Topic: [ANN][RIC] Riecoin: constellations POW *CPU* HARD FORK successful, world record - page 135. (Read 685207 times)

We're waiting for dga to merge the stratum code. I'm hoping he could make new binaries for Linux and Windows: that would leave miners no excuse for trying other pools.

I've started working on the superblock thing which feels like the new priority, I'm thinking on the easiest and safest way to implement the difficulty adjustments taking superblocks into account. I'm determined to break at least one record.


Thanks for keeping pushing me to write the updates!

They have a page to address exactly this question but it's a bit vague.

Any updates this week Gatra?

+1!


yep, we are talking about millions of digits... so only verification can take days. For single primes, maybe not Rabin-Miller but this: http://en.wikipedia.org/wiki/Lucas%E2%80%93Lehmer_primality_test or this: http://en.wikipedia.org/wiki/Lucas%E2%80%93Lehmer%E2%80%93Riesel_test

agreed... our primes all have lot's of consecutive 0's when written in base 2.... not an interesting feature though.



yes, many of us still have this question... did they lost interest, or did they find a secret algorithm? probably they just knew the records would be broken soon and decided to use the money for other things...

thanks, and let me know if you find that paper!


I use gitian to make the builds. I usually use the same as the latest available bitcoin release, since it's supposed to be well tested. Here are the latest deps: https://github.com/bitcoin/bitcoin/blob/v0.9.2.1/doc/release-process.md they still use qt 4 for compatibility reasons.

May be the solution to the so called Multi-armed Bandit Problem can provides further insight. See: https://en.wikipedia.org/wiki/Multi-armed_bandit
Also read: Overcoming Limitations of Game-Theoretic Distributed Control by Jason R. Marden and Adam Wierman
www.cs.caltech.edu/~adamw/papers/price_of_BB.pdf

Yes, this coin got me looking into ARM and SMID which I never had before.  I am pretty sure I've got some very clever tricks in my sieve which no one else has considered (well, aside from riecoin there isn't much use for a fast p6 sieve).

I've said it before but this coin is a side project, making profit would be a bonus.


There are some quite well known Montgomery multiplication design patterns for FPGA's and have been for years, same for binary exponentiation.  I have been wondering if those old crypto accelerator cards could be leveraged into riecoin (that is my pipe dream).  Also the main FPGA makers are supporting OpenCL.  Yes; write OpenCL and compile it into hardware.


This looks like a job for a pool where state across multiple miners can be maintained.  Maintaining validation and reward allocation across a distributed network would be PhD material.

Regards,

--
bsunau7

This kinds of discussions makes the value of Riecoin, it's lot more interesting than mixing and mashing hash algorithms!

The value of the coin eventually matter in that if it were high enough, it would incentivize some dedicated prime computing silicon, which eventually could have other uses. I guess NSA & similar have such silicon, but it's definitely beyond the budget of us mere mortals.


I am surely moving the goalposts here, but would it be possible to amortize the verification over many blocks?
Sort of a distributed partial Rabin-Miller (is that practical and can it be made provable?)

This would obviously have DOS issues, if a superblock is found incorrect after many blocks, you can't go back in time to try another, so that superblock would be "lost"...
It could be alleviated if for the superblock, up to X candidate prime clusters could be "registered" as meta-data, apart from the difficulty requirement, those would only have to pass low-probability primality tests, cheap enough in terms of computing but hard enough to filter spam (possible?). The registration could be allowed for a few blocks after the super-block (to minimize the network stall).
At this point the reward for the super-block would be zero, with tx vout with an amount of zero per candidate (just to keep track of their addresses).

Over the next N blocks, partial rabin-miller tests would be performed on those candidates alongside the usual PoW.
After the N blocks, candidates that failed primality would be rejected, and the superblock reward would then either be split among those that passed the test or attributed randomly.

It all hinges on the provability of a partial Rabin-Miller test, which may be a pipe dream

the font sucks but the ideas for different types of primes with unusual properties, i.e. with x number of consecutive zero's or any number

or with y number of non consecutive intigers (would probbly use a lot more memory)

the variations are endless, not really useful from a 'scientific' standpoint,

do you remember when the rsa used to post the bounties for primes ??

then they just stopped,  i often wondered if they came up with a much more advanced way of factoring and deriving primes that they never bothered to share with anyone else,  probably far fetched but they did pay those bounties and some were for big money ,

they paid them at the time because they needed to know how quickly people could find large primes and factor thier numbers for 'security'

lol

looking back with some of the things we know today it seems kind of funny


i often point out to people that in the 1980's 8 bit encyption was consider ok

lol imagine a few primes 8 digits long was considered 'secure' 

i also often joke that a child with a smart phone can crack that encryption now,


i wonder if in 5 or 10 years if technology will continue to leap ahead and what access 'normal' people like us will have who are doing things because we want to
(not because of the value of the coin like that other person said)

take your time and think it over, the superblock is a pretty neat idea, and i certainly favor anything that can take riecoin ahead of prime coin in the record books and personally i don't care about the value the coin unlike almost everyone else......

i remember reading a paper about 3 or 4 years ago about possible scientific applications of 'unusual' prime numbers but i honestly can't remember what the application was , if it was 'real ' or just a proposal and exactly what they were looking for,  maybe i can find that paper !!

whatever you decide you have my support, especially now that i am comfortable building wallets on the btc 9.x codebase

the most recent build i did as a test of the new iX coin client on 9.2 (coded by ahmed bodi) used open ssl 1.0.1i , qt 5.3.1 and min gw 4.9.1

if you fork the client and don't mind please suggest which versions of the tools you are using if different from the most recent bitcoin tools,

i remember the 8.x series of riecoin used qt 4.8.3

thanks much

yes, thanks! I didn't pay much attention to it before because I hate it how they use the "comic sans" font...

ive been following, i agree to verify proof of work of just large prime could be a very long time

would be a great proof of work algo but block times would be very long

edit 1 i am sure you are familiar with this work

http://primes.utm.edu/bios/code.php?code=G13

http://primes.utm.edu/index.html

edit 2 another idea http://primes.utm.edu/curios/first.php

Quick turn around & thanks!

