Topic: [XPM] [ANN] Primecoin Release - First Scientific Computing Cryptocurrency - page 63. (Read 688812 times)

1 x Parallella = 32 GFlops - $99 -  0.32 $/GFlop
1 x i5 4670K = 86 GFlops - $240 - 0.35 $/GFlop
And this is overlooking the fact that you would also need to buy the rest of the computer (case, memory, hard drive, etc).

So yes, i think it's quite an efficient processor for its price.  But would you mine a whole lot at 1 Parallela? probably not.  
You would need to get at least 10 or so.  They sell them in packs of 4 with all the extra cables required for $575.
http://shop.adapteva.com/collections/parallella/products/parallella-cluster-kit

Estimated shipping date of October.  By which time i don't know if mining primecoin with a CPU will really be a thing anymore.  But it might be a nice rig to have to mine other things that come along.

If you bought 2 of those 4 packs, you would have 256GFlops for $1150....

@Sunny: Why not merge your code with the high-performance version of mikaelh?

You stated that you can't do this due to licensing problems with libgmp, right? I think that that you are wrong: libgmp is LGPL, not GPL, and can therefore be used even in closed-source projects provided that you link to it only dynamically. Which primecoind does, so you would be fine even if primecoin would be closed-source.

(this is kind of a double-post as I am trying to speak to both developers at once..)

Exactly. The main reason we see Bitcoin ASIC miners is that every block is worth $2500. Or, the best protection from ASIC mining is low valuation if you want to put it that way.

If I had to venture a guess which one is the more complex circuit, scrypt or finding Cunningham prime chains, I would guess the latter. The sieve is also quite memory heavy. That does not directly translate to an ASIC performance gap of course, it's just an observation.

yes.
but the number looks low, maybe you need to update your miner if you are not using the latest of this https://bitcointalksearch.org/topic/xpm-ann-primecoin-high-performance-hp14-released-255782

I wonder what kind of PPS a Parallella can achieve. At $99 I'm about to buy one.

http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone

Bravo

I have extensive knowledge of prime number theory and how modern computers can be used to find them but this is not going to 'unlock' more efficient methods of mining the coins. First mining involves more than finding primes. Second and more important in an imaginary world where we had asic chips mining prime coin as the difficulty level shot into the sky the awards of coins per block diminish very low. The design of the coin is such that it yields 'less' the 'more' you throw at it and we would all be in the exact same place we are now.

Personally I think the beta testers of the coin may have perfected some 'better' builds to mine than we have available in public now. I have no proof of this and anyone who helped make this source has a right to use their knowledge to their advantage. We all have the right to not mine the coin.

Until prime coin has some regular uses other than just mining and selling on an exchange it will remain a commodity only. I admire the development team for it's creation at a time when a new coin is launched every other day.  Prime coin is popular now because it is unique and while the long term scientific value to be expected from the project is low it is a great step in a direction to design a coin that does something more than drive up electricity use. My advice is to just enjoy it as it is now. Give it a year or two to develop. I suspect if any 'advanced' methods' of mining prime coin are developed they will not be posted for everyone to use. Anyone that gets a powerful edge is most likely to keep it quiet and mine for their own benefit.

Bitcoin was designed to be a form of currency beyond traditional corporate and government control for the better of all but a majority of posts I read give me the impression most are mining to just make a quick buck. The current money system of the world revolves around greed. The true long term benefit to everyone would be a permanent change in how money changes hands in the world and negation of any entity to just 'print more bills' like all nations do now. I may be a minority opinion here but I think this result if achieved would be a much bigger win for everyone than just a few of us making some money in the short run.

So can we relax about GPU, FPGA, ASIC mining and think of the big picture?

What's the latest popular speculation on coin exchange value? 

Just did that and it worked. Thanks

I had this problem too, it seemed to come and go away by itself. It was happening while I restarting the client a lot while testing different sievesizes. Adding many nodes did not help. Have you tried just leaving it running for an hour so until it can find some connections, that's what I ended up doing?

