RTX3070: Black Screen + 100% GPU Fans

[SmoothOperat0r]

I'm getting this random crash while playing some games, happens at least once a day, where I lose video signal (black screen), my GPU fans go 100% speed and I need to do a hard restart. After doing some research, it sounds like this is a display driver crash. My thermals are good and specs listed below. Has anyone found a fix for this?

 

Specs:

CPU: R5 3600

CPU Cooler: AMD Wrath Prism

GPU: RTX3070 (EVGA FTW3 ULTRA)

Display: LG CX OLED (G-Sync Compatible)

RAM: 2x8GB 3600MHZ (Corsair Vengeance RGB Pro)

MOBO: X570 (ASUS Tuf Gaming X570-Plus (Wifi))

SSD: 1TB M.2 (Intel P665) 

PSU: 750W Gold (EVGA Supernova G3)

Case: Phanteks P400A Mesh

Case Fans: 6x Corsair 120QL

Windows: Latest Rev of 10

Bios: Latest Rev (Feb 2021)

 

Any help is appreciated, 

 

Thank You

[SmoothOperat0r]

Still no luck finding a solution, seems like there are a few people with the same problem.

 

Could it be an outdated BIOS issue? My current version is from August 2020, it just occurred to me that the RTX3070 came out after the revision date on my BIOS, could this be the issue and should I update my BIOS regardless?

[Heliian]

It's a good start to update the bios in this case. 

 

Make sure you have the proper power cables for the 3070, I've heard splitters can cause problems on the power hungry versions.

 

Are your CPU thermals good? 

[c00face]

On 2/28/2021 at 10:22 AM, SmoothOperat0r said:

I'm getting this random crash while playing some games, happens at least once a day, where I lose video signal (black screen), my GPU fans go 100% speed and I need to do a hard restart. After doing some research, it sounds like this is a display driver crash. My thermals are good and specs listed below. Has anyone found a fix for this?

 

Specs:

CPU: R5 3600

CPU Cooler: AMD Wrath Prism

GPU: RTX3070 (EVGA FTW3 ULTRA)

Display: LG CX OLED (G-Sync Compatible)

RAM: 2x8GB 3600MHZ (Corsair Vengeance RGB Pro)

MOBO: X570 (ASUS Tuf Gaming X570-Plus (Wifi))

SSD: 1TB M.2 (Intel P665) 

PSU: 750W Gold (EVGA Supernova G3)

Case: Phanteks P400A Mesh

Case Fans: 3x Corsair 120QL

Windows: Latest Rev of 10

Bios: Could update bios still (last updated in August)

 

Any help is appreciated, 

 

Thank You

3600mhz ram is considered overclocked. Another member had the same issue as you and it was wifi driver incompatibility, ram speed, and not enough power to draw on for graphics card.

 

follow instructions on your motherboard how to install a 3600mhz ram or adjust the speed to lower setting in bios. disable / uninstall or update the WiFi driver, and go to your pc power settings and switch from balanced to performance.

[Mark Kaine]

Also funny enough some EVGA PSUs have issues with EVGA 30xx GPUs, you may want to check if you have one of the affected ones, also make sure to use the proper cables, no splitters or daisy chaining, each pcie port needs its own pcie cable. 

 

[SmoothOperat0r]

21 hours ago, Heliian said:

Make sure you have the proper power cables for the 3070, I've heard splitters can cause problems on the power hungry versions.

 

Are your CPU thermals good? 

Yes, dedicated single jacks (no daisy chain 2nd jack).

 

CPU thermals (and all thermals) are good. 

Idle Temperatures:

CPU: 40c

GPU: 31c

RAM: 34c

 

Heaven (1440P Ultra Settings) Temps:

CPU: 55-60c

GPU: 55-60c

RAM: 40c

 

Cinebench R23 Multicore Temps:

CPU: 75c

GPU: 35c

RAM: 38c

[SmoothOperat0r]

20 hours ago, c00face said:

3600mhz ram is considered overclocked. Another member had the same issue as you and it was wifi driver incompatibility, ram speed, and not enough power to draw on for graphics card.

 

RAM:

Its actually a CL18 4000MHZ kit which is listed on the QVL. I have it running at CL16 3600MHZ with custom timings derived from DRAM Calculator. I have a set of Gskill CL16 3200MHZ which is also listed on the QVL I can swap in for testing.

 

Wifi Driver:

Good suggestion, I will try a clean install for the driver. 

 

PSU:

This was my first thought, but I've already swapped PSUs, originally had the Gigabyte P750GM, but after reading about all the DOAs I decided to replace it with the EVGA PSU. While I'm much happier with the electrical performance, more stable voltage rails and higher wattage delivered, this did not solve the problem.

 

Going to start with the Wifi driver, if that doesn't work I'll swap in the gskill ram.

 

Thanks for the feedback, hope this works.

 

[SmoothOperat0r]

19 hours ago, Mark Kaine said:

Also funny enough some EVGA PSUs have issues with EVGA 30xx GPUs, you may want to check if you have one of the affected ones, also make sure to use the proper cables, no splitters or daisy chaining, each pcie port needs its own pcie cable. 

Will have to look for this, is it published by EVGA or just reports? Cables are good, 2 dedicated cables (no daisy chain 2nd connections) just straight homeruns to the PSU for each. The PSU has 4 PCIE terminals, I couldn't find any information if the connections need to be staggered for dual power socket cards. I believe I have them plugged into PCIE 1 and 2.

