Question GPU failure - black screen, fans 100% ?

[yliats]

Hi,

My GPU is failing under load. When I launch a game or any other GPU-intensive program after a few seconds my monitor turns black and GPU fans turn 100%. The rest of the PC works fine since my friends can still hear me through discord.

I've found many different suggestions on how to deal with this issue (including this forum) none of which solved it. Things I've tried and didn't help:

• cleaning the entire PC and reconnecting all of the cables
• new PSU with new cables
• different nvidia driver versions
• stress test on windows and ubuntu (I have a dual boot) which all failed

The only thing that helped is restricting the power consumption of the GPU to 50% using MSI Afterburner. Obviously it drastically reduced the performance to an unacceptable level.

I didn't find any defects on the GPU including the power connectors.

What should I try next?

Thanks.


** edit **
My specs:

CPU: Ryzen 3900x
CPU cooler: Cooler Master MasterLiquid ML360R
Motherboard: Asus ROG Crosshair VIII Hero
Ram: 2 x 16 GB G Skill TridentZ F4-3600C18
Main SSD: Sabrent Rocket 4.0 1TB (1000 GB, PCI-E 4.0 x4)
Second SSD: Corsair MP600 PRO XT (1863 GB)
GPU: NVIDIA GeForce RTX 2080 SUPER
PSU: EVGA 650 GQ, 80+ GOLD 650W (3 years old)
Chassis: Cougar Panzer-G
OS: Win 10 / Ubuntu 18
Monitor: AW3423DW

 

[Lutfij]

When posting a thread of troubleshooting nature, it's customary to include your full system's specs. Please list the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model.

 

[yliats]

When posting a thread of troubleshooting nature, it's customary to include your full system's specs. Please list the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model.

I've added this info. Thank you.

 

[Lutfij]

new PSU with new cables
You stated this yet you also state that the PSU is 3 years old. How can it be old and new at the same time? As for your GPU/PSU, did you try powering the entire system with a higher wattage, reliably built PSU that can deliver 650W towards the entire system, borrowed from a friend or neighbor?

different nvidia driver versions
Can you walk us through how you did this? Did you use DDU?

 

[yliats]

new PSU with new cables
You stated this yet you also state that the PSU is 3 years old. How can it be old and new at the same time? As for your GPU/PSU, did you try powering the entire system with a higher wattage, reliably built PSU that can deliver 650W towards the entire system, borrowed from a friend or neighbor?

different nvidia driver versions
Can you walk us through how you did this? Did you use DDU?

Thank you for your reply.

PSU: the 3 year old one is my PSU which I tried replacing with a new one (that my friend just bought). The friend's PSU (I don't recall the exact model) was 650W as well. During my check my PSU powered the entire system except the GPU while my friend's PSU power only the GPU, so I doubt I was anywhere near the power limit.

My system remained almost constant for over 3 years and worked great, so I don't see a reason for a sudden high increase in power consumption. I also expect the OPP of the PSU to kick in in a way that shuts down not only the GPU.

Drivers: Safemode -> DDU -> restart -> windows 10 automatic drivers download. After making sure it didn't help I've updated the drivers through geforce experience which didn't help either.

I also assume that since this problem started to occur on my other OS (Ubuntu 18) this means that the nvidia drivers are not related to the issue.

 

[Lutfij]

Safemode -> DDU -> restart -> windows 10 automatic drivers download
You should install the latest GPU drivers sourced from Nvidia's support site and manually installed it in an elevated command, i.e, Right click installer, as opposed to letting the OS source the drivers for you.

 

[yliats]

Safemode -> DDU -> restart -> windows 10 automatic drivers download
You should install the latest GPU drivers sourced from Nvidia's support site and manually installed it in an elevated command, i.e, Right click installer, as opposed to letting the OS source the drivers for you.

Thanks for the recommendation. I've tried manually installing every available driver released in the past year (including the latest).

 

[Phaaze88]

Gpu VRM/voltage regulator probably can't hack it anymore and puts it's foot down when you put a high enough load on it.

The only thing that helped is restricting the power consumption of the GPU to 50% using MSI Afterburner.

Yeah, that's one of the workarounds for it, especially when the solution would require replacing onboard components.
Try turning down the graphics eye candy and capping the fps(60 or something), and see if it yields the same behavior.

 

[yliats]

Gpu VRM/voltage regulator probably can't hack it anymore and puts it's foot down when you put a high enough load on it.


Yeah, that's one of the workarounds for it, especially when the solution would require replacing onboard components.
Try turning down the graphics eye candy and capping the fps(60 or something), and see if it yields the same behavior.

It does, but the reduction in graphics required for stable operation leads to unacceptable quality. My GPU is still under warranty, I just wanted to make sure that the GPU hardware is the problem and there nothing else I can do except replacing the GPU.

 

[yliats]

This is still an open issue. What else can I try? Thanks

 

[Phaaze88]

There's really nothing else beyond the software based options of lowering power limits, capping fps, and turning down gpu eye candy.

 

[Polico]

Dude, there's is a lot of threads like this in the last month, me included. Could be something they broke with the last driver update? I installed other drivers, de 517.48 but it happens anyway. Changed the thermal, changed the PSU.... It just have the same pattern in very different PC's, it has to be software.

 

[Polico]

Hi, well, for those looking for the same problem. I tried 2 things:
- Installing a new disk to make a fresh windows installation and drivers installation from scracth. The problem persisted.
- In the same Pc as always changing the GPU for another one and the problem is gone. So it's clearly the GPU with some connection problem. I will try to check the connections and resolder them.

 

[yliats]

BTW, it took me over 2 months, but eventually Gigabyte replaced my card with a new one (it was under warranty). My conclusions are:

1. This was a 100% graphics card issue, since now that it was replaced with a new one, I had no issues at all.
2. Gigabyte warranty sucks. They did everything they could to avoid replacing the faulty card under warranty. It took me over 2 months overall and tens of hours of invested time. I will be avoiding Gigabyte products in future.