New pc crashes under load.

[DG458]

My new pc keeps crashing when playing games. The build itself went fine and I can use the pc normally, but whenever I play games it'll crash to desktop, or even crash the entire system. The games that crash to desktop are fallout 4, no Man's Sky, overwatch, and Diablo 3. There are no error messages the game just closes. I've tried playing Cyberpunk as well and that will completely freeze and crash the entire thing. Ive tried running benchmarks and stress tests in 3Dmark, and only maybe 1 in 5 attemps succeed, it usually just closes and says "an error occured". Ive installed and reinstalled the GPU drivers through the geforce experience software. I've updated my motherboards bios to the latest version. I've ran MSI afterburner and watched temps and usage, the temps all stay at a reasonable level for the hardware. All stock no overclocks, ram running at docp, issue persists if i turn docp off. Build specs are as follows: 

Operating system: Windows pro 64bit

Motherboard: Asus Strix X 570I mini itx (latest bios, issue persists pre-updated)

Cpu: Ryzen 9 5900x 

Cpu cooler: Phanteks glacier one 240

Ram: 32gb Corsair Vengeance pro 3600mhz 

Storage: 2 Samsung Pro Nvme M.2 2tb each

GPU: Nvidia 3090 TI FE 

PSU: Evga SuperNOVA 850w 80+ gold

Case: Phanteks Evolv Shift XT

 

[mehbodo]

Are you running the GPU from two seperace PCIe power cables, or two plugs from the same cable line. If it's the latter, then you need to run the second cable

[DG458]

53 minutes ago, mehbodo said:

Are you running the GPU from two seperace PCIe power cables, or two plugs from the same cable line. If it's the latter, then you need to run the second cable

It would be from 2 seperate cables. The gpu has the 12pin converting to 3 8s. My psu only has slots for 2 gpu cables so 1 8pin has its own and 2 have to share the other. This was the biggest psu I could find that would fit in a mini itx case. 

[BlackManINC]

3 hours ago, DG458 said:

My new pc keeps crashing when playing games. The build itself went fine and I can use the pc normally, but whenever I play games it'll crash to desktop, or even crash the entire system. The games that crash to desktop are fallout 4, no Man's Sky, overwatch, and Diablo 3. There are no error messages the game just closes. I've tried playing Cyberpunk as well and that will completely freeze and crash the entire thing. Ive tried running benchmarks and stress tests in 3Dmark, and only maybe 1 in 5 attemps succeed, it usually just closes and says "an error occured". Ive installed and reinstalled the GPU drivers through the geforce experience software. I've updated my motherboards bios to the latest version. I've ran MSI afterburner and watched temps and usage, the temps all stay at a reasonable level for the hardware. All stock no overclocks, ram running at docp, issue persists if i turn docp off. Build specs are as follows: 

Operating system: Windows pro 64bit

Motherboard: Asus Strix X 570I mini itx (latest bios, issue persists pre-updated)

Cpu: Ryzen 9 5900x 

Cpu cooler: Phanteks glacier one 240

Ram: 32gb Corsair Vengeance pro 3600mhz 

Storage: 2 Samsung Pro Nvme M.2 2tb each

GPU: Nvidia 3090 TI FE 

PSU: Evga SuperNOVA 850w 80+ gold

Case: Phanteks Evolv Shift XT

 

You have a spare CPU to try? GPU? 

[DG458]

7 hours ago, BlackManINC said:

You have a spare CPU to try? GPU? 

I have another pc that is a 3950x with a 2080, 1000w psu. The whole things is liquid cooled though so I was hoping it would be a last resort. Didn't want to have to drain the loop to start pulling parts unless absolutely necessary. 

[BlackManINC]

5 hours ago, DG458 said:

I have another pc that is a 3950x with a 2080, 1000w psu. The whole things is liquid cooled though so I was hoping it would be a last resort. Didn't want to have to drain the loop to start pulling parts unless absolutely necessary. 

Well, spare parts is always the best way to rule out exactly what is causing these types of issues. I'd start with the CPU since its a full system crash. 

[DG458]

I pulled the 2080 hydro out of my other pc and put the stock air cooler on it and put it into the pc with the 5900x. Ran the speed way stress test and it passed on first try. Something the 3090 ti had yet to complete once in its 10+attemps. Pulled the 2080 out and put the 3090 ti back in, than pulled the 1000w psu out of the other pc. The 1000w has 6 vga ports so all 3 of the 8 pins for the 3090 where able to get their own cable. 

[DG458]

So I downloaded a few more tools instead of just afterburner. I don't know if you can see gpu vram temp or not but I couldn't find it so I got TechPowerUp GPU-Z. How hot is safe ram temps and can it cause the instability issues that I'm having? Vram is getting as hot as 112C and "Hot spot" on gpu seeing as high as 98C with the overall gpu at 89C under full load stress tests. I'm also worried that because of the way this case is configured that it's baking the ddr4 system ram directly on the other side of the small wall that seperates the two. Can hot system ram cause random errors as well? 

[BlackManINC]

20 hours ago, DG458 said:

So I downloaded a few more tools instead of just afterburner. I don't know if you can see gpu vram temp or not but I couldn't find it so I got TechPowerUp GPU-Z. How hot is safe ram temps and can it cause the instability issues that I'm having? Vram is getting as hot as 112C and "Hot spot" on gpu seeing as high as 98C with the overall gpu at 89C under full load stress tests. I'm also worried that because of the way this case is configured that it's baking the ddr4 system ram directly on the other side of the small wall that seperates the two. Can hot system ram cause random errors as well? 

According to some nerds on the Linus discord, yes, 112c is insanely hot. They said something about replacing the thermal pads, but I have no idea if a GPU even has something like that. Not sure what they even mean.

[Cyberat]

90 C is insanely hot. First step is to ramp up 100% Fans at 60 C in drivers.