[Mark Kaine]

1 minute ago, SmoothOperat0r said:

Will have to look for this, is it published by EVGA or just reports? Cables are good, 2 dedicated cables (no daisy chain 2nd connections) just straight homeruns to the PSU for each. The PSU has 4 PCIE terminals, I couldn't find any information if the connections need to be staggered for dual power socket cards. I believe I have them plugged into PCIE 1 and 2.

yeah they've been exchanging affected PSUs afaik, but I don't remember the models affected. 

 

 

[SmoothOperat0r]

16 minutes ago, SmoothOperat0r said:

Wifi Driver:

Good suggestion, I will try a clean install for the driver. 

Just did a clean install of the latest driver from Intel's website. Sounds like this could be the issue, I remember having to reset the wifi connection following reboot after each crash. I just figured it was from improper shutdown, but that make sense if its a driver issue causing this.

 

Loading up Valheim, usually crashes in the first 5 minutes and then runs stable for the rest of the night. Hopefully this works.

[SmoothOperat0r]

Unfortunately the wifi driver clean install did not fix the issue.

 

I checked the QVL on the spare RAM kit I was going to swap in and even though it says AMD compatible on the box, its not on the QVL. So I ordered another kit that is. My current installed kit is on the QVL, but maybe running it at a lower frequency and faster timings is an issue, I would run it at the rated 4000mhz, but the fclk is limited to 1800mhz. 

 

New RAM will be here Monday.

 

I have not updated my BIOS yet, so it could still be that.

[0H0]

Hey, (french sry for the mistake)

 

I reply to this thread for people who eventually still have the same issue, with a EVGA 3090 RTX Ultra I got the same problem, I was 100% sure the card was dead because it was running fine for 2 week before it happen.

I was going to RMA but then I notice one of the 8pin cable was a little hot, the PSU is Seasonic 1300W so i didn't bother use a third cable as the old card was using 2x8pin and there is a splitter on each cable. I run multiple benchmark with 2 cable and the splitter when I installed the card it was running fine.

 

So I put a third cable and since then I got no problem I also set the TDP to 119% instead of 107% with the firmware upgrade and it work fine for now, pass all bench again, no fan issue.

I don't guarented the card is fine maybe it still slowly dying( I will update if it does)  but for sure it seems to draw a lot of power.

[SmoothOperat0r]

 

2 hours ago, 0H0 said:

So I put a third cable and since then I got no problem I also set the TDP to 119% instead of 107% with the firmware upgrade and it work fine for now, pass all bench again, no fan issue.

 

I have separate cables for each power input, no daisy chain, One cable is connected to PCIe #1 on the PSU and the other is connected to PCIe #2 on the PSU.

 

An update on the situation.

 

I've swapped RAM, did DDU in safemode and installed latest drivers (microsoft store beat me to it, forgot to disable the wifi driver before rebooting). I also rechecked my event viewer and found that dwm.exe was crashing either right before or during the black screen crash, everytime!!!! I also found a lot of crashes for xbox game bar and had warnings for DCOM permissions. I checked with EVGA, they said to reseat the card, if that doesn't work, try moving it to a second slot if available. 

 

I've reseated the card, but upon doing this I noticed the rear case brackets for the GPU didn't line up perfect with my case (P400A), I tried loosening the mobo screws to get a better fit, but still couldn't make it line up right without pushing on the case. So it could still be a seating issue, even though the retaining clip is fully engaged.

 

Next step would be to move the card to the second slot, and see if that makes a difference. If not I still have the option to RMA the card, but I'm trying to avoid that since the lead time is 6+ weeks and there's no convincing evidence that this is not software related. It could be related to the AMD USB issues which will be getting patched in the next BIOS update. After countless hours of searching the internet for answers, I've found many people with the same issue, and lots of things to try, but no definitive fix.

 

P.S. forgot to mention BIOS was updated to the February 2021 revision, will update again when the AMD USB hotfix is released.

 

[SmoothOperat0r]

On 3/2/2021 at 8:47 PM, c00face said:

and switch from balanced to performance.

Must have missed this, I'm currently running ryzen balanced power plan, I thought high performance was mostly for benchmarking, don't you want the cpu to turndown when not under load?

[SmoothOperat0r]

On 3/16/2021 at 5:56 PM, SmoothOperat0r said:

but still couldn't make it line up right without pushing on the case. So it could still be a seating issue, even though the retaining clip is fully engaged.

THIS ^^^^^^^^

 

The GPU was not seated fully, the two prongs on the bottom of the mounting bracket were getting caught on a metal seam below the motherboard. The prongs were not perfectly straight, slight angle on the tip towards the perimeter (seam) of the case. Not sure if that's by design but it compounded the issue with the seam. I was able to correct this by slightly bending the prongs straight again, If your going to do this, use 2 wrenches, one to bend the tip and one to stop the rest of the bracket from bending. I also had to fix the top bar on the case bracket which had been bent upwards to accommodate the GPU misalignment. 

 

Once I was done, I put the card back in and it seated perfectly without effort. Everything lined up perfect and you could see from behind that both prongs made it past the metal case seam. I tried to reproduce the crash all last night and today, but it seems to be gone now. Normally I get this crash on games like valhiem and stranded deep in the first 15 minutes of gaming each day and sometimes multiple crashes. Now it runs perfect.