Regards,

--
bsunau7

So the test is needs to include a "p4 or p12 not prime" test as well.

Sorry for any false hope!

Regards,

--
bsunau7

Yep they deserve my respections:)

yes, 6 to 9, to 10, whatever.

Moreover, if my math is right and can be generalized this way, we would break the twin prime record in 11 minutes and the largest prime number in 16hs.
The problem is that verification of each of those blocks would take some minutes, so the client would take a lot to sync. That's why I choose 6-tuplets: easy to verify, hard to find. But if we made it superblocks only once a month, syncing wouldn't be so bad, so it would be possible!
Also, since block verification would take so long, we would be prone to DoS attacks by someone submitting fake blocks that would take a lot to verify.... if we could mitigate this, we would break records from twins to many-plets. Or even the largest prime number. We would be sacrificing a bit of the usability of the currency aspect in favor of the distributed computing aspect of the riecoin project.... and... there's a 150k USD price for primes larger than 100 million digits... I'm starting to think some reengineering of the mining process, and strengthening of the DoS protection could make us get there...

getting too excited, need to think more about this...

edit: verification would take prohibitively long for single primes and twins, even if we did it once per month. Probably for triplets too

Yes, you're right.


In that scenario would it be possible to accept 6 to 9 tuples?
Or more precisely allow the current miners to keep working on the "old" 6 tuples, while allowing new miners to mine greater tuples? perhaps with a small bonus to give an incentive to upgrade

The appeal of world records is strong

might be, but this would not only be a hard fork: all the miners would have to be updated too.
if we only added the superblocks, those mining in a pool wouldn't be affected - only the pool server would have to update the riecoin client.
You are proposing a more aggresive move.


We would need a difficulty of 1060 (in 7-tuples). Current computing power (1462 in 6-tuples) would translate to a difficulty of approx 705 for 7-tuples. So changing block time to 4 minutes (or having 4 mins superblocks) would do the trick. If we had 4 min blocks we would find a world record size 7-tuple every block and a record octuple once or twice a week. Some years for a nonuplet, but it would be one hell of a nonuplet, exceeding current record by orders of orders of magnitude.
edit: I'm tired, my math is shaky... it's not 4 minutes, it's 2 hours. Still doable with superblocks, but forget about finding nonuplets - or octuplets - this way.

If we were going to change the PoW in such way, we'd better make it accept alternatively 7-tuples, 8-tuples and 9-tuples each with different difficulties, so we could easily break all records every day.

I made a lame attempt at giving room for this kind of change in the future by adding a "primes" value hardcoded to 6 in the "getwork" and "getblocktemplate" calls, thinking that maybe in the future this could be variable. But the thing is that we would have to update all the mining software, and see which patterns we would look for, not only the number of primes. So we would have one block be 6, the next 7a, then 7b, 8a, 8b, 8c, whatever, then repeat.

edit: looks like for 25-tuples, there are 18 possible patterns all with diameter 110... I guess we wouldn't go that far

Drats!

According to https://en.wikipedia.org/wiki/Prime_k-tuple the septuplet constellations

0, 2, 6, 8, 12, 18, 20

might sometimes be octuplet and nonuplets.

Would it be worth changing the PoW from sextuplets to "septuplets with a chance of meatballs octuplets and nonuplets" ?
Would the current computing power in Riecoin mining be enough for world records then ?

cool, thanks for the updated list!

I'm afraid that doesn't work. Imagine that you did find p-2 and p+18 prime. So you have a septuplet. But, you started from a sextuple so p+12 is also prime. This means that your septuple is also an octuple: it has 8 primes in the range p-2 to p+18. A difference of 20. But Athony Forbes tells us that the minimum distance possible for 8-tuples is 26, so the 8-tuple with distance 20 cannot exist. But we said it did... this absurd comes from assuming you could find p-2 and p+18 both prime.

The thing is that ignoring the 12 is cheating, you won't have this:
but actually this:

and that is not possible because there is a prime q, with q<20 where any of the p(i) will be a multiple of q.
edit: let me see which one it is.....
ok, it's 5. Let's add 2 again for simplicity, so we have:
now:
if p+0 has the form 5x then it's not prime
if p+0 has the form 5x+1 then p+14 is actually 5x+1+14 = 5x+15 = 5(x+3) then it's not prime
if p+0 has the form 5x+2 then p+8 is actually 5x+2+8= 5x+10 = 5(x+2) then it's not prime
if p+0 has the form 5x+3 then p+2 is actually 5x+5 = 5(x+1) then it's not prime
if p+0 has the form 5x+4 then p+6 is actually 5x+10 = 5(x+2) then it's not prime

so it's not possible to find primes with that pattern...