Question New GPU Crashes/loses signal every time any amount of moderate stress is put on it

Jun 23, 2020
9
0
10
Hi,

I recently got a 2080Ti, and immediately put it on a water block. After installing it, I noticed that my monitor loses signal, or the card is crashing under moderate loads (tested on games such as CS, Minecraft and Terraria). Initally I had assumed that there was an issue with my PSU, however I replaced it with a higher wattage one and I'm still getting the same problem.

So far, I have tried:
-Reinstalling drivers
-Installing older drivers
-Using another monitor
-Using HDMI instead of DisplayPort
-Swapping power supplies and PCI-E power cables

Is there anything I should try before RMA-ing it?

The rest of my specs are:
-i9 9900x
-32GB Corsair Vengeance RAM
-Asus X2999 Prime
-Two EK waterpump/resevoir combo

thanks for any help in advanced.
 
Jun 23, 2020
9
0
10
Have you kept your eye out for temps while gaming? Did you check that theres nothing wrong with your pump, waterblock etc.
I don't even have time to check temps in game before the graphics just crash. It sits idle at about 52C, which isn't abnormal if I understand correctly. My pumps should be fine as I've had the running for a couple years with a watercooled 1080.
 
52'c idle (ie no or minimal GPU load) for a 2080TI sounds like a lot to me, even if it was air cooled I'd consider it a quite high idle temp, unless ambient temps are very high. With such a significant idle temp, I suspect load temps can get really high quickly, which could potentially be causing the crashes you are experiencing.

You could try to set a monitoring app to log values, and see if you can then open the log after the PC crashes - it might save the last recorded GPU temperature, and maybe provide an indication wheter your GPU is running hot or not.

I did a quick google search for the 2080TI Founders Edition idle temp, because of the stock Nvidia air cooler (instead of a highly optimized custom cooling solution), to get an idea what a more general idle temp would hover around. There's a guy in this thread with a watercooled 2080TI which doesn't even reach 50'c at full load. I know you can't compare custom cooling loops directly, but to me it is an indication that your idle temp is rather high (considering it is watercooled) and your card could possibly be overheating under load.

I know some 2080 Super GPUs doen't get AS hot as some 2080TIs, and I know they use different chips, but my aircooled 2080 Super idles at 32'c during the summertime with a room ambient temp of approximately 24'c (fans spinning as slow as the software allows).
When it is not hot outside, the card idles at 30 - 31'c

Even if they are different chips, a 20'c difference in idle temp, especially when considering it is air vs. water cooling, is more than I would expect to be normal

EDIT :
I have seen other reports of 50'c-ish idle temps for a 2080TI, but those are with air coolers using idle fan stop features, which mean that they are basically passively cooled when idle, fans only kicking in at a certain temperature.
 
Last edited:
Jun 23, 2020
9
0
10
52'c idle (ie no or minimal GPU load) for a 2080TI sounds like a lot to me, even if it was air cooled I'd consider it a quite high idle temp, unless ambient temps are very high. With such a significant idle temp, I suspect load temps can get really high quickly, which could potentially be causing the crashes you are experiencing.

You could try to set a monitoring app to log values, and see if you can then open the log after the PC crashes - it might save the last recorded GPU temperature, and maybe provide an indication wheter your GPU is running hot or not.

I did a quick google search for the 2080TI Founders Edition idle temp, because of the stock Nvidia air cooler (instead of a highly optimized custom cooling solution), to get an idea what a more general idle temp would hover around. There's a guy in this thread with a watercooled 2080TI which doesn't even reach 50'c at full load. I know you can't compare custom cooling loops directly, but to me it is an indication that your idle temp is rather high (considering it is watercooled) and your card could possibly be overheating under load.

I know some 2080 Super GPUs doen't get AS hot as some 2080TIs, and I know they use different chips, but my aircooled 2080 Super idles at 32'c during the summertime with a room ambient temp of approximately 24'c (fans spinning as slow as the software allows).
When it is not hot outside, the card idles at 30 - 31'c

Even if they are different chips, a 20'c difference in idle temp, especially when considering it is air vs. water cooling, is more than I would expect to be normal

EDIT :
I have seen other reports of 50'c-ish idle temps for a 2080TI, but those are with air coolers using idle fan stop features, which mean that they are basically passively cooled when idle, fans only kicking in at a certain temperature.
Thanks for your reply,

I gave my radiators a clean and changed the coolant. My GPU is sitting at a cool 40C now when idle and about 48C under moderate load (I was able to catch the temp of the GPU just before a crash). The only thing is, it's still crashing. I ordered new thermal pads, could this make a difference or should I start to explore RMA?
 
I doesn't look like overheating is the issue, now that you've improved cooling and get good results.

What is the make and model of the PSU you upgraded to?

Is the 2080TI connected by two seperate PCI-e cables, and not just one PCI-e cable with two connectors?

I can't think of much other than what you've already tried, so it could be, that RMA is the only thing left to try, unless someone else can spot something obvious
 
Jun 23, 2020
9
0
10
I doesn't look like overheating is the issue, now that you've improved cooling and get good results.

What is the make and model of the PSU you upgraded to?

Is the 2080TI connected by two seperate PCI-e cables, and not just one PCI-e cable with two connectors?

I can't think of much other than what you've already tried, so it could be, that RMA is the only thing left to try, unless someone else can spot something obvious
The power supply is a Corsair HX1000i, and I know there's nothing wrong with it as I bought it second hand from a coworker who pulled it straight from their build, which is very similar to mine.

As the PCI-e cable, it's just one cable with two connectors.
 
The power supply is a Corsair HX1000i, and I know there's nothing wrong with it as I bought it second hand from a coworker who pulled it straight from their build, which is very similar to mine.

As the PCI-e cable, it's just one cable with two connectors.
Change it so it uses 2 seperate cables rather than just 1. That might be your limiting factor.
 
Jun 23, 2020
9
0
10
Change it so it uses 2 seperate cables rather than just 1. That might be your limiting factor.
I did that, and it didn't seem to work.

Before I start the whole RMA process, I'm going to try re-applying thermal pads when they arrive.

I appreciate everyone's replies and help, I'll keep you all updated.