Good morning all!
I’m chasing my tail, but I think I have enough data for someone with more knowledge than me to draw an informed conclusion. I won’t bore you with the little details, but here’s the gist:
Gigabyte orus Elite Wifi x570 Motherboard
Ryzen 7 5800x
Corsair Vengeance LPX 32GB (2X16GB) DDR4 3200mhz
NZXT Kraken X63 AIO
Thermaltake TT-1425 140mm case fans x5
Thermaltake TT-1225 120mm Case Fan x1
Universal RGB LED strips x2 (5w total consumption)
Corsair RM850x from 2015
-2080ti, bought used December 2020
-PO said he had to replace a fan
-Used for 1440p144 games since I bought it
-Possibly reduced to 1080p240 for gaming
– Card working perfectly, not a single issue playing at 1080p240 and having a second monitor (21:9) 3440×1440 144hz 34″
-Fast forward to January 2022, graphics/GPU driver crashes and causes reboot when playing anything graphically intense. I don’t know how long it lasted because I mainly play Valorant, Phasmophobia, etc. and these don’t challenge the map at all.
-March 2022 I have a 3070 on hand now, swap it out and do a full teardown of the 2080ti on the PCB to inspect for damage, solder issues and reapply thermal paste, just in case.
-What I find is that the previous owner certainly did more than replace a fan, he had taken everything apart before. Thermal paste looks like finger paint from kindergarten. No obvious damage anywhere, extremely dirty.
-Iso bain, brush every square mm, Q-tip tracking, brand new and shiny card. Apply Micronics MN-GL4 thermal paste (8.5w/mk, stability -40c-150c)
This is where it gets interesting
– Meticulously reassembled everything, could probably use new thermal pads and such a small block of foam for the back plate was missing when disassembling. I’m just careful to tighten everything so as not to put pressure on the PCB against the back plate
-Insert into the top PCIe slot and launch Heaven’s benchmark. I know it’s not the best performance indicator for today’s GPU, but I thought it would underscore it enough that if it still fails, it will definitely fail
-He definitely fails. Immediately. Don’t even start the benchmark. As soon as the program opens I get a black screen and a hard restart of the whole PC
-GPU-Z and try again. This time recording all the data
-After a few unsuccessful attempts, I decide to move it to the lower PCIe slot
-Charge Heaven’s and it doesn’t crash. I left it idling throughout monitoring temperatures and eventually the memory temperatures hit 83C and the card shuts down. As far as I know, the 2nd PCIe slot is powered by the x570 chipset at only x4, although HWinfo showed me x16 in the 2nd slot. I expected the thermal shutdown as there was maybe only a few mm between the board and the bottom of the PC case; he literally had no room to breathe.
-I saved all this data and put my 3070 back in the top slot alone, opened the sky and sure enough it works without any problems.
Key Points: 3070 only draws around 220w according to GPU-Z during benchmark. 2080ti in the first slot is inconclusive due to failure. 2080ti in 2nd slot drew 260w, but the readings are all over the place and I’m no expert. I have a new RM1000x and 3080ti on the way, and I already have a 2nd full PC but I can’t use it for testing as it only has a white label Aresgame 500w PSU. The plan is to swap the 850x into the 2nd PC, test the 2080ti on its motherboard (Asus B450m A/C) and pray. Before swapping them out, I’m going to put the 3080ti through its paces on my 850x in the top slot to see if it’s a power supply issue.
Can anyone look at the attached excel sheet and tell me what I’m looking at? GPU failure? PCIe slot failure? power supply ? Combination?
Edited by iSniperpls, Today, 06:26.