[NemosMiner] multi algo profit switching NVIDIA/CPU miner

[dragonmike]

Hi I've been using Nemosminer for a few weeks now and I am really liking it.

I was hoping that you would be open to suggestions of how to perhaps improve it.

The miner switches algos very often but by the time those rewards mature and make it through the exchange the momentary increase in estimated BTC/day that made the miner switch has long past and you get very little of those rewards. 

Could it be possible to have a variable in the start bat that would set how often the miner checks profitability and how often it switches algos?
Ideally I'd like to be able to set it to check every few minutes but to only switch every hour based on what on average was most profitable during the past hour.  I think that working on those longer term averages will give a better chance of getting a more stable return by the time the rewards clear the exchange.

So to sum up:
NemosMiner should switch only every hour, and should switch to what was most profitable on average for the past hour.

no... im no going to make it work off hour averages or 24hour averages   its works off the realtime stats...

i recommend letting benchmark complete, then edit start.bat and change (-Interval 90   to  - Interval 300)  thats how i get the best results out of it...

Basing a mining decision on past performance is indeed not the way to go at all. You need to exploit current low diff. There is a clear problem however with the fact that coins need time to mature and more time again to be exchanged, so prices will usually have moved massively (against us, mostly) and closed that arbitrage window.

Yesterday I ran Nemosminer for 36 hours with payout in DGB. I monitore the last 24. My rig got approx 500 coins paid out. That's unfortunately not much. I switched to mining DGB only for 24 hours and already know it's going to turn out better.

As much as it annoys me, the problem with the switching software is that if coins that are mined move sideways or downwards, you lose.

In order for an auto-switching software to outperform in crap markets, it needs to be smarter, i.e. anticipating market moves (technical analysis, trend recognition etc) in order to find a way to alleviate the delay caused by maturing and exchanging the coins.

And then apply the same knowledge to predicting the weather…

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

[1714681966]

1714681966

[1714681966]

1714681966

[minestro]

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

Where does this strange 66% come from? Is your take away, there are 3 possible outcomes (rise, fall and stay the same) and they are all equally probable? That's quite wrong and it also doesn't consider what are the possible losses and gains and how much they impact your profits.
I think you are obviously ignoring other effects impacting the profitability of mining. It's not only the price at the exchange. There are things like drops in difficulty which make mining at least short term more profitable.

[dragonmike]

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

Where does this strange 66% come from? Is your take away, there are 3 possible outcomes (rise, fall and stay the same) and they are all equally probable? That's quite wrong and it also doesn't consider what are the possible losses and gains and how much they impact your profits.
I think you are obviously ignoring other effects impacting the profitability of mining. It's not only the price at the exchange. There are things like drops in difficulty which make mining at least short term more profitable.

Of course I'm simplifying the probabilities. But I was hoping you'd get the bulk of the message:

Difficulty arb doesn't guarantee outperformance when you factor in market prices and coin maturing latency. As a matter of fact, when markets go downwards or sideways -ceteris paribus- you loose.

It's not a scientific approach, only an empirical one. I did experience the same back in 2014 when I was chasing profitability with my rigs.

[jimmykl]

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

Where does this strange 66% come from? Is your take away, there are 3 possible outcomes (rise, fall and stay the same) and they are all equally probable? That's quite wrong and it also doesn't consider what are the possible losses and gains and how much they impact your profits.
I think you are obviously ignoring other effects impacting the profitability of mining. It's not only the price at the exchange. There are things like drops in difficulty which make mining at least short term more profitable.

Of course I'm simplifying the probabilities. But I was hoping you'd get the bulk of the message:

Difficulty arb doesn't guarantee outperformance when you factor in market prices and coin maturing latency. As a matter of fact, when markets go downwards or sideways -ceteris paribus- you loose.

It's not a scientific approach, only an empirical one. I did experience the same back in 2014 when I was chasing profitability with my rigs.

I guess with some wrangling you could grab historical data from an API, look at a longer timeframe and apply weighting to favour stable or coins which are trending up for a set period. But it wouldn't be simple and would add a lot more moving parts for the dev.

