Bitcoin Forum

Due to a coming GPU miner, this bounty has now been withdrawn!


Hello! I'm offering up 2.5BTC of my own money to the first person who releases a standalone CPU miner for primecoin that performs as well if not better than the primecoin implementation already in the client. Irritant has also pledged 1 BTC.

FAQ: "Why is ypool not getting this bounty?"
A: Does not run native on Linux (requires wine).
Requirements:
Compiled windows binary
Compiled linux binary (or put it in an easy-to-use git repository, we want to make this available to everyone)
Published source code (github preferred!)

If you want to add to the bounty, just respond in this thread. I'm not going to collect for the bounty, we can just send it to the winner when they win.


Reasons for a CPU-standalone miner:
Pooling computing power to leverage against luck
Ability to submit shares to pools, 'guarantees' income (smooths out the bumps, etc.)
Not all miners need a full blockchain and client, etc.

Cheers!

Bump.

Basically. As long as it supports shares and would work with pools, that would be golden. And this bounty has been here all night, I thought it'd be done sooner xD.

Right now, there is no established pooling protocol to follow.

Hey guys, I've been using ScryptMiner GUI to solo mine primecoins. It can submit shares and work with pools too. 8)

Do you mind sharing how you have managed to do that?

How on earth?

bullshit

seconded.

Pools?...  ???

How would a pool even do shares with a PoW like this?

what is bullshit? I simply meant the program was designed to work with pools and submit shares, I didn't realize that primecoin was not designed to use with pools. I still have been solomining using scryptminer which i believed was the initial intent of OP.

So you're solo mining using scrypt for primecoin?

Yeah, that's the tricky part. Perhaps a proof-of-not-prime of some sort? Seems like a high-bandwidth option...

Pool support isn't up to the client/mining software, though, the pool software would need to be modified to support this. Having a standalone cpuminer fork would be a step in the right direction.

Pool share can be implemented as lower difficulty prime chains, similar to hashcash proof-of-work I think.

I guess so, I didn't even check last night to see if it was scrypt or not(i've had a terrible cold the past week). My primecoin wallet says i've found 5 blocks already, and they are currently being confirmed, only my first block is at halfway to being fully confirmed. I was just in a rush to get mining last night and figured to use scryptminer for my CPU because it was the only standalone CPU miner i have. my cpu usage is at 100% on 6 cores and has mined for the past 12 hours. I found my first block almost instantly when i launched scryptminer using a solo mining config.

So is primecoin a scrypt coin? I am terribly confused by all of this now. ???

quoting for posterity.

just changing one of the loops to optimise the code a bit, or changing from -O2 to -O3 will satisfy the three requirements specified in your bounty.

Will

No, primecoin is not a scrypt coin, you are a cunt liar thou.

Welcome to negative reputation baby.

You have found 5 blocks using a scrypt miner for primecoin?

There are two options here:

1. You are a liar.
2. You are quite literally, and I mean this in the nicest way possible, dumb as rocks.

So which is it?

I do not understand all the harsh negativity, I guess I am simply clueless of primecoin however, in my primecoin wallet I have found 5 blocks and the only miner I have used is the scryptminer. I simply figured it was the reason why I am getting primecoins. I have no idea how else I am getting them then.

Geez, I wish there were less assholes in this community

Perhaps... and only perhaps, you made a primecoin.conf file and set in this file autostart mining configuration with your wallet, without write anything in debug window.

After you start cgminer and thought you were mining with cgminer instead with wallet...

It is a possible explanation...

If not... I don´t know what did you done

Do you realize that nothing that you say makes sense? It would be like going into a spanish class and speaking german the whole time and then wondering why people are giving you a hard time.

Does "setgenerate true -1" mean anything to you? If it does, there is your explanation. Or do you just randomly copy-paste code without knowing what it does or why you're doing it? And if you are simultaneously scryptmining and unknowingly generating peercoins because of the aforementioned code, you are incredibly lucky, and living-proof that luck always trumps smarts.

That was it!
thanks a bunch, sorry if I'm a noob guys, but it doesn't give you all reason to be asshats.

You guys can have your fun and call me an idiot all you want, I really don't care.

Um, primecoin isn't a scrypt coin. its a prime number calculating algo. completely new. you didn't set an auto start in your conf. you are  liar. they called you on your bullshit.

Who cares. Until Primecoin gets out of the 'basement dweller coin' status and has actual miners that 90% of people can use without 'hacking together' something, its just a nichecoin with no value.

just for the record.  cgminer --scrypt will connect to a primecoin-qt client and look like it is mining.. cgminer will report kh numbers for each gpu.... and it will report new blocks on the network in the console.....but it will not find blocks for primes... and i do not think conman will add the support either!

