Topic: [ANN]Bminer: a fast Equihash/Ethash/Cuckaroo29z miner for AMD/NVIDIA GPUs 16.4.9 - page 131. (Read 148347 times)

Okay, comparison #3 started about 20 minutes ago concurrently mining ZEC on Flypool using a fixed difficulty of 10000 (by using the switch "-p 10000").

If you like watching paint dry then you will surely love to check on the status of the miners:

dstm: https://zcash.flypool.org/miners/t1aoQJwvJqYT32xHVpYezWBxKuxPSwaJB1h
bminer: https://zcash.flypool.org/miners/t1dU7Gve41A3b4mxL7a3oVAnBLsF9kMgDgA

I created these two addresses just for this test so they are starting from 0.

Пpивeтcтвyю! Pyccкoязычныe ecть в этoй вeткe?

well @MagicSmoker you was asking about other poll sadly for equilash algo there are not soo many good one each have they'r flow & issue but best is to find what it best for you & your mining power

Personal i wish nanopool to implement more coin coz from all my mining experience i think is one of most stable, reliable & with good info about statistic but well this is my personal preference, flypool is same good one but they came in my second preference

What is important is to have a good steady Hpower (preferable to not encourage pool hopper or horde algoritm jumper), steady amount miner & good ping for your mining rig (this is most important)

few other pool what i use:
- suprnova is fine but ocminer it have a very nasty api / web interface what bring info / statistic what can't be use at all
- miningpoolhub it have bad & good thing in same time a extremly customize amount of cash out, in what type you want to cash out (auto exchange), api is a little better but with some big delay sometime ... what i not like is fact it encourage horde jumping algoritm & if you want to be a focus miner to a specific coin when big pack of jumper came it can ruin your work if you are a small miner

away find what is best for your needs, what me or other say it can be subjective perspective

p.s. sorry for off topic speaking other way about bminer feedback

Aww, thanks, but no tips are necessary. I was going to build a 6x 1080 rig with these 2 1080s, but these were all I could buy before prices went full retard so I decided to use them to do a bunch of head-to-head comparisons of miners and pools while waiting to get 4 more.

Because of the comments @cryotoyes made about Luckpool not being the best choice for this test (or, really, the best choice of pool at all...) I'm going to repeat the comparison Yet Again, except mining ZEC on Flypool. It turned out that what @cryptoyes warned about occurred - that luck on a per-share basis, as well as variable difficulty, can skew the results enough to render them pointless.

In this case, the average hashrate reported on the pool side was 547 for bminer and 532 for dstm, or 2.8% higher for bminer. However, dstm earned slightly more ZEN, at 0.10776779 for dstm vs. 0.10643226 for bminer, of 1.25% more for dstm. Both of these metrics can't be true unless dstm got slightly more difficult shares and/or had slightly better luck at solving them.

Interesting, i didn't think until now there would be much interest. I had coded the entire proxy quite a while ago, and have been using it internally in our farm. It's a pretty general stratum proxy that works for most equihash pools and some ethereum ones, probably others too. Will just need to add per-client stats. then you connect one rig running dstm, another running bminer, another running ewbf, etc (and whatever else you want) and have running stats stats over a set of predefined windows, or since connection time.

We'd have to think about how to prevent "fake stats" (sad...). Open source normally means anyone can submit bogus data and I wouldn't want the miner's buddies to make some miner look better than it actually is. Some form of validation would be desirable. Maybe do it invite-based at first to folks here who are interested before opening it to the public. Or issue auth keys (sort of like unique invites).

make sure you add your crypto address so we can drop you some donations. this is something i'm surprised no one has done yet and would love to see ACTUAL numbers.

pre-thanks!

I know, I know - the whole point to what I am doing with these two cards right now is co I can avoid underutilizing my one real mining rig (which is still not all that impressive on account of the GPU shortage - just 6x GTX 1060). As for luckpool, I was holding out some hope that I would eventually win a block myself (yet I scoff at playing the lottery... go figure).