I doubt anything can fully replace having the time to do background research then double-check the switcher results yourself and choose to override the results if the charts aren't looking good

[minestro]

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

Where does this strange 66% come from? Is your take away, there are 3 possible outcomes (rise, fall and stay the same) and they are all equally probable? That's quite wrong and it also doesn't consider what are the possible losses and gains and how much they impact your profits.
I think you are obviously ignoring other effects impacting the profitability of mining. It's not only the price at the exchange. There are things like drops in difficulty which make mining at least short term more profitable.

Of course I'm simplifying the probabilities. But I was hoping you'd get the bulk of the message:

Difficulty arb doesn't guarantee outperformance when you factor in market prices and coin maturing latency. As a matter of fact, when markets go downwards or sideways -ceteris paribus- you loose.

It's not a scientific approach, only an empirical one. I did experience the same back in 2014 when I was chasing profitability with my rigs.

I guess with some wrangling you could grab historical data from an API, look at a longer timeframe and apply weighting to favour stable or coins which are trending up for a set period. But it wouldn't be simple and would add a lot more moving parts for the dev.

I doubt anything can fully replace having the time to do background research then double-check the switcher results yourself and choose to override the results if the charts aren't looking good

Yes, I think the question here is not to find a solution to make market predictions I think of Nemos Miner more as a set-and-forget solution. IMHO the switching just minimizes my risk to mine hours on a coin which is just going down, especially without having to do anything It's for sure not the most profitable solution but this fast (even setting the interval to 1h) switching is minimizing the risk.

[dragonmike]

There is predictive work involved, sure. But face reality: if a multi-algo-switching miner cannot outperform mining a single coin in 66% of market conditions, it has no reason to be.

Where does this strange 66% come from? Is your take away, there are 3 possible outcomes (rise, fall and stay the same) and they are all equally probable? That's quite wrong and it also doesn't consider what are the possible losses and gains and how much they impact your profits.
I think you are obviously ignoring other effects impacting the profitability of mining. It's not only the price at the exchange. There are things like drops in difficulty which make mining at least short term more profitable.

Of course I'm simplifying the probabilities. But I was hoping you'd get the bulk of the message:

Difficulty arb doesn't guarantee outperformance when you factor in market prices and coin maturing latency. As a matter of fact, when markets go downwards or sideways -ceteris paribus- you loose.

It's not a scientific approach, only an empirical one. I did experience the same back in 2014 when I was chasing profitability with my rigs.

I guess with some wrangling you could grab historical data from an API, look at a longer timeframe and apply weighting to favour stable or coins which are trending up for a set period. But it wouldn't be simple and would add a lot more moving parts for the dev.

I doubt anything can fully replace having the time to do background research then double-check the switcher results yourself and choose to override the results if the charts aren't looking good

Yes, I think the question here is not to find a solution to make market predictions I think of Nemos Miner more as a set-and-forget solution. IMHO the switching just minimizes my risk to mine hours on a coin which is just going down, especially without having to do anything It's for sure not the most profitable solution but this fast (even setting the interval to 1h) switching is minimizing the risk.

That's a lot of profit you're willing to mitigate.
24 hours of mining straight dgb returned over 700 coins (vs yesterday's 500 using nemosminer). Difficulty was pretty much the same.
To me that means the whole profit-switching concept still needs a bit more work.

[minerx117]

thanks for your Opinion dragonmike

go mine DGB then...  some of us are Lazy and don't like dealing with exchanges.. i am one off those people thats why i made this, to get paid straight to my BTC wallet...  you can usually always make more if your mine coins and exchange them yourself.... plus i got tired of trying to keep up with new coins and algo's so if something like bitcore or skunk has a big run for 24 hours and im away on work my rigs will automatic switch and mine them...   at the end of the day it is up to you, if you have better luck mining straight DGB then mine Straight DGB...

Best Regards
Nemo

[dragonmike]

Don't get me wrong, Nemo, I'm not criticising at all, just making pertinent observations and sharing experience and opinions. After all, that's what these forums are for.

For now I'm back to mining dgb but that might very well change again later.

Keep chugging all!

