Hey guys need some help. This rig has been mining for 6 month straight. Out of the blue it has started to randomly go offline. The only symptom I have to go off of is the miner stops mid operation. The problem is it hangs HiveOS to the point that it loses all functionality. I have to hard reset via power reset. Trouble shooting so far: Swapped motherboard, memory, cpu, and PSU’s so far. Next I will be testing each psu one by one. My thoughts are it’s a bad riser.
As you suggest, testing each GPU one by one to find the weakened link in the GPUs chain.
Assuming Operating System has been re-flashed to new OS drives, and new miner downloads confirmed.
Good luck on the next phase…if we had an extra dollar for every aging GPU/riser/etc. we could have on heck of a party.
Think I found the smoking gun! I have a Saphire Pulse 6700xt that I RMA’ed. This GPU just isn’t made for mining I guess. As soon as I took it out everything stabilized. I tried to put it back in with a new riser and lowered the OC’s and it still crashed. Might be sending this one back too…
I have a tortured affair with Sapphire 6xxx series, as you have seen from the discord posts.
I’d get it in a single GPU test rig if you have one. It might just want “higher” voltages to be stable. This 6800 is my “trouble maker” but it is rock solid when I finally bumped the power variables:
I would agree if it wasn’t stable for 6 months straight on default settings and conservative OC. I had a couple of power outages and I just don’t think the hardware is strong enough. Going to RMA it again.
If RMA is available and a GPU is flaky, I’d take it…
Exactly… Do you happen to have the specific settings you used for this specific model? What are your settings for under volting and OC on HiveOS?
Do you happen to have those settings for Windows also? I would like to try both.
My 6700XT’s are all XFX brand but I would start at the medium’s here, keep voltage there, but potentially clock up the other values to test limits with a flaky card:
https://www.hashrate.no/6700xt#ETH
Windows memory frequency doubles linux values in most cases, but I would tune the Radeon tools first.
Tried to pull the voltage down and it still crashed on Windows. Lol oh well it was worth a try.
fwiw: I tend to move voltage up with my flaky GPUs…
Hey Grea do you shop at parallel miner by any chance?
I have not bought anything, but was eyeballing server supplies and breakout boards there previously.
I ended up adding more riserless boards and Bitmain power supplies for last round. I am a sucker for reducing my various SKU counts to fewest possible. Have you seen my rigs 
Lol I get it hit me up if you are interested in anything before you purchase. I can send you a discount code for the help.
1 Like
Don’t spoil me, I am too needy
Ha I don’t have enough time to spoil anyone
Could bump memory controller voltage a notch, say 825.
Memory voltage is likely fine with 1050 memory clock (absolute mem clock max is 1075)
I’ll bring my rig down and move in another 6700XT as well, arrived pre-owned. Who knows.
This GPU runs fine on its own dedicated system with those setting on Team Red miner. As soon as I put it back in the big rig it crashes. Ugh
This topic was automatically closed 416 days after the last reply. New replies are no longer allowed.