rieMiner - The reference Riecoin miner

[Pttn]

rieMiner is a Riecoin miner supporting both solo and pooled mining, and can also be run standalone for prime constellation record attempts.

Learn more (also with guides): https://riecoin.dev/en/rieMiner
Download: https://riecoin.dev/resources/Pttn/rieMiner.php
Source Code: https://github.com/Pttn/rieMiner



Suggestions are welcome here, and please also report bugs here.
However, don't ask for technical help, for this browse the Help subforum or open a new topic there. Be sure to have read and followed correctly the guides!
Go to the Mining subforum to discuss about optimal settings for your system or read this guide.

[Pttn]

rieMiner 0.92 Revision c is out and binaries are available on Github.

Changes:
- New assembly optimizations by Rockhawk, improving performance by 10-15% on AVX2 machines - if you like this be sure to donate some RIC to him (his address can be found on his profile)!;
- For Solo Mining, it is now possible to set Donate to 0 without workarounds and donations now go to the Riecoin Funding address ric1qr3yxckxtl7lacvtuzhrdrtrlzvlydane2h37ja rather than my personal address;
- New advanced option RestartDifficultyFactor option to customize when the miner restarts. Notably useful to optimize the Primorial Number a bit by reducing the restart margin, in this case it is a good idea to use GeneratePrimeTableFileUpTo to precompute the Prime Table and make restarts much shorter;
- Now that we have rieMiner-Qt, it makes the configuration assistant basically pointless, so it was removed;
- Some other minor changes.

With the new assembly code, it is no longer possible to compile generic binaries with AVX2 (I am using -march=core2), so one must now compile rieMiner to enjoy the AVX2 benefits with the EnableAVX2 option. This also means that the Enable AVX2 check on rieMiner-Qt will also no longer work.

I would gladly build 0.92d and later using -march=haswell or so, but apparently some still mine on old Ivy/Sandy Bridge or even Nehalem CPUs, even recent Pentiums/Celerons do not have AVX2, and only a minority would bother building rieMiner...

[Pttn]

Darn, I realized that one of the modifications causes an issue when running multiple instances of rieMiner. So I fixed it and am already releasing the Revision d. It also adds Primorial Offsets for 7, 8 and 9-tuples, allowing up to 64 Sieve Workers.

[pzktupel]

Hello, why do not sieve the program 10 condotions and prove the first 7 and than 8,9,10,... so the ratio is not ~60 , it is maybe just 30. So 8,9,10-tuplets are earlier ?!

BTW, I'm the successor from Tony Forbes' prime k-tuplet page.

http://www.pzktupel.de/ktuplets

greetings

pzktupel

[Pttn]

Welcome here Norman, and glad to see that you are regularly updating the list with new 9-tuples and large 8-tuples from Riecoin.

I am moving this discussion to the rieMiner's topic as it is more about the its default settings than the Riecoin Protocol itself.

The Riecoin Protocol only requires 7-tuples for its PoW, so the goal for a miner is to find as many 7-tuples as possible. It is up to them to choose whether to sieve for longer tuples.

With the current Difficulties, doing so will decrease the 7-tuples rate a lot. I ran some simulations at Difficulty 1024 and if sieving for 8-tuples with good settings, it is ~2x slower to find 7-tuples while ~1.6x faster to find 8-tuples and longer for my computer. If sieving for 9-tuples, it is ~4.5x slower to find 7-tuples while ~2x faster to find 9-tuples and longer. The calculations assume a ratio of 64 if not sieving (about equal to the number of 7-tuples divided by the number of 8-tuples found so far). I did not ran for 10-tuples, but we would find even less 7-tuples, and maybe about 2x more 10-tuples as well.
The results would be worse for current Difficulties around 920.
So if we sieved for 9-tuples, we might get more 9 or 10-tuples, but not much more. And they will be smaller because the Difficulty would adjust to the lower 7-tuple rate, so less interesting.

Things would be much more interesting if the Difficulty increased a lot. Around 1536, we could start considering sieving 8-tuples by default on rieMiner as it would be only ~1.3x slower to find 7-tuples while ~2.6x faster to find 8-tuples and longer. If sieving for 9-tuples, it would be ~2.2x slower to find 7-tuples while ~4.6x faster to find 9-tuples and longer. The calculations assume a ratio of 1.5*64 = 96 if not sieving, maybe it would be a bit lower due to the use of larger Primorials.

At 2048, the performance loss of sieving 8-tuples would be negligible (a few %) while 3.2x faster to find 8-tuples and longer so we could definitively do it by default. Sieving for 9-tuples would be only about 1.3x slower, while increasing the rates of 9-tuples and longer by about 10x. Calculations assuming a ratio of 128 if not sieving.
11-tuples (if we changed the default Primorial Offsets) might still always require decades to be found regardless of the Difficulty and sieving...

