECS:
Operating System - Microsoft Windows 10 Pro 64-bit
Processor - RYZEN 3970X 4.50GHZ PROCESSOR (8% Overclock-stable)
Power Supply - EVGA - SuperNOVA 1300W Watt 80 Plus Titanium ATX Power Supply
RAM - G.SKILL Ripjaws V Series 128GB (4 x 32GB) 288-Pin PC RAM DDR4 3200
Motherboard - ASUS - TRX40 ROG Zenith II Extreme Alpha AMD sTRX4 EATX Motherboard
Graphics Card – Zotac gaming 3090 (STOCK CLOCKS)
Solid State Drive - 1TB Gigabyte AORUS M.2 (2280) PCIe 4.0 (x4) NVMe SSD (x2 in raid 0)
Hard Drives - 5TB Western Digital BLACK time x in RAID1

 

[ohio_buckeye]

Don't have a 3090 but a quick search turned up this nvidia forum of people who had similar problems. Look like some had issues with ram and another with an issue with a turbo boost setting in bios. Might check to see what bios version you are using as well.

RTX 3090 Keeps Crashing | NVIDIA GeForce Forums

 

[rickypicky5]

Is it just one game? Assuming the 2080ti and the 3090 require different NVidia drivers, it could be an issue with the 3090 driver you are using and the game you are playing.

 

[Eximo]

Role back to an older driver would be my suggestion here.

Check your GPU for vBIOS updates, unless they did it as part of the RMA.

 

[LORYT699]

Hi,

I have a GTX 3090 and over the last few months it has been randomly crashing in Overwatch and Warcraft.

Initially I sent it back for RMA but it was rejected as it passed stress testing.

When I got it back the problem seemed to have gone so I though it was just a reseating issue but recently it happens everyday and reseating no longer works,

I have done all the stress testing for GPU, CPU & memory for 24 hours and it never fails.

If I play Overwatch it will probably happen 2-3 times a day

The reason I am 90% sure its the GFX card is when I had to use a 2080Ti as a replacement for the RMA (about 2 weeks) there was not a single crash and I was playing the same games

What can I do to get a repeatable crash so that I can get it RMAed?

PC SPECS:
Operating System - Microsoft Windows 10 Pro 64-bit
Processor - RYZEN 3970X 4.50GHZ PROCESSOR (8% Overclock-stable)
Power Supply - EVGA - SuperNOVA 1300W Watt 80 Plus Titanium ATX Power Supply
RAM - G.SKILL Ripjaws V Series 128GB (4 x 32GB) 288-Pin PC RAM DDR4 3200
Motherboard - ASUS - TRX40 ROG Zenith II Extreme Alpha AMD sTRX4 EATX Motherboard
Graphics Card – Zotac gaming 3090 (STOCK CLOCKS)
Solid State Drive - 1TB Gigabyte AORUS M.2 (2280) PCIe 4.0 (x4) NVMe SSD (x2 in raid 0)
Hard Drives - 5TB Western Digital BLACK time x in RAID1

it seems a driver issiue

 

[massop]

Thanks for the input guys,

I have tried the following since my last post

• Restored CPU overclock to stock speed.
• Disabled boost in BIOS.
• Updated Motherboard BIOS.
• Updated chipset drivers.

I still get random crash in Overwatch (now I am officially banned in season 3 ranked matched because of the amount of crashes)

I need to try some other games to see if the issue happens there, do you have any suggestions?

I will try rolling back the video driver to 2020 as I had no issues then and let you know how it goes.

In regards to vBIOS updates, how do I go about doing that?

Thanks

 

[massop]

I have downloaded the March 2021 driver from HERE,

Unfortunately I still got the crash after 90 minutes of gameplay, definitely starting to look like issues with the card.

I just need to find a benchmark or similar test that can get this to crash every time so I can get the card RMA, any suggestions would be much appreciated

 

[LORYT699]

I have downloaded the March 2021 driver from HERE,

Unfortunately I still got the crash after 90 minutes of gameplay, definitely starting to look like issues with the card.

I just need to find a benchmark or similar test that can get this to crash every time so I can get the card RMA, any suggestions would be much appreciated

well, just a stupid question.
Is all the other things updated to the last firmware?like windows for example, cause I remember a few time that my pc crashes without doing nothing and the problem was that not all the hardware/basic software were updated to the last firmware, then also search for similar things in the overwatch forum cause it happen only with that right?

 

[massop]

Ok so there has been a lot of testing in the background on this and I think I found the issue.

Looks like it was not the card at all.

I am running the OS on 2 NVME drives in raid 0, they have big heatsinks on them because they run hot.

One of these drives sit directly under the 3090 and is even in physical contact with the card. I think what was happening was the card was getting hot during gaming and was heating up that SSD so it was causing it to fail thus crashing the system because its an OS drive.

I moved the OS to new SATA SSDs far away from the card and I have not had any crashing so far.

The reason I was unable to see this sooner was because I cannot see the NVME S.M.A.R.T information (temps etc) when its in raid so I had no idea it was getting too hot.

Ill keep testing and update if anything changes but I think this was the issue all along.

Thanks again for your help