What I'm saying is (and correct me if I'm wrong) you can set in primecoin.conf  an autostart mining (if you put gen=1 in primecoin.conf file then client will start mining automatically after start) so, when you start your wallet, it start mining automatically... If it is wrong so I don't know what Lloydimiller4 had done, but I prefer to think one person is innocent before launch my war dogs...  ;)

Hint, don't make claims you don't know anything about and we could have avoided this problem in the first place! Ignorance is not an excuse when you are proactively sharing bad information as fact. A proper post would have been:


As opposed to:

+1  ;D

Thank you, in the future I will try to not give advice when I am ignorant about the topic, but in this case I wasn't sure that I was ignorant until vingaard pointed it out.
OP, sorry for sidetracking your thread, I will stop posting in here unless its on topic.

this gui made me lol

Crossposting for exposure;

If anyone would like to give it a shot, there's an extremely optimized sieve of Eratosthenes implementation here:
https://primesieve.googlecode.com

From the software:


From the source, the sieve size is small (1 * 10^6) by default if I'm interpreting it correctly.

I'm wondering if a massively parallel GPU implementation with higher memory bandwidth also benefiting from the reduced memory requirements will see much better performance.

Also,

If the current implementation is the standard C++ one (I haven't looked at it thoroughly but it seems to be), whoever can race to implement this first may benefit immensely.

For those interested, there's a CUDA implementation here too: http://www.mersenneforum.org/showthread.php?t=11900

Any progress on this? I may be willing to add to the bounty if someone is actually at work on a miner...

I'm not loaded by any means but I offer up 2 LTC for whoever creates an improved CPU miner (the first miner to be better than QT)--and 5 LTC for whoever creates the first GPU miner.

Bump.

I would be willing to kick in some LTC/BTC for a CUDA implementation. There are already projects that has proven that CUDA is apfar superior at Mersenne solving over general processing power and CUDA implementations of Mersenne solving has already been used to verify for example the 14th Mersenne solved.

Of course, being a natural born Cynic, as always there will be people who do this on their own (or already has) and will have an order of magnitude advantage in mining over the "general public" rather then sharing it with the forum.

Ill joining that one, and i offer 1BTC with the keyword CUDA in it. Got a 580GTX that need work.

Posted a hint on cudaminer author post. https://bitcointalk.org/index.php?topic=167229.940

this.

(assuming in your favor that you're not a big-mouthed liar)

Onkel Paul

Would a mod be interested (Salty? :)) in doing escrow for this bounty pool?

Maybe I should make a thread with a bounty for a CUDA miner as well? :)

I offer 1 XPM for the first CUDA Miner of Primecoin.

I also have a GTX580 that would like some work other than scrypt in cudaminer and oclvanitygen.

bump

The code works with native C++ integers and not CBigNum.  Not sure if CBigNumber is needed for Primecoin.  How big are the primes anyway?

[
The code works with native C++ integers and not CBigNum.  Not sure if CBigNumber is needed for Primecoin.  How big are the primes anyway?
[/quote]

they are between 83digits and 268digits (decimal number) so far.

BFGMiner's "prime" branch seems to work, at 2000pps for me (on a system that primecoind was doing well under 100pps on, unoptimized).

It's not user friendly, and not intended to be easy to use, as it's still in development.
I'm not sure if I'm going to finish it, as I'm pretty sure Primecoin is a scamcoin - but maybe if others want to join in the development we can share the bounty? ;)
Basically it mostly just needs cleanup at this point.

What would designate it as a scamcoin? 

it's not bitcoin?  ;D

Care to explain?

There's no reason for people to adopt it.
If the prime POW proves to be demonstratably better than SHA256d, it could be adopted by Bitcoin with a hardfork.
Things like this (enhancement to a specific technical detail) should be done as testnet branches - as proposed changes to Bitcoin.

Without a hope of adoption, Bitcoin and forks function in a pyramid: when people get tired of it, the ones stuck with the coins in the end lose out.
Bitcoin is different because its innovation makes it possible to achieve a status quo where everyone wins (by having a usable decentralized currency).
Any new system would need to have a viable way to compete with Bitcoin (ie, something Bitcoin couldn't just adopt as an improvement).

Examples of altcoins that don't function as scams:

• Tonal Bitcoin: Compatible with BTC, uses the same blockchain. Whether it succeeds or fails, nobody loses value so long as Bitcoin as a whole doesn't fail.
• Namecoin: Not a currency, but a domain name system.
• PPCoin: Uses proof-of-stake instead of proof-of-work; arguably this is so fundamentally different that Bitcoin could not adopt it. Note that last I heard, PPCoin was being used as a scamcoin despite this, however, and there were some major problems with centralization/security tradeoffs.
• Freicoin: Features demurrage, which trades the "why spend it?" problem for a new "why acquire it?" problem. It would violate Bitcoin's social contract to make an economic change like this.