But I do loathe scammers, and I especially loathe paying to be someone's beta tester. Check back through my posts to this thread for evidence of that... 



This sounds exceptionally useful. Something like TCPView except filtering out all of the traffic that isn't related to the miner and tallying up the results?

@MagicSmoker - It doesn't really matter what you mine as long as it's similarly profitable (just sell it on the exchange and buy what you want). Even if not, two days with a single card makes almost no difference, especially if the primary interest is in getting a more accurate test result which could help you improve your profits far more than that in the future. Mining zec is quite all right. Give Flypool a try I say

As for luckpool: 2 MH/s is still measly. Interesting that you prefer an almost sure loss of 9% but are at the same time bothered (like I am) about 5% difference in miner performance ... or perhaps you dislike scammers as much as I do

I'll be busy the coming week, but will eventually release that proxy to allow more accurate comparisons of miners without wasting profits. I'm even thinking of having the proxy upload anonymous results on a public space at github for everyone to see (it will all be open source). If anyone is skimming, they need to be exposed. This community has been taken advantage of far too much.

Thanks for pointing that out... now if only they had pools for ZEN and ZCL.


Yeah, I knew that my average payout would effectively be reduced by 9% (1 block out of 11) but I chose them because - up until very recently, anyway - they had the shortest time to find a new block and my experience thus far, particularly with PPLNS pools, is that if there is too long between blocks you get royally screwed on your earnings. The most egregious example of this is VTC on give-me-coins where the time to find a block varies from 30 hours to over 100; you can literally mine there for days with no increase to your balance, leave the pool for something more productive and then it finds a block, say, 20 hours later and you still get nothing because the PPLNS window has closed.

So I figured an almost guaranteed loss of 9% was better than, say, randomly getting nailed for a 100% loss.

could you add the list of device in the next release? for eg. I want to run some device in zec, some device in zen. Thanks you

@MagicSmoker

- Flypool reports 24h hashrate right there in the graph. It's the rightmost orange dot or the value reported at the top for average, e.g. https://i.gyazo.com/2026ef989969e01c7003d854512bedc5.png (the orange dot is better because you can read it with 2 decimals more or less, the value at the top only gives you a single decimal)

- Luckpool only gives you a 12h average, which is shit (and all miner stats are lost when their GUI goes down, which it does almost daily), it's the small number at the bottom of the first box: https://i.gyazo.com/611445b6a74f35539321d1adf621be35.png

To all of you: Please STOP using Luckpool. You're aware that it gives a full coin to the block finder (that "jackpot" shit they have), and reduces the payout for everyone else, right (block finder gets 1 ZEN, everyone else shares only 10 ZEN instead of 11 ZEN)? This advantages the large miners. You will lose profits with it unless you have a large hashrate larger than the majority and you mine for very long periods uninterrupted. Don't listen to the idiots who claim that everyone has the same chance to find a block and get the jackpot. That only holds if *everyone* mines an infinite amount of time. Otherwise, guess who will find more blocks per unit time, and who is more likely to leave the pool before they find a block? Anyone with some statistics background can see the scam behind this. If that bullshit were true, casinos would be going bankrupt, instead of thriving (they have more money than you).


Awww, isn't that cute. There had to be someone who'd resort to the fabulous and very constructive argument of "do it yourself if you don't like it" ... sounds like you're either a genius who can do everything and needs nobody or are content with everything thrown at you. If you're happy paying bminer 2% despite some suspicions of fraud, good for you. Some of us aren't. Some of us care about frauds being exposed if they are indeed frauds.