Another entirely plausible possibility is that windows firewall is blocking it. You need to allow it for each different version you run because the file you are executing is probably in a different folder for each version, so windows firewall considers it new/unknown. Windows firewall on the first run will ask if you want to unblock or keep blocking. Its possible you either clicked the wrong button, or your firewall did not notify you, or you have some other software firewall that is blocking it.

I just came back from vacation and sadly my miner is 8 days behind . I downloaded hp7 but I can not connect to the network. Any help?

I have restarted windows and the client. Also tried to add a couple of nodes but that does not seem to be working.

You need someone who knows all three. Those people are rare overall, not to mention here.

I wouldn't be so sure that is actually true. I'm far from an expert, but modern process technology is remarkable. You get a lot of silicon in a modern chip. Of course you could design something that brute forces scrypt faster than a GPU if you had the money. Would the performance difference be as great as with SHA256? Probably not. Would it be a greater difference than if we were calculating primes instead? Probably.

Why wouldn't it be? It is after all possible on a GPU, and designing a custom circuit for a workload is much much easier (albeit a lot more expensive!) than implementing it on a GPU (which is specialized for a very specific task, insanely parallell vector calculations).


If course it is possible, I don't think anyone disputes that. But the performance difference between CPU and GPU is not going to be anything near what it is for hashing.

Had not seen that thread. Good luck to 'em.

We need a hyper geek who has:

* Designed processors
* Knows how SHA-256 works on a machine level
* Knows how prime number searching works and how to implement that at the machine level

They should be able to say for sure.

http://en.wikipedia.org/wiki/Cunningham_chain
http://en.wikipedia.org/wiki/Generating_primes

My brain just hurts looking at these pages and the links that follow.  

So you doubt that mtrlt can do the trick? -> https://bitcointalksearch.org/topic/xpm-working-on-a-gpu-miner-for-primecoin-258540
And another step into the direction GPU miner was taken here: https://bitcointalksearch.org/topic/xpm-cuda-enabled-qt-client-miner-for-primecoins-source-code-inside-wip-258982
There might be no GPU miners atm, but I doubt it will take long. Primecoin is suprisingly attractive. Even a pool is available although it had to be built almost from the scratch due to the completely new Proof-of-Work approach.
I suggest we'll have a GPU miner that outperfporms the current CPU miners for Primecoin (and that is more energy efficient) before we have FPGAs for Litecoin (that are more energy efficient than GPU miners)

This question comes from a complete non-computer engineer, but I'm curious. Since ASIC is designed to accomplish a specific task, wouldn't it be possible to create an ASIC to "mine" prime numbers? [This isn't a question about whether it would be a marketable device like the ASIC USB miners, just about the engineering side.]

Before I knew about cryptocoins, I ran a lot of GPU's for BOINC distrubuted computing projects. There are/were(?) heaps of CPU based Prime number finding projects but none of them ever used GPU, so I am agreeing that branching is not suitable for the hardware and that is why no one ever bothered to make a program to mine on them. For this reason I suspect we might never see a GPU or ASIC prime number miner.

Primecoin mining is not another scrypt.

Scrypt is just another hashing algorithm, and hashing algorithms are generally the best candidates for GPU to do. Because they tend to have no branching, and the width of the input& output are fixed in *each* step.
That's why sha256 on GPU is so easy.

Scrypt tried to be GPU-proof by introducing random memory access, and it failed probably due to this (http://en.wikipedia.org/wiki/Scrypt):

"The large memory requirements of scrypt come from a large vector of pseudorandom bit strings that are generated as part of the algorithm. Once the vector is generated, the elements of it are accessed in a pseudo-random order, and combined to produce the derived key. A straightforward implementation would need to keep the entire vector in random access memory so that it can be accessed as needed.
Because the elements of the vector are generated algorithmically, each element could be generated on the fly as needed, only storing one element in memory at a time and therefore cutting the memory requirements significantly..."

Big number calculation used by Primecoin is a totally different beast. it's doable, but i'll definitely be much slower.