The only reason Bitcoin would hardfork is the security of SHA256. There is no "better" in any other coin which would make Bitcoin switch to any other solution, so your criteria which is scamcoin and which is not is completely wrong.

I think you just proved my point here.

IMHO, bitcoin can adopt most of the features that based on bitcoin code, that is for sure.
However bitcoin is too big to make any decision to adapt new changes lightly without causing issues/mass panic.
any changes to bitcoin would affect tons of investors (eg, asic miners, manufacturers).

By the time bitcoin realized certain features (let say primecoin's POW ) is worthy to be implemented, the coin might already gaining too much momentum to be even stopped or slowed by bitcoin.

 

Primecoin is at least doing something that has use outside of itself.

I didn't say there is no logic in much of you've said, on the contrary, your explanation why other coins fail is interesting. I only think you underestimate the inertia of Bitcoin fundamentals, it *may* fork in a few decades if SHA256 is undermined, and it may switch to some new hash function but everything else will stay the same.

Also, you are underestimating the new fundamental advantage of Primecoin: it will be very hard to implement it in GPU to be faster than CPU, unlike other coins. The advantage to mine with only CPU, gives many people a chance to mine, taken away from them by specialized GPU mining monsters.

If Bitcoin had to change the PoW, that would probably end up killing it.

By the time a reason comes to change the PoW, the amount of money invested in double SHA256 silicon would be much greater than it is today, and changing it would alienate the current mining community due to how expensive it would be to actually mine.

That's a bad thing.

For starters, mining exists to serve the users, not vice-versa.
Cryptocurrencies aimed at appealing to miners are fail by default.

Insofar as POW works in general... you cannot stop ASICs, period.
It simply is impossible.
With SHA256d, the availability of GPUs and FPGAs made it so the leap from CPUs to ASICs was bearable, and did not compromise Bitcoin's security.
Without the step of GPUs and FPGAs, this leap is almost certainly to be deadly to any cryptocurrency.
In the meantime, before ASICs make sense financially, you have a system that gives criminals (computer crackers) the ultimate authority.
So, to conclude, GPU-resistant is very bad for proof-of-work systems.

I disagree, it's not the point of stopping ASICs, they are the best thing happened to Bitcoin because they are ultimate protection from 51% attack. But there is no reason to develop ASICs for alter coins, regular people need access to cryptocoins for large adoption. People love lottery. You can't have global economy without very large adoption of currency, so Primecoin may actually be very, very good thing. There is a reason for such avalanche adoption comparing to other altcoins.

I didn't claim that XPM is GPU-resistant, to my limited knowledge it should be, but it may not be the fact. Certainly will be for some time, giving it a chance to be really broadly adopted.

luke, your words:
now think about differences in hashing sha, and calculating the primes.. are you realy realy trying to come out that ignorant idiot?

I don't follow - why the name calling?

so the bounty remains uncollected?

I'm not sure this is the case, or at least it is not as simple.  With hashcash proof-of-work it is impossible to look for lower difficulty shares without also looking for higher difficulty shares.  In Primecoin, on the other hand, as I understand it, one could look for chains of length 7 and find them with much greater frequency than they would find chains of length 7 while looking for chains of length 8 (i.e. pool miners would maximize their share submission by hurting the pool; the tragedy of the commons ensues).

I assume that when the mining algorithm executes it first executes the Sieve of Eratosthenes to build a list of possible primes.  If one finds that there is a list of 7 numbers that passed the sieve and form a chain then they could be checked to see if they form a valid share, even if the sieve eliminated the next value, proving that a block of difficulty 8 or higher is impossible from that start (I am assuming a share difficulty of 7 and a network difficulty of 8 or more).  A miner optimized for finding valid blocks as fast as possible would save computational time by ignoring the chain of length 7 when the difficulty is 8 or higher, while a miner optimized for finding valid shares would check every chain of primes that passes the sieve that is at least (share length) long.

This could be circumvented by requiring the numbers after the share's chain up to the integral network difficulty to all pass a sieve, but I believe that that would break the requirement that shares be fast to verify by the pool host.  Additionally, it would set stringent requirements on how the numbers would have to be sieved which would limit improvements to be made in that area (which seems to be where most of the improvements are being made).

You've made a really innovative coin, Sunny, and I trust you to come up with an innovative solution to this, but it isn't as simple as it may appear at first glance.

