Topic: [XMR] Monero Mining - page 20. (Read 264880 times)

The PoW algorithm should be modified so that botnet will be discouraged.

I believe the ccminer developer who goes by the username sp_ has an optimized private cryptonote GPU miner that he sells for 0.1 or 0.2 BTC.
Something like that.

I don't know for a fact how much better it is than tsiv's GPU miner, but it is reportedly faster.

Unless I am wrong and it is actually a CPU miner...

Basically, someone is mining with access to thousands of PCs. This could either be botnet or very lucky system administrators with system users that don't need 100% of their CPU time.

In either case, someone is getting boatloads of Monero for free. My assumption is that these coins are dumped to extract the most profit. But who knows. They could be hodlers.

My only hope is that these operators are getting this message - if you want to continue reaping profit from one of the only CPU mineable coins that is profitable, you should really donate a large percentage of these coins to Monero core development. After all, the only reason your whole enterprise is possible is because Monero exists.

To me, this is just further evidence that we need better GPU software. Yargh. I don't know what the best solution is. I support the whole everyone should be able to mine thing. But yargh.

this is a hard coin to mine , that is all I know

if someone has a better miner it would not be the first coin where it was not shared with others

I do remember seeing some chat about a 5-10% increase in hashrate with a private version of ccminer posted either here or on irc. Probably worth it to give the person that releases that speedup 10% of the bounty, whether that's the author or just someone who paid for it and now just wants to release the binary for extra $$$XMR. I'll solidify my contribution to the bounty at 35 XMR, so a 10% speedup would be worth about 3.50 XMR

well, an update: the operator of monero.net contacted me and provided information suggesting that there is not some super amazing optimized code.

I agree with jwinterm - a 2X would be significant enough to warrant full bounty, and subsequent percentages of 2X that get corresponding percentage of bounty.

However, we should be aware that a 10% improvement already exists - somewhere in the ccminer thread, but the dev hasn't released it publicly. Ah i can't find the link.

The 40.4x factor does seem a bit ridiculous and arbitrary. To be honest, I think even a 2x improvement at this point is asking for the near impossible. Maybe the way to structure it would be to have a 2x or 4x improvement take home all the cash, and so then a 10% increase in hash rate would be worth 5 or 10% of the current bounty. Just a suggestion, because I (from a pretty uninformed perspective) really think 40x is asking for well beyond the impossible.

Maybe those high hash miners use a proxy. They send the hash combined with many miners.

okay. As I mentioned, the theory is that some ridiculously optimized miner may exist.

chocolatebar mentioned goalposts for the bounty i proposed. I've never done bounties before, so I have no idea. So I'll just throw something together and we can start hashing it out.

I'm particularly interested in nvidia (so the cuda miner), mainly because the maxwell architecture is very efficient hashes / watt, at least with the last version of tsivs ccminer.