Antminers doing better Accepted/Hour after a couple of restarts

[Ameador1]

Is this typical? And more importantly - does it matter? I have been trying to look at things to give an indication of a miner having problems. I mine with Nicehash and the current speed is always all over the place - so that stat does no good to pay attention to. I use AwesomeMiner and it will tell me when a miner has "died" or has not had a work unit Accepted after a certain amount of time, but these are not always that useful. "Dead" yes - but sometimes AM and NH both show they are working - but the accepted rate or HW errors have blew up. So I started watching Accepted Per Hour rate to see how they are doing. My S9's seem to be doing a little better than 22A/H for 13.5T to 14.5T units. BUT, sometimes when they re-start - they may only end up at 16 or 18 A/H after a couple of hours. I restart them, usually once or twice and they'll get going again back in the 22+A/H range.

So basically at this point - when one re-starts I will watch it's A/H rate for an hour or two - if it is under 22 - I'll restart it. Rinse and repeat until it goes over 22A/H. Twice now (over the last couple of weeks playing with this) I have gotten a 13.5T to go from mid 22A/H to over 25A/H and the same with a slower than usual 14T unit. My 14.5T unit usually runs at mid 24A/H.

One, am I right in thinking this is a good stat to monitor for performance changes - it is Accepted work units they are paid on - correct?

The TH/s rate on all of them seem somewhat non-related. Like the 14T that is running fast right now - when it had ran for almost 2 hours after a restart - was showing TH/s(Avg) at 14.17TH/s - but at like 21.3A/H - After two restarts it's TH/s(Avg) is at 14.07TH/s - but it's A/H is at 25A/H!

Also, why is it that this happens? Probably half of them will run too slow after a restart - requiring me to watch their stats and keep restarting until they get up to par.

And keep in mind - I always give them an hour or two before paying too much attention to their A/H rate - as they fluctuate a lot right at first. So all these numbers are after giving them a little while to "settle" down and get consistent.

I just wanted to get some feedback, let you guys know I am seeing this, and to confirm that this is, in fact, worth paying attention to.

[fanatic26_]

Try pointing your miners to a real pool and not that nicehash junk and see if they act the same way, which I doubt they will.

[Ameador1]

Try pointing your miners to a real pool and not that nicehash junk and see if they act the same way, which I doubt they will.

I tried multiple pools and Nicehash was paying the best. If there are others that do better ROI, I'm game - but I also value my time and hopping around everywhere is too time-consuming. I'm always open to better options though. In fact, I was looking around just the other day and have been spending a little time talking to the guys in ckpool. But not sure about taking the risk of not getting frequent enough payouts to pay the power bill. I run my miners as a business and they have to pay for themselves - so still looking. I have 4 - 13.5T, 3 - 14T, 1 - 14.5T, 2 - L3+ (525GH/s @), and 2 - L3++ (580GH/s @) units running on Nicehash and am getting payouts, after fees, of 0.00555BTC per day at the moment (well, last payout). No fees (including miner fees) to do instant xfer to Coinbase - free xfer to Coinbase Pro - free conversion to $ - free xfer to the bank. All that has to be considered in the overall cost - plus the ease. What do you recommend that is better?

[fanatic26_]

No i dont mean for profit, I mean as a control test to see if the miners act the same way on a standard pool, sorry I should have been more clear.

[Ameador1]

No i dont mean for profit, I mean as a control test to see if the miners act the same way on a standard pool, sorry I should have been more clear.

Oh, I see. My understanding of Nicehash is that the buyers bid on the amount they want to pay for xTH/s processing speed. They then give the specific pools and pool settings they want to use. When the seller's miner picks up work from Nicehash, Nicehash literally has the seller's miner hashing directly with the specified pool the buyer chose to mine at. If this is the case - then they are effectively mining at varying actual pools - with Nicehash acting as an intermediary.  Plus, it seems like this has more to do with the miners to me. Like one of them will come up having higher HWE/H and a lower A/H accordingly - after a restart or two - the board with the higher HWE/H might be better and the A/H goes up as well. My suspicion is that it is at the board and chip level where some of them are operating more/less effectively from one restart to another - and that by restarting a few times - you will hit a better state of operation on the miner as a whole at some point and that this shows up best looking at the A/H rather than the TH/s or HWE rates. At least that's my take on it.

[fanatic26_]

If you are using nicehash and your hashrate is being moved to different pools and pointed in different places the accepted shares is a pretty meaningless stat. Different pools have different difficulty targets which change the accepted shares. Hashrate will look like its going all over the place as you bounce around between ones. I recommended you getting on a stable pool to test for these reasons.

[Ameador1]

If you are using nicehash and your hashrate is being moved to different pools and pointed in different places the accepted shares is a pretty meaningless stat. Different pools have different difficulty targets which change the accepted shares. Hashrate will look like its going all over the place as you bounce around between ones. I recommended you getting on a stable pool to test for these reasons.

Yeah - that is a question I had about that too. However, my miners show a very consistent difficulty - it doesn't bounce around. You would think it would bounce around - but I suppose it has something to do with them being an intermediary? I questioned them on the difficulty a couple of months ago when they had an outage - and when they came back up - all my miners increased in the difficulty of the accepted shares. Since then and until now, my difficulty is 500K! But when they first connect, they connect at 1M, then after a few minutes, they scale back down to 500K. Anyway, they said they don't determine the difficulty - that the endpoint pools do. But I have never seen any of my miners change to anything else while mining. Used to be 128K before they jumped to 500K, which is what spurred me to contact them. Isn't the password on pools supposed to be for passing difficulty rates along to the pools? Maybe they set the stratum and password while buyers only specify the user.worker? I don't know. The Accepted rate would also change if the difficulty rate changes too - to A/H would be all over the place and it is not. My hash rate per worker on the online stats dashboard is all over the place - but the miners GUI show that hash rates as pretty stable.

Do you have a pool you suggest testing on? My biggest concern is losing earnings/delays in payouts - my miners are paying their power bill - barely - I don't want them to put me in the red by bouncing around.

I think my earning rate at Nicehash is still good compared to others, so leery to shuffle stuff to test this issue. BUT, also looking at other pools to see if this is still true. It's so frustrating trying out different pools - with all the payment method variants, pool mining fees, pool payout fees, btc mining fees, etc... It's hard to compare them for net income.

[fanatic26_]

Yea I was more recommending testing on another pool for a day just to see if you see the same strange behavior, as it would give you a clue if it was poolside or something else.

I run my test miners on Kano.is pool personally, its an easy no hassle pool and you can set the difficulty in the console if you need to (which I have had to do in the past with ebit machines)