[ranastic]

Antivirus detected Win64/Coinminer.BB. does it harmful?

[Windozxpert]

I happen to find this miner extremely useful and profitable, probably the most profitable I have tried so far. Granted, I have only been doing this for about two years, but I've tried many different algo switching miners and with a few tweaks, this code is almost perfect.

Nemo, one thing I noticed is the Lyra benchmark is off. It shows a much higher hashrate than what is actually possible with the hardware provided. I think I saw this issue mentioned in past posts, and I though it to be fixed in the last release, but it's still off. This throws off everything, so I had to manually change the benchmarked values, or just disable Lyra.

Thanks for the work Nemo! We appreciate it.

[Windozxpert]

Antivirus detected Win64/Coinminer.BB. does it harmful?

Most AVs tag miners as viruses. There is no malicious code in these scripts, it is safe.

[Converse]

no... im no going to make it work off hour averages or 24hour averages   its works off the realtime stats...

i recommend letting benchmark complete, then edit start.bat and change (-Interval 90   to  - Interval 300)  thats how i get the best results out of it...
[/quote]

Please tell me the value of speed in the process are updated in the statistics and counted in the next interval the re-scan or just the initial test speed taken into account ?
What are the optimal settings for initial test interval ?
What settings are optimal then put obynochny to work program ?

[JaredKaragen]

FWIW to everyone:


Sometimes profit switching mining earned more.

When the market went into serious flux, mining one algo definitely earned more over time.

So with a grain of salt, do some tests under varying market conditions and form your own result as I have.


Each method has its place......  for varying reasons.

[repla64]

Skunk algo %10 - %20 reject last 2 days.  Does not validate cpu error.

[BitBustah]

So I use Nemos with the HashRefinery pool. I will install a second and third GPU next week. Is there something I need to modify/tweak or will Nemos just mine with all GPUs on thesame algo?
Thanks for the time folks. Appreciate it.

[Real_Unhooked]

So I use Nemos with the HashRefinery pool. I will install a second and third GPU next week. Is there something I need to modify/tweak or will Nemos just mine with all GPUs on thesame algo?
Thanks for the time folks. Appreciate it.

How did you manage to change the pool? Are there other pools compatible with this software?

[cinderz45]

Hiya Nemo/minerx117:

Thank you for your hard work. I want to know if this workaround will work for me.

I run 4 1080 Tis in my rig. I prefer Alexis 7.5 CUDA ccminer for many algorithms. We all know its limitations.

So, for fastest switching, I run one Nemosminer program for GPU 0 and 1, and another Nemosminer program for GPU 2 and 3. At first I thought it worked very well.

However, it messes up benchmarks when a Nemosminer reads the data from the wrong ccminer.

Please let me know if this would work:

After benchmarking each program SEPARATELY (very important, do NOT run both Nemos at same time to benchmark), I add this to all algorithm command lines in Miners folder:

Nemosminer 1: -b 127.0.0.1:4070

Nemosminer 2: -b 127.0.0.1:4071

Would this solve my problem? No more reading port 4068, so no more updating benchmarks? Benchmarks stay accurate, and fast switching means more profit?

[agapes]

how you choose GPU for mining?

[cinderz45]

how you choose GPU for mining?

it isn't easy, but...

you can add " -d 0,1,2" to algorithm command lines for the various versions of ccminer, each of which you have to set manually, each of which can be found in the "Miners" folder and edited via Powershell.

for EWBF, you need to set it as "--cuda-devices 0 1 2" with each integer being one of your devices, starting at 0

[agapes]

how you choose GPU for mining?

it isn't easy, but...

you can add " -d 0,1,2" to algorithm command lines for the various versions of ccminer, each of which you have to set manually, each of which can be found in the "Miners" folder and edited via Powershell.

for EWBF, you need to set it as "--cuda-devices 0 1 2" with each integer being one of your devices, starting at 0

oh man im really appreciate in last 3 weeks i asked 100 times no one answer me
it was really helpful , i wanted to disable my GTX 760 because had a problem with some algorithm and stopped my mining and i had to disable my graphic card to make mining work
THANK YOU!!