Author

Topic: [ANN][RIC] Riecoin: constellations POW *CPU* HARD FORK successful, world record - page 245. (Read 685216 times)

sr. member
Activity: 616
Merit: 321
LOL Huh? is this funny or what ??


are we all mining for others ??




{
"height" : 2284,
"bestblock" : "5985761e09c43d6692e5125dc89779d8a746778106ea0d5ed090027b1a8909d0",
"transactions" : 2120,
"txouts" : 2172,
"bytes_serialized" : 148284,
"hash_serialized" : "c9c648d7272b2c387c01e0bdab2fc9985fd3f72afdcf6625742cb21b2c65cf98",
"total_amount" : 71024.99999744
}


HuhHuhHuh
legendary
Activity: 1176
Merit: 1134

Riecoin has been listed on Newaltex.com (RIE/BTC)

https://www.newaltex.com/exchange/rie_btc

Help us increase the trading volume

poloniex would be a much better  small exchange to be on. they have google authenticator 2FA and Tristan the dev is top notch. You can email him [email protected] and ask to add RIE (or is it RIC?)
dga
hero member
Activity: 737
Merit: 511
So difficulty is currently at 774 bits or 234 decimal digits and I haven't found a block in over 24 hours. At this number size, expect about one (1) out of every 540 integers to be prime. Well, until we optimize the segmented sieve of Eratosthenes further for finding prime sextuplets, it is back to Primecoin mining for me. The miners released with this coin are a joke.

So many people have pointed out that unless you have massive amounts of processing power, this coin is not for you. It is a shame because a lot of people will be left out.

It is going to be interesting to see how a pool miner will be implemented - there are a lot of  2 - 4 chains, but very few 5 - 6 chains.

EECS in your .signature tag and not interested in doing the optimization? *grins*  If only the release happened during IAP.

The difficulty is high, but - solo mining is hard, period.  An optimized sieve won't change the difficulty of solo mining.

What the optimized sieve will do is level the playing field.  Right now there's a factor of 14x difference, at least, between the best private miners and the released one.  It's great fun for people who like to tweak the code, but probably not the best long-term thing for the coin.
newbie
Activity: 14
Merit: 0

Riecoin has been listed on Newaltex.com (RIE/BTC)

https://www.newaltex.com/exchange/rie_btc

Help us increase the trading volume


oh no, pls DONT ADD THIS COIN, we only prefer cryptsy or bter or mcxnow.
i'm very afraid of small exchange with ton of zero, it could kill the potential coin.

This makes no sense at all, please elaborate.
sr. member
Activity: 278
Merit: 250
So difficulty is currently at 774 bits or 234 decimal digits and I haven't found a block in over 24 hours. At this number size, expect about one (1) out of every 540 integers to be prime. Well, until we optimize the segmented sieve of Eratosthenes further for finding prime sextuplets, it is back to Primecoin mining for me. The miners released with this coin are a joke.

So many people have pointed out that unless you have massive amounts of processing power, this coin is not for you. It is a shame because a lot of people will be left out.

It is going to be interesting to see how a pool miner will be implemented - there are a lot of  2 - 4 chains, but very few 5 - 6 chains.
hero member
Activity: 781
Merit: 501

Riecoin has been listed on Newaltex.com (RIE/BTC)

https://www.newaltex.com/exchange/rie_btc

Help us increase the trading volume


oh no, pls DONT ADD THIS COIN, we only prefer cryptsy or bter or mcxnow.
i'm very afraid of small exchange with ton of zero, it could kill the potential coin.
sr. member
Activity: 604
Merit: 250
What value are u using for the --max-prime? Using --benchmark I got the best results (number of found blocks, not knumber/s) with --max-prime=100
Feel free to donate if this helps to you Smiley

RIC RT6Yu5tK3FNrW9m6vb4WUc6Wk5j3VPp6fR

I've been using 11000. Gives me stats like these on a slow duo core:

[2014-02-13 12:03:20] Expected days to block: 91.6 337.5 2-chains/hour
[2014-02-13 12:03:20] accepted: 0 of 0 in 31 seconds
[2014-02-13 12:03:29] thread 0: 914358273 hashes, 90 Mh/s
[2014-02-13 12:03:29] thread 1: 854589441 hashes, 90 Mh/s

I think if it is too low you will get a lot of false positives for primes.
newbie
Activity: 14
Merit: 0

Riecoin has been listed on Newaltex.com (RIE/BTC)

https://www.newaltex.com/exchange/rie_btc

Help us increase the trading volume
newbie
Activity: 4
Merit: 0
Run 48 cores since the lauch. 1 Block found 50 coins, i need to do somenthing or the coins send automatic to the wallet?.