In these situations, we could find so much more interesting constellations, but there is a really long way before reaching 2000 levels, it would require about 1700x more mining power compared to now, unless one finds significant improvements for the Sieving phase...
Everyone, keep mining and spreading the word!

[Pttn]

A new stable release is out: rieMiner 0.93. Updates since 0.92 Vanilla:

- Improvement of assembly optimizations (+10-15% with AVX2), removal of the EnableAVX2 option (now always enabled in AVX2 builds);
- Now possible to configure rieMiner via Command Line;
- Bug fixes, especially for Windows;
- Remove the configuration assistant as we now have rieMiner-Qt;
- Add pauses to reduce confusion (notably, giving most Windows users a chance to read messages about wrong configuration);
- Support of the Stratum Client.Show_Message method;
- Many more Primorial Offsets are now hard coded (useful for people having a lot of Cores);
- Replacement of the Test Mode by a separate testing server for easier and much more realistic and complete testing;
- Networking code rewritten and much improved, Jansson replaced by NLohmann Json;
- There are no longer separated Light and Android branches, everything was reunited in Master. Easier build procedures;
- Other minor and aesthetic changes.

The rieMiner binaries are now hosted on Riecoin.dev. Find the links to the latest version here. https://riecoin.dev/resources/Pttn/rieMiner.php

Happy Mining!

[Pttn]

Time for a small revision 0.93a.
As always, download the newest binaries here. https://riecoin.dev/resources/Pttn/rieMiner.php

Changes from 0.93:

- Fixed a bug;
- Total Physical Memory is now a factor used to determine the default PrimeTableLimit, which should no longer have to be set manually if having less than 8 GiB of Ram;
- No more donation;
- Added a basic rieMiner Api, initially to help HiveOs integration;
- Build system significantly changed, now using Cross Compilation not just for Android, this makes the creation of official binaries much easier. Msys is no longer used/supported;
- Some other code improvements.

[TheRealCharlesMorris]

As far as I am aware there are 3 methods to donate.

1) When a miner has found an admissible block, and is also donating, there is an output in the found block to send the donation.
2) Miners disable automated donation in rieMiner, but then donate later in bulk at their leisure.
3) This new idea for automated donation, and it would save perhaps 10% to 15% of the size of an otherwise empty block.

Decreasing speed of growth for the blockchain is generally healthy for a currency- considering accessibility and counter-party risk. At this point in time the Bitcoin ledger is 454 GB, this fact vastly impacts accessibility, leading users to accept counter-party risk, and increase funds centralization, to simply to use the currency. In other words, this accessibility problem leads users to prefer services like Coinbase.com and blockchain.com or currency exchanges rather than free and decentralized wallet nodes.

When the funds are centralized, those funds can be the target of theft. I read claims today that between 3 and 4 billion USD in cryptocurrencies were stolen in 2022.

The idea is simply: Given a target donation of X%, instead of dividing the monies each block; X% of the time the miner assigns the block base address to the donation address instead of the miner's address.

One drawback of this idea is that the block finder of donated blocks is unknown, since it looks like the donation address mined it. If that does not bother someone, then it is not a problem.

[Harv563]

How much decrease in growth of blockchain do you estimate this would yield?

[Pttn]

The Donation system was removed in the latest version (months ago) for the reasons you mentioned, and notably as several people forking the code were changing the donation address rather than removing the donation system, bloating the blockchain for nothing.
Some could also argue that it was a kind of "forced" or "hidden" donation even though there was a way to disable it, so now I can also be sure that nobody can ever make such criticism.

To donate, simply send coins to donation addresses from now on.

The second output adds 31 bytes per block so in the worst case, that is about 17.9 kB more per day, or 65.2 MB per year. As a reminder, there are 576 blocks per day.
However, that is only if 100% of these 576 blocks had such donation every day, in practice, only a part of Solo Miners (which are themselves a minority of the miners) donate so, the actual size used is much less than these upper bounds.

There are several other measures that could be taken to reduce the blockchain growth. Some could be considered in a third Riecoin Hard Fork. Others would require the cooperation and responsibility of the Riecoin users.

[Pttn]

After a while, it is time to release rieMiner 2501, which will be joined by Riecoin Core 2501 very soon.
Latest binaries are available from https://riecoin.xyz/rieMiner/

Updates since 0.93a:

- Cookie Authentication for Solo Mining
- Proxy Connections Support
- Several bug fixes
- Logging improvements
- Option to stop rieMiner on finding a tuple in Search Mode
- Removal of the Primorial Number Option
- JSON-RPC 2 support for Riecoin Core 2501
- Complete code refactoring, the number crunching code is basically standalone now and may eventually be made into a library