Go build your own cuda miner

I'm getting annoyed by these guys charging a ridiculous 2% of all our profits. I coded an rpc proxy for equihash which i'm using in our farm. I will add functionality to report hashrate for each client connected to the proxy and will then open source it. The proxy impersonates a single miner and hides the actual miners from the pool. It relays the job sent by the pool to all proxy clients. At least you can be sure they all mine the same share, start at the same time for each share and have the count done locally. Sadly, because these guys closed sourced their miners, you can't control the seed they start with for each share, so luck will STILL be different between dstm and bminer ... i.e. you will still need to wait a long time for reliable results, but at least I can compute total average hashrates, rather than only 12h or 24h.

If bminer is cheating, I want to expose it. Will look into this next week when hopefully I'll have more time. I stopped using bminer after doing my dstm vs bminer test.

Actually, you did not specify pool or share luck in your previous message, but I understand your point now and it makes sense. I still believe that my methodology would be okay *if* I had a large enough number of shares for each miner, even if they were variable diff, but...


I look at payout because I don't know how to get average hashrate from a pool except by looking at the SMA line on the graph, and that doesn't strike me as terribly accurate.


I don't currently have the expertise to set up a test network/private chain... 

@MagicSmoker - you're missing the point about luck. It's not the pool's luck I was talking about (see my message again) but each share's luck. As I explained a while back, even if you are mining with 2 rigs simultaneously, the 2 rigs will receive different shares to mine, and those shares will have different luck. Because you have a single card in those rigs, the payout will thus be highly variable.

The biggest flaw is that you look at payout instead of pool reported hashrate. The second flaw is that you are using a single card (a single card would be ok if the pool could give you 1 week average, but none of them do). The third is that you are using a bad pool (bad because it goes down often, has a highly variable ping, only reports 12h average, and has a very low hashrate -- yes, 2MH/s is very low, especially given you're using a single card).

If you don't have the expertise to setup two identical private chains or divert rpc packets, then the best you can do is to use Flypool (226 MH/s) and look at the pool reported hashrate after 24h. It will be more reliable than luckpool+payout, but still slightly flawed (again, single card, different lucks for each card).

All that being said, I'm now pretty sure bminer artificially inflates numbers. Oh well, what do you expect - these guys want to stay anonymous and use shady https connections back to their own servers...

Sorry to hear that. :-(
Contrary to your experience, I've read many people get less earning with NH and now switching to MPH. I don't know, I guess it is just each one experience.
I use MPH for few months now, and quite satisfy with it. It has transparent fees and rewards, pretty straightforward.
Also you can choose to auto-exchanged to many other coins besides BTC (I choose LTC myself).

I did a concurrently run comparison between MPH and NiceHash for 1 week which ended a few days ago and MPH lost, earning 10% less than NH; that's not exactly a good showing, given that NH more or less sucks.

MPH's website is a confusing mess, as well. For example, they don't list the coins available for each algorithm like on most Yiimp-based auto-switch, multip-algo pools. Manually checking those coins I know use Equihash it looks like the only one that finds blocks frequently enough for a 24 hour test to be valid is Zclassic. I also don't have a wallet set up for Zclassic yet, but I've been meaning to do that so this isn't a real obstacle.

I don't want to use auto-convert to BTC mode, however, as I noticed my BTC balance actually going down while coins were still waiting to be auto-exchanged.

@RealBminer :
I need to setup Bminer to run with Awesome Miner, coz I need the Profit Swicthing feature of AM.
Could you kindly advise ? Thank you.

I would prefer Miningpoolhub. :-) There are 4 coins of Equihash there to be mined, which are mostly top of profitable Nvidia mining these days.
(Just suggestion. :-) )

Thanks for the constructive criticism @cryptoyes and @sir_blacks. There's only 4.5 hours to go on the current test so I might as well let it run but I will start a new comparison on Flypool. Still with just 1 GTX 1080 for each miner, unless one of you is willing to sell me 2 more 1080's at MSRP or less  

EDIT - so, looks like Flypool only has ZEC for Equihash, which isn't my preferred coin to mine. Anyone have a suggestion for a high volume ZEN pool?