did you try other miners than nbminer ?
try trex or gminer.
try to set fan to 90%
Before encountering the problem, I was using TRex. I changed with NBminer and Gminer, same problem.
I crossed the problem card with a non-problem card, same result.
I changed the pcie then the usb cable then the riser, same result.
My bios is ok.
The card has no errors under windows and runs fine.
I also reversed the power supply with a no problem card, same result.
As I said, I will try the card alone on the rig.
On the other hand, Keaton, I donât understand how I can launch Hive under windows. Thatâs to say?
Thank you for your time for me 
Edit: On the other hand, I ask myself the following question: My 6 gpu which work without problem have an oc vram at 2500. Can these oc influence my problem card? If I drop from 2500 to 1900 (for example), could that solve my problem?
Try do downclock the memory.
Stick pendrive with hiveos into the windows rig and boot 
Hi. Iâm only coming back now because itâs quite busy.
Regarding my problem, it canât come from the GPU. You can see it on the following screens.
Personally, I donât know what to do anymore.
My last action was to reset the ocs, in maintenance mode, restart then put the ocs back. And restart. It didnât change anything of course.
So, I stay with 6 GPUs unfortunately
look at the gpu-z chart.
Power consumption 46%
perfCap Reason: Vrel
its all you need to know. theres probably thermal pad/ cooling problem - you could check that lead.
try to use this chart under windows, to invastigate whats wrong.
My 3080 had thermalpads on the memory problem, other one, just wont work on high clocks.
Run furmark once and look, than user trex mining eth to see if its memory. search in the internet for a guide chow to change pads. see if youre gpu has simmilar on it.
So the card mines fine under windows at the same clocks?
If so, boot hiveos on that system and see if itâs a power supply/cable etc issue on your rig or if itâs the card itself.
Same issue for me with gygabyte 3060 ti lhr oc pro v3.
I RMA the card to vendor who test the card and dont find any problem on windowsâŚ
Hynix memory.
Hi Keaton, no I havenât mine on win10. However, I uninstalled the graphics drivers for my 2080s and then installed the 472 drivers for the 3060ti. I launched a bench and the card works very well.
Here is the result after 10 minutes
I will do cross tests on the rig with the power cables. Iâll plug it into the riser of a card that mine without problem. I will come back in the morning to give you the result
On the rig, I reversed the GPU0 card without changing anything. The result remains the same, fan 0%.
Iâm tearing my hair out!
I donât think itâs a hardware issue.
- Riser => Ok
- USB cable => Ok
- PCIe => Ok
- Power supply => Ok
- Motherboard => Ok
- GPU => Ok
- Flash SSD => Ok
- Drivers => Ok
- HiveOs Version => Ok
Is beyond me
Hi, mine is a Gainward Ghost V1.
On the other hand, it seems to me that the V3 causes problems, to be confirmed.
Try mining in windows.
Show us screenshot with a miner and gpu-z like you did with furmark.
In my opinion, its thermalpads problem on memory and/or power section. Its mining at ~90% - so is the powercap seen on GPU-Z
Apparently, even under w10, it does not want to know anything beyond +550 in memory.
Here are some screenshots
Launch of TREX
+400 memory
+550 memory crash
That makes more sense, you can put it back on hive since itâs not an os issue. try using lolminer, it will at least repair the dag as much as possible.
Reduce mem oc when it crashes until itâs stable as mentioned before.
Thank you very much for your expert eye
I went down to 800mhz and itâs been working for several minutes.
On the other hand, Lolminer, it consumes a lot for less hash.
So can I switch to another miner?
Will my card be throttled to 800Mhz until it dies?
Keep lowering the mem clock and reboot after each crash. You can even go negative on the mem if needed.
And nbminer (teamblackminer and phoenixminer also) artificially inflates the local hashrate 1-2% compared to some miners. Even though it looks lower with lolminer, it will still be just as much or higher at the pool (what matters)
Thank you for all those informations.
The gpu has been running at 500mhz for a while now.
Hoping it doesnât crash again
Hello Friend, I haved the sam problem with Gainward 3060 TI.
I have test for many days.
I have this for you, when the Gainward makes problems.
Are you reducing mem clock each crash and rebooting?