RX5700XT rig stability

So, I had my rig running for about 24 hours, it has been stable at 250Mh/s, however.
Today I login and I found this:

And practically all ETH shares goes incorrect from all GPUs.

Does it mean it crashed 1h40m ago and somehow glitched? or?
Any ideas guys?
Could it be the internet line problem?
Even after rig whole rebooting, all shares goes incorrect anyway

It looks like it restarted at least 5 times. Memory is almost always the problem so I would drop the memory to 900 or less and see what happens. A new version of Phoenix miner came out recently too. I’m not sure when or if HiveOS pushed it out but NiceHash did it last night so that is a potential cause. I use the Phoenix miner log files to go back and look at what happened. I’m not sure how to do that in HiveOS. Dropping the memory a bit has fixed every instability issue I have had so far. Beyond that, rolling back the AMD drivers to 20.10 might be something to try. Your experience with HiveOS so far is making me appreciate Windows a little more. :slight_smile: (I still plan on giving HiveOS a try someday)

1 Like

No the other 4 restarts I did that, the only abnormal one was the last pointed at 1h40m.
However, I changed the pool address to another server url, and now it works smoothly.

I have no idea… eu pool is closest to me, but it keeps going incorrect shares past that abnormal reboot, changing to usa pool, even with +200ms now all shares are correct.

Perhaps the eu pool of nicehash is bugged?
How is possible that usa pool gives me 97% integrity and eu pool 70% dropping down to 13% ?

After 2h30m of running it started again to put incorrect shares.
Changing pool again solved the issue.

I don’t understand why it does that.

1 Like

Sounds like either you or the pool are just having network issues today.

@vlck After messing with these for a few weeks now it seems correcting stability issues has been as simple as dialing things down or simplifying my configuration. I see post elsewhere that are utilizing all these new extra voltage settings and it’s great that these options exist however, I found that in most cases they really are not necessary and should be used if there is a identified need to do so? That said, I have only two settings on these, 1350 core and 920 mem, I can set these higher which will indeed yield a higher hashrate but this come with a trade off where you will experience more frequent error causing reboots.

I’m sure these settings vary between brands/vendors however I have found from surfing around that these are a pretty save place to start, with stock bios, and you can work from there. These settings are stable for me getting just over 53MH/s on ETH and almost 22MH/s on RVN.

*Note: Blue temp is core and red temp is mem temp and still researching if this is normal?

if you click the error message it will show you the last few pages from the mining agent screen so you can see what went wrong

1 Like

invalid shares are almost always a result of clocks speeds too high and I would start with memory

1 Like

Updating my situation, it was likely a pool error giving me invalid shares.
I still have to reboot the whole rig once in a while, up to 2 times per day, as them seems to crash and provide only invalid shares and zero accepted, after the reboot, them go back to normal.
As for now, them produce an average of 0.00072000 BTC ≈ €5.81 per 24 hours, so I’m kinda satisfied.
Of course the crashes and reboots are an hassle, but I think everyone has to deal with it, right?

no that isn’t normal your rig should be able to run for days without error or longer

1 Like

Seems something is off then :thinking:
Because randomly them all crash and only gives incorrect shares them all or atleast a few of them

if you click the error message it will display a window with the last few pages from the miner agent screen and you can see which GPU failed first and start to walk that GPU clock down a bit starting with the memory, for Ethash on those cards you want to downclock the core and the memory just don’t go too far or you will get instability.

Also do not be afraid to mess with the VDD maybe try 850 or 900 also instead of only tweaking core and memory

The thing is it doesn’t auto-reboot, it looks like it still normally runs, but all shares are incorrect and none accepted.
Also it floods them, like 50+ per second, all incorrect.
As soon as i reboot the whole rig, it goes back to normal

As soon as you see 1 invalid share you have a problem when I get them it is because I have core or memory clock too far from default a few rejected shares are normal from time to time but not invalid shares those are red flag

Mmm I see, in 12 hours I get an average of 700+ incorrect shares :thinking:

this should look like this

That’s how mine look after 12 hours

If i go down on core tho, hashrate will drop below 49mh/s :expressionless:

that is better than mining for nothing you do not get paid for those invalid shares if you factor in all the mining you did on invalid shares your speed would be much lower than 49

let me see if I can find someone with those GPU"s online that has stable overclock settings

Uhm, I can give it a try, maybe 45mh/s for 24 hours won’t hurt me anyway