@BChydro
Are you new to mining ? These are khash per thread.
Of course they don't vary with the number of threads ...
If you're using more threads than number of cores your cpu has, mining speed per thread should decrease. That instance has 32 cores, mining speed should max out at 32 threads. But the client reports the exact same speed per thread, even when I stuff 32 threads per core (1024 threads) That makes no sense. That happens only in the first round when threads are initialized and tested. Leave it running 10 seconds and you'll see that if you have more threads than cores and you keep adding threads then each one starts going slower and the total mining speed does not improve. |
|
|
|
|
Please don't fall for the virus nonsense.
The external miners trigger some antivirus false positives because miners are sometimes included in trojans used for botnets.
The windows wallet was cross-compiled from linux, and I verified that the .exe files downloaded from sourceforge are exactly the same that were generated by my ubuntu. So unless it's a new kind of linux/windows supervirus then it's clean.
If you have your doubts, compile it yourself. That's the beauty of gitian: you can verify the binaries.
|
|
|
|
updates: - while riecoin.org is down please see riecoin.biz- official windows wallet not published yet, will be there in some more minutes - if you execute 'getwork', there is a field called "diff". That is the length of the primes in bits. Starting difficulty was 304 and in changed to 395 in block 289. the difficulty reported by getinfo and getmininginfo seems to be always 1, a bug I'll fix in the next release 395/304 = 1.299 The value in getinfo and getmining info is reported as 1 but it really is almost 1.3 thanks for your patience and don't give up! |
|
|
|
|
please hold on, launching in a few minutes
|
|
|
|
I have a question:
Is there no reward for the first 24hr mining (the first 576blocks)?
If so, who is willing to mine in the first day ??
Am I wrong with sth?
Frist 576 blocks without reward, that's correct. However it's impossible to know how long will that take. Once the difficulty adjusment and the computing power stabilize then 576 blocks should be about 24 hs, but the first blocks will certainly be found in much less than 24hs. It's impossible to tell how long, because it depends on the computing power of the whole network, and little bit on luck too. I'll be mining from the start with many PCs and you are encouraged to do that too because: - You can set up your miners and go do something else instead of being constantly watching the number of blocks in order to see if it got to 576 or not - It will make the launch more fair - It will probably be only a few hours, not 24, at least 1 or 2, and definitely less than 10. For reference, at minimum difficulty it takes about 5 hours for my i5 laptop to find a block. But I expect many VPSs to join, so everybody can do their guesstimates... so close now... |
|
|
|
|
There's a countdown at riecoin.org, it's less than 14 hours from now.
I'll try to put up a mini mining guide, but it's basically the same as if you wanted to solo mine with CPU with any other coin.
there will probably be no pools, so the external miner should be pointed at the computer that runs the wallet - generally the same one so it's localhost or 127.0.0.1 - using the user and passwords that you choose and set in the file riecoin.conf
Probably YAC should have a very similar guide.
|
|
|
|
I didn't try it this way, but if you remove the asserts for the genesis block in InitBlockIndex() and the obvious "assert( 0 );" then you should be able to generate the keys. Sydney? I'm in GMT -3, so we must be in almost opposite time zones... yes, 14 hour difference...  |
|
|
|
is this just a failur because the coin hasn't started yet?
yes! it's because full source code will be released at launch time. the good news is you were able to compile/link, so you have everything correctly set up. At launch time, just update your source files from the repository, recompile, and you're ready to go |
|
|
|
I hope compiled versions for windows will be available at launch.
Yes, they'll be available. I'm having problems with some antivirus detecting it as malware because it's based on code from the cpuminer that may have been used for botnets (I even deleted the sha and scrypt mining code but they still complain). But of course it's a false positive and executables will be available. |
|
|
|
Excited for this project! Any ballpark estimates for memory throughput requirements? How large of a sieve is used?
In the external miner, sieve size can be configured via command-line parameters but the default is 1Mb (per thread). I believe best results are achieved when the sieve fits in the processor's cache. But this will depend on your hardware... |
|
|
|
That's interesting. But since we don't actually know whether there are infinitely many sextuplets (and certainly can't assume such sextuplets exist in a given range of values), doesn't that mean that it's possible to hit a difficulty with which no block will be possible to find?
