Nbminer always crashes with code 8 after a few hours

[szepesif]

Hey guys!

I read all discussions but couldn’t find any useful infos or tips how to prevent it. I’m using the latest nbminer (40.1) with 3 nvidia cards and nbminer is always crashing after a few hours. Sometimes 2, sometimes 6, but I always find this message in the logs eventually:

There’s no invalid shares, sometimes I see a few rejected ones but not much. My OCs are perfectly normal, nothing abusive, all cards are on the safe side. The nvidia driver is 470.103.01 at the moment.

I would run Trex but it’s already running in this flightsheet (2 other nvidia cards for a friend) so I can’t run it again and I’m stuck with nbminer. I would choose a different miner but in my experience all of those are way worse and more unstable than Trex and nbminer.

I just want to know how to prevent the crashes and I would like to figure out why it’s crashing at the first place. Which card is causing it and why. Unfortunately the logs are not telling me much and I have no idea what error code 8 means.

Do you have any idea what could I do or try to make it run stable?

[keaton_hiveon]

Can you post a screenshot of your worker overview screen?

Also, I’d recommend gminer over nbminer as it doesn’t fluff the local hashrate and has a lower dev fee on 2.73 or older, or enhanced lhr unlocking on the newer versions depending on what your needs are.

[szepesif]

Sure, here it is.

GPU 2 (the only LHR card in the rig) and 4 are running Trex to a different wallet and the other ones are mine with nbminer. So I don’t have any LHR card at the moment with this nbminer config. Feel free to let me know if you spot any issue with my OCs. Unfortunately my cards are really unique, none of the popular settings are working for those, I had to figure out my own core/memory/PL values for almost each card.

So your suggestion is to try gminer 2.73 to see what’s happening?

[keaton_hiveon]

Id imagine your stability issues are from your ocs. You don’t want to use core offsets on any modern cards.

For non lhr cards, the goal is to run the lowest locked core clock that maintains full hashrate. No power limits needed.

For lhr you just need a high enough core clock to keep it from lhr engaging. 1500mhz is a good starting point for all lhr cards.

[szepesif]

I tweaked it a little bit. Is it any better?

Let’s see if nbminer still crashes or not after a few hours.

[keaton_hiveon]

The 2060 super you’re gonna want a locked core clock on as well. Which cards are submitting rejected shares? the 2060?

[szepesif]

Yes, the current rejected share is from the 2060s. I will try to set up a locked core on it as well, but that card is a mess, nothing really works with this one. The setup that you see there is the most profitable one that I could figure out and it’s still far away from the optimal mh/s for that card. Silicon lottery I guess.

[keaton_hiveon]

What does it get with 1100 to 1200 core clock?

[szepesif]

Previously I set up 120 core clock on that 2060s and I got around 39.2mh/s. That’s the best that I could get with this one. Now doesn’t matter which locked value I try I can’t get more than 38.5. I don’t understand it. I tried with 1100, 1200 and even lower/higher. Basically no reaction, 38.5 and nothing is increasing or decreasing. This card is a strange one. Or maybe the changes are not activating for some reason.

[keaton_hiveon]

Yeah should fall under the same min/max core/mem rule like every other nonlhr for best results.

If you run nvtool --throttle Does it show any throttling?

[szepesif]

In fact it does.

[szepesif]

I removed the powerlimit completely for that card but nvtool is still saying it’s powerlimited. Strange.

[szepesif]

Something is fishy here… it says different core clock as well (I set it up to 1100). I also tried to reboot the rig completely. It seems the power limit and the core clock is not changing for some reason. Do you have any idea what is going on with this card?

[keaton_hiveon]

set the power limit to the max that bios can do, in your case 175w. then try again

[szepesif]

Another issue. Anytime when I stop the miner the system is dropping the 2060s GPU until I reboot the rig. If I try to run any commands it says 0 Mhz Unknown error and unsupported GPU.

If I reboot the rig it works again but everything is the same. As soon as I stop the miner this one is lost again and I have to reboot to get it back. I really need to figure out what is going on

[keaton_hiveon]

reduce mem clock until its stable

[szepesif]

I still don’t understand this GPU. I reduced the memory clock to 1600 but it’s still saying it’s throttling and the reported core clock is 990 instead of the configured 1100. Can it be thermal throttling? But the reported temperature is only 53 celsius. I’m lost.

Sometimes the core goes up by itself to 1095 but I never seen a higher value, doesn’t matter what I enter in Hive.

Also, nbminer is still restarting after a few hours, even after modifying the OCs as you suggested.

[keaton_hiveon]

Did you set the power limit to 175?

[szepesif]

Yes, I tried setting it to 175 and even tried to remove the power limit completely (0). No change. This card is not behaving correctly.

[keaton_hiveon]

When set to 175w and rebooting after, does it show it’s throttling?