I get intermittent streak of excessive reject rate, from a usually high-ish 2% that jump to 30% to even 100% for several minutes every other hour or so.
A system to protect rig owners from bad pools would be nice. I get it buyers want to pay for accepted shares but sellers don't want to lose 10-20% of their hashrate because buyers don,t know what they are doing.
Also a setting to sign up as a vanguard for miners to be switched would be good. Some people have miners that will do okay if they are immediately switched to highest profit order, while some people not.
There is nothing to worry about this. Please, take 15 minutes and carefully read this FAQ:
How exactly does pay-per-valid-share and get-paid-per-valid share work?
https://www.nicehash.com/index.jsp?p=faq#faqg2I can assure you this is a
100% fair game for sellers and buyers.
If there is a buy order with a pool for extremely low diff and possibly even on a unstable or overloaded pool, you still
will be paid for the valid work you provided, because NiceHash validates your work before it is sent to the pool. You can make sure about this if you check your payouts and your statistics.
It is true - sometimes buyers risks and submits order for extremely low diff and possibly even on a unstable or overloaded pool - but this is their risk (I guess they know what they are doing - hint: biggest profitability can be made when you try to "attack" coin with massive hashrate, maybe even try to fork it or take the majority of network hashrate for yourself). However, they will of course still have to pay for the valid work that was provided by providers, even if pool can't "eat" all the work that has been provided (it's just the same risk as having own rigs or rent dedicated rigs for fixed time). But since there are so many benefits for buyers with NiceHash (no contracts, no time boundary, no risk for failed rigs, etc) they prefer to mine on NiceHash instead of renting dedicated rigs.
It is also true that even if we validate all of your work when we're hitting some very low diff coin, it will happen that the reject rate will be a bit higher on providers side (because rigs actually aren't capable to produce all valid work on very low diff) - but here we're talking about reject rates 2-5% instead of normal 1-2%. But this is the exact same reject rate that a provider would have to "swallow" if you would be mining on any other pool or multipool which is hitting high-profitability low-diff coins. All you have to take into account is this: you might see 2-5% higher reject rates at some intervals (again, look at the graphs on our site, not your sgminer) but you'll get paid 25% more than multipools pays. So it's up to you weather you'll go for a bit lower reject rate, or for a significantly better payout
I hope this clarified your concerns at list a bit. Since NiceHash is an innovative approach to selling/buying hashpower I do understand that there are lot's of thing to be clarified. And probably there will be some exceptions where will have to improve the algorithm - therefore your feedback (from sellers and buyers) is highly appreciated!
Sure, but here i'm not talking about 2-5% i'm talking about 8 and 16% overall reject. So i still don't think that part, that i did read previously, is relevant. I've had full rounds with 3%(both rigs) reject then blam it spike up for a whole round. I might be wrong but this tells me its the client that is sucking, if not its my rig config which seem unlikely... Which would leave NiceHash to be the cause. I prefer blaming the client. Sometimes reject spike to 100% and by looking at random key, i'm not the only one.
You seem to be blaming proper configuration, but don't give any actual information, therefore i would assume that if i get 0.4-1% reject rate on e.g waffle pool, CM, coinfu, etc, then i am properly configured.
Is there special configuration needed for your "pool"?
The 16-20% reject is on cudaminer, there isnt much configuration to be done there, but its not on supported miner so i can just be glad it work at all. But what about my main GPU running on SGminer at 7-9% (r9 290 @900khs 850 average WU, get 790 WU on your pool)?
The only difference from my usual multipool settings is i'm specifying a diff variable (128-256) that has to be same for all my miner.
I could just shrug it off as 10% overall loss of accepted share by pointing at Nicehash is still a 15%~ improvement over the best multipool i,ve found, but it doesnt feel proper.
Thoughts?