My p2pclient will beat this problem. By making a sub blockchain with difficulty one less than the main chain's, payment is based on all the blocks you have submitted during the round. The P2P chain will end up producing many many more primes than the main chain does, maybe somebody will take it upon themselves to create a listening node and just store them, since I won't be making a site for the pool it won't be me, which also means I can make it a no-fee pool.

I found a Prime generator for CUDA from Nvidia website, and I am wondering whether it is helpful for our Primecoin mining.

https://devtalk.nvidia.com/default/topic/493679/prime-generator-for-cuda/

If people want me to I will, I'm a fan of Primecoins, and would like to see further development.

Unless I'm mistaken, the source code links on that page are dead ends.  Also, unless I'm mistaken, it is using a different algorithm and is running on smaller numbers.  A useful thing to find would be a CUDA (or, ideally, an OpenCL, which runs on both nVidia and AMD) program that performs fast modular exponentiation of large numbers (i.e. larger than 64 bits; the numbers needed are on the order of 256 bits or larger).  Modular exponentiation is at the heart of both the Fermat Primality Test and the various forms of the Euler Lagrange Lichfitz Test.  It would also be good to find a sieving method implemented in a GPU-specific language, as that is the other math being done in the search miners are doing.

It looks like the linked page implements (or, would have implemented, if the links weren't broken) the Ulam Spiral, which is a method of finding lots of primes and is a strong proof of primality (as opposed to the Fermat Test, which only shows probable primality).  It, like the Sieve of Eratosthenes used in the generation algorithm right now, can be used to calculate every prime number up to a certain number.  If it can be stopped early and still give useful information about a set of numbers then this sieve is applicable to Primecoin (i.e. if you can run some number of iterations of the Ulam Spiral where you only care about a very sparse subset of numbers and when you're done it has eliminated a lot of the composite numbers).  However, without source code it isn't really a big help to developers right now.

This is spot on.   Both luke-jr and I have both had success replacing Openssl's default big number exponentiation function with the one from GMP.  I recommend perhaps people look there :)

That's enough hints for now!

Will

Far from enough hints  :P

People from Tsukuba University already implemented GMP on CUDA:
http://www.hpcs.cs.tsukuba.ac.jp/~nakayama/cump/index.php?The%20CUDA%20Multiple%20Precision%20Arithmetic%20Library

If someone wants to roll his own implementation here's where he can start:
http://individual.utoronto.ca/haojunliu/courses/ECE1724_Report.pdf

If somebody already implemented primecoin client on GPU he should not have that advantage for too long or he can kill a coin mining. From some comparisons I've seen advantage can be even bigger than in BTC case.

CUMP only implements addition, subtraction and multiplication... doesn't implement mpz_powm which is needed for the primality tests.  There's a ton of efficiencies that can be obtained just from optimizing the existing code, and without even trying to re-implement modular exponentialization in CUDA.

Will

How does it beat the problem? I think having difficulty one less does not solve it: a miner optimized for the sub blockchain could generate valid blocks which would not pass the sieve if they were working on the main chain. This means they get payed for work easier that what they should be doing, "stealing" from honest miners in the same pool. I guess that if all miners did this "cheat" the problem wouldn't be that serious, however lot's of hardware would be doing "useless" work, something that we were trying to avoid. this gets worse as the difference between the main chain's difficulty and the pool's difficulty gets higher.

I add 1 BTC to this bounty
'pledge' = 1 BTC

Forgot about this thread, xD!

Anyhow, ypool, from what I understand, does have a stand-alone miner. However, since their code won't work as-is on linux, and many people are doubting their authenticity, I am holding the coinage for now. However, this is a community bounty, so if the majority of people want me to send the coins to ypool, I'm cool with that. :)

i dont know if it performs same or better..

Another valid point :)

https://github.com/jh000/jhPrimeminer

1Jh1ejB8RDJowzW2fuqyjGzqHAjQ5fXEyb

something i found :P

you'd have tobe the creator to claim it  ::)
If anything it should go to mikaelh for all his work on the high performace mods he's made too the orginal source

also there are these, just came out like yesterday

v0.2.1.1 - 7/18/2013
freeze bug fixed
added slight performance tweaks
http://test.dsync.net/jhPrimeminer-0.2.1.1.zip - x64 build

http://test.dsync.net/jhPrimeminer-0.2.1.1_x86.zip - x86 build

and of course the creator(s) should be paid, but a little bit my way for helping you guys sounds fair too doesn't it :D

in the Op:


So you want a little bit your way for a miner we already knew and didn't contribute anything too?  :-\

When I put up the bounty I was more expecting BFGminer, native linux support would be awesome.

apologies, just trying to help :P

Is this up yet? I'll create a poolers CPU miner fork for Primecoin if it is