Sorry for the bad english... thx  Grin
newbie
Activity: 47
Merit: 0
What value are u using for the --max-prime? Using --benchmark I got the best results (number of found blocks, not knumber/s) with --max-prime=100
Feel free to donate if this helps to you Smiley

RIC RT6Yu5tK3FNrW9m6vb4WUc6Wk5j3VPp6fR
sr. member
Activity: 616
Merit: 321
Current n-1 improvement (already mentioned elsewhere):

Run rminerd with a larger sieve.  12% improvement in going from 1Mb (default) to 2Mb on xeon class machines (i.e with large caches).  For example

Code:
./rminerd -q -o http://10.10.10.10:28332 -O awallet:apassword -t 8 -i 2097152 

Regards,

--
bsunau7

Next n-1 improvement is a small one (1-2% depending on how sieve size used) and was found before I focused on the time critical code:

Replace all

Code:
  if( mpz_probab_prime_p ( bnTarget, MR_TESTS) )

With

Code:
  if( mpz_millerrabin ( bnTarget, MR_TESTS) )

mpz_probab_prime_p is a wrapper around mpz_millerrabin which includes tests for small primes.  The sieve which riecoin uses has already sieved the numbers so no need to double sieve.

I would expect this to make more than a few % points on small memory systems.

Regards,

--
bsunau7

could you compile a windows version pls ?

I compiled a windows version of the fermat build if you want to give it a try:
https://www.dropbox.com/s/hz6ns6s85silqw0/cpuminer-rminerd-fermat.zip

So far it is the fastest variation I have found. It is based on this code:
https://github.com/iddumied/cpuminer-rminerd-fermat

ok i am testing now , thanks i let you know how it works out, btw now 95 mhash/s
sr. member
Activity: 604
Merit: 250
Current n-1 improvement (already mentioned elsewhere):

Run rminerd with a larger sieve.  12% improvement in going from 1Mb (default) to 2Mb on xeon class machines (i.e with large caches).  For example

Code:
./rminerd -q -o http://10.10.10.10:28332 -O awallet:apassword -t 8 -i 2097152 

Regards,

--
bsunau7

Next n-1 improvement is a small one (1-2% depending on how sieve size used) and was found before I focused on the time critical code:

Replace all

Code:
  if( mpz_probab_prime_p ( bnTarget, MR_TESTS) )

With

Code:
  if( mpz_millerrabin ( bnTarget, MR_TESTS) )

mpz_probab_prime_p is a wrapper around mpz_millerrabin which includes tests for small primes.  The sieve which riecoin uses has already sieved the numbers so no need to double sieve.

I would expect this to make more than a few % points on small memory systems.

Regards,

--
bsunau7

could you compile a windows version pls ?

I compiled a windows version of the fermat build if you want to give it a try:
https://www.dropbox.com/s/hz6ns6s85silqw0/cpuminer-rminerd-fermat.zip

So far it is the fastest variation I have found. It is based on this code:
https://github.com/iddumied/cpuminer-rminerd-fermat
member
Activity: 98
Merit: 10
You'll need more than $100 credit... there was someone with 120K cores on AWS, costing $1,200 an HOUR, and he wasn't finding much

this is what i'v been thinking lately :


    



Good you've only been thinking it , not writing it down , cause man , is this a blasphemy towards the English language.


well ......... people see what i ment with this right ?

By millionaire you mean $100 USD and an AWS account?

Edit: And I forgot, Amazon will give a free $100 evaluation credit.

Can you pm me where to get this credit? The power of 60 promo seems to be over.