It's not proven, but sextuplets are believed to be infinite. If we ran out of sextuplets I would be very happy because Riecoin would have served it's purpose of helping science! It would be an important result for mathematicians. And the coin wouldn't die because we could switch (hard fork) to prime septuplets or octuplets, etc. We can't be sure such sextuplets exist, but it is conjectured that distance between sextuplets is O( (log p)^6 ) and the max distance between them is O( (log p)^7 ). An interesting article on this can be found here: http://arxiv.org/abs/1301.2242This max distance (believed true but not proven) adds an interesting feature: in theory if you are very unlucky you could do sha256 hashes in Bitcoin style for years and still produce no blocks. But this maximum indicates that if you mine long enough then you will find a block: as long as k*(log p)^7 fits in the nonce filed, you could assume that there exists a nonce that produces a block. Bitcoin does not have a theoretic background that allows you to claim that. We do. It's only conjecture and not proven, but lots of mathematicians believe it to be true. I admit this is not very useful because not only you need to find it, you also need to find it before 2.5 minutes. But being (almost) sure that the nonce exists feels good. Also, with pooled mining, since finding a quintuplet is O(logp) easier, that'd mean share value would need to adjust with the difficulty (since greater diff - bigger p - easier to find a quintuplet compared to a sextuplet). Which would be quite a nuisance (though I think XPM does that, it's still subject to readjustments on, for example, YPool).
It's true that share value has to be normalized for difficulty. But that's not a problem, since we have a formula for that. Moreover, the relative value of sextuplets to quintuplets, quadruplets or whatever the pool decides to accept, can be computed too using Hardy Littlewood k-tuple conjectures. Ypool based relative scoring of XPM shares of different chain lengths based only on trial and error. |
|
|
|
gatra which libraries are going to be needed for compiling? The same as bitcoin or do we need something else too?
Will we be able to solo mine with the client or are we going to need an external miner?
Whether the cpu-miner (not wallet-miner) will be available during the launch?
The client/daemon require the same libraries as Bitcoin. The qt client has an embedded miner, but it's slow, the external miner is recommended. The external miner "cpuminer" (executable file is "rminerd" to differentiate from the original "minerd") requires the same libraries as Pooler's cpuminer (libcurl) plus gmp. To compila, you will need autotools too. |
|
|
|
I just translated the launch post into Chinese as below: https://bitcointalk.org/index.php?topic=454135.0By the way, I registered this ID for protective purpose, and the domain http://riecoin.net too. I will only use them to promote riecoin in China, and will be pleased to return them if necessary. Finally, have been looking forward to this for quite a few months, and wish riecoin stand out. Thank you for your effort! Promotion is welcome! I only ask you to please make it clear that users "riecoin" and "gatra" are 2 different persons. I hold riecoin.us, riecoin.com, riecoin.biz and riecoin.org. I'm not responsible for others. Thanks and regards, gatra |
|
|
|
|
The explanation of Riecoin's proof-of-work algorithm is now published in riecoin.org
So this is the secret: prime sextuples. Each block represents six prime numbers of the form p, p+4, p+6, p+10, p+12, p+16.
If you expected just finding prime numbers, one per block, and you fill cheated, please try to understand this:
It's not practical to just look for prime numbers, because verification would be almost as hard as mining. So there has to be a catch. Something to make finding more difficult than verifying.
There are many ways to do this, for example requiring that the primes follow some arbitrary rule like they should be k1 mod k2. Or spell my name backwards, or whatever, but this would waste resources arbitrarily discarding primes that don't fit, and it doesn't make the accepted primes more interesting mathematically speaking.
Requiring sextuplets make it harder to mine while still keeping interesting properties regarding the distribution of prime numbers.
About the Riemann hypothesis:
I don't expect to prove it or disprove it. I'm a computer scientist, not a mathematician, so I want to be very careful about my claims... otherwise mathematicians will come and hunt (haunt?) me down.
I might be wrong but I understood that the Riemann hypothesis implies a tighter bound on the error term of the prime-counting function than the Prime Number Theorem. It is possible that some mathematician could use Riecoin blockchain data in order to draw some conclusions about the Riemann hypothesis. But I don't expect that to really happen, and it's believed to be true anyway...
Maybe I should have named it "Hardy-Littlewood k-tuple conjecture-coin" since we are directly testing it... but it's not catchy... maybe LittleHardWoodCoin... oh, well, too late now...
|
|
|
|
Congratulations on completing the project. thanks! but I consider it far from complete... so much left to do after launch! sorry. GPU or CPU?
CPU only at launch. Maybe, eventually, GPU. |
|
|
|
Difficulty adjustements are every 576 block or every 288 block?
Because the ANN is contradicting
It was 576 but I thought it was too much so I reduced it to 288. Thanks for noticing, I'll check the ANN |
|
|
|
|
Show Posts
