RX 6900 XT Flickering and Re-starting

J20Joe

J20Joe

J20Joe

Associate
Joined
30 Aug 2022
Posts
9
Location
Leeds
Hi,

Bought a RX 6900 XT from Ebay (used and from a decent seller) and started to see issues, Flickering in most games and restarting the pc very frequently, wondering if anyone has the same issues and a cure?

Ive tried overclocking the GPU and at stock stats and it re-boots. sometimes i can get a stable over clock and run super position bench mark @4k over and over, i can game and then a couple of days later it crashes and i can no longer run the bench mark anymore? i feel like i have tried everything.

Specs:

Sapphire Nitro + RX 6900 XT (has 2 bios modes, quiet and performance, performance crashes often)

Ryzen 5 5600x

Corsair Venegance RGB pro 2 x 8GB 3600MHZ

Evga p2 1000w Plat PSU

AS rock B550M Mboard

2 x M2 Gen 4 and gen 3

PCIE Riser cable



Temperatures are very low as it is a closed loop, the guy who i bought it from says he had no issues and has been very helpful so i trust him.



Current OC (stable for 2 days now but i doubt it will last):

350W override

2020MHZ DRAM speed

Fast timing Ram

PBO +200MHZ - 20 all cores (cinebench R23 ran fine many times)

SAM enabled



What i have tried:

DDU (done this about 50 times)

Clean install with a NEW SSD of windows 11

ULPS disabled

Power plan set to max

return to stock settings (not stable)

Tried a couple of previous drivers (worked fine and then flickered)

turned off windows and AMD updates when stable but then after a couple of days it came back

Disable Free SYNC

updated Mboard bios



To fix the issue i dial the clock speeds to around 2000MHZ and then turn back up, very strange behaviour, when the card works it works and it performs very very well but the constant disappointment when it doesnt work is killing me, its as if i need to turn my pc on and wait 15 mins before any sort of stress for it to work properly.



If you've managed to read all this well done, if you can help that would be much appreciated
 
Interesting you should say Remounting the water block, what are your thoughts with that? the guy i bought it from changed the thermal pads and re built the waterblock before i bought it, and he said he had no issues before, dont think he tested it out after rebuilding it.

Cannot try without the riser cable, i could if i were to take off the water block to be fair but the crashing is too in frequent to check this, id have to breadboard it which is not ideal, however i did use a PCIE 3.0 known good riser cable and that also had same issues
 
Riser cable - just worth removing that potential issue from the equation, at least you can tick it off.

Do you have the original cooler to try? How are the temps, all look normal?

One other thing that I have had in the past - I had a memory issue that was not apparent until I put a more powerful gpu in, leading me down the path of the GPU but it was actually a ram problem. Do you have memtest hci you can run to rule it out? It's probably unlikely but worth checking.
 
I can try the original cooler, but my issue is its a custom desk PC so height is a real issue, a triple slot does not fit so ive have to strip it out and breadboard it.

Temps are great, got spot under full load is 60° max.

I havent run a mem test to be honest, only thing i havent benchmarked, so i will give that a go. its just really strange that it works fine for a couple of days then i cant bench mark, i then have to reduce everything below stock and then build it back up and my original settings work again. very weird
 
Sounds like a painful issue, start with what i've suggested where you can and others may be able to provide some more insight / testing.

Do you have anyone local to you that could test the GPU in their rig if issue remains with original cooler?
 
Yeah i could potentially do that, might be a good start, need to make sure they have a good enough PSU for it.

Could you elaborate on your question about mounting the Water block? is there anything the previous owner could have done to cause issues? are temps the only indication that it wasnt done correctly? not sure if a drop of coolant in the wrong place could have caused a short of some sort, but again why would it bench mark fine one day and not the other. its a vicious circle of possible solutions. Seems there are a few people on forums that have similar issues, bit i can never find a solution. im hoping my current OC is stable and remains stable, so Far 1 day :D
 
I'm not 100% to be honest, someone else may be able to clarify, but my experience of watercooling GPUs I have had bad contact which has caused instability, despite temps being OK.

I think it's definitely worth checking, especially if they've re-done the pads etc, to make sure everything looks good. But see if others think this is sensible.
 
Maybe the cooler temps it's now running on the loop is pushing it to clocks which are on the edge of stability which is why the original owner assuming he was air cooling didn't experience.
 
Original owner had it Water cooled. im not overclocking the frequency at the moment, letting the card boost it, he did say he managed 2700mhz without issues, ive set it below and seen issues
 
And the card is not working again today, booted it up and ran a benchmark that I've ran 10 plus times yesterday and I get the same crashing.

Video of the crashing is here, it also crashes at stock settings

Dropbox

www.dropbox.com www.dropbox.com
 
When did you get the card?

One way to look at it is eBay buyer protection exists for a reason, and you have already gone above and beyond a typical user troubleshooting.

May be worth returning & choosing a different card?

Just food for thought.
 
Bought it with block installed, the guy said it worked fine before, i havent touched the block.

Wonder if it could be the riser cable? im not sure if there are any good brands but the one i got was around £30 so im not sure if that is a budget one, it says it supports RX 6900 xt and is PCIE 4.0, and is set to 4.0 in the Bios
 
You must log in or register to reply here.
Back
Top Bottom