I was thinking of the Powerof60 credit - I actually used that for "real" work (EMR) a long time ago, didnt know it had went away.
sr. member
Activity: 441
Merit: 250
You'll need more than $100 credit... there was someone with 120K cores on AWS, costing $1,200 an HOUR, and he wasn't finding much

this is what i'v been thinking lately :


    



Good you've only been thinking it , not writing it down , cause man , is this a blasphemy towards the English language.


well ......... people see what i ment with this right ?

By millionaire you mean $100 USD and an AWS account?

Edit: And I forgot, Amazon will give a free $100 evaluation credit.

Can you pm me where to get this credit? The power of 60 promo seems to be over.
member
Activity: 112
Merit: 10
has anyone found some coins here? i was mining 24 h and no luck

Yes, I found a 35 RIC block with my i7 4770 (before block 1152), since then, nothing.  Grin

do you mine with wallet or standalone miner?

With the official standalone miner (win64).
sr. member
Activity: 392
Merit: 250
Quote
[15:08] jh00: @coinhunter not really, but I want to launch Riecoin support before I go to bed in 14 hours

jh00 is the owner of ypool Smiley


He also said there is huge improvment possible in the Riecoin algo, up to 100x perf
legendary
Activity: 2912
Merit: 6403
Blackjack.fun
You'll need more than $100 credit... there was someone with 120K cores on AWS, costing $1,200 an HOUR, and he wasn't finding much


Yeah , I heard that NSA was trying to mine this with 250 billions cores but didn't manage to find any block.
newbie
Activity: 47
Merit: 0
Because of the variance in success and some quirks of the search, here's a patch to let people have a little more visibility into how the miner is doing:

http://www.cs.cmu.edu/~dga/rie_log.patch

To use, download the miner source and
  cd (your-source-directory)
  patch < rie_log.patch

You should see log output whenever the miner finds a chain of length 2 or longer.

Look for roughly a 10x difference in length 2, length 3, ..., blocks in your output log over time.  If you're *not* seeing any length N chains -- because, e.g., you've broken something during optimizing -- then you know something is wrong.

I've found that the rate of generating length 2 or 3 chains is a good backup benchmark to use.

Gatra, please feel free to have full ownership of the patch - I release any copyright claims to it - if you want to include it or something like it in the default miner.

  -Dave

How to redirect output to the file? standard > doesn't work for me
hero member
Activity: 583
Merit: 505
CTO @ Flixxo, Riecoin dev
Because of the variance in success and some quirks of the search, here's a patch to let people have a little more visibility into how the miner is doing:

http://www.cs.cmu.edu/~dga/rie_log.patch

To use, download the miner source and
  cd (your-source-directory)
  patch < rie_log.patch

You should see log output whenever the miner finds a chain of length 2 or longer.

Look for roughly a 10x difference in length 2, length 3, ..., blocks in your output log over time.  If you're *not* seeing any length N chains -- because, e.g., you've broken something during optimizing -- then you know something is wrong.

I've found that the rate of generating length 2 or 3 chains is a good backup benchmark to use.

Gatra, please feel free to have full ownership of the patch - I release any copyright claims to it - if you want to include it or something like it in the default miner.

  -Dave

nice, will take look at it. Have in mind that the 10x may vary with difficulty.
hero member
Activity: 583
Merit: 505
CTO @ Flixxo, Riecoin dev
thank you all offering servers, I'll get back to you when I have some code to share

Could you generate / manage addresses for donations?  Not for silly crap like spamming Cryptsy, but for code, pools, etc.

I'd be willing to throw some RIC and/or BTC into the fray.

Donations are always welcome. The addresses are the ones listed in riecoin.org:

RIC: RByJXMhtRa2Jc2ix6sWoVRZq3kyK3zb8nY
BTC: 1Ud6xgvXwKGksbguVgke8UTbJ9sYr9AuH

I do code, and I'll maintain the client, miners and give support to pools and p2pool. I'm against paying to get into exchanges.

The cpuminer works fine, no errors on scanhash_riecoin, but I realize some optimizations could be done. I'll look into that tomorrow but don't expect huge improvement.

Thank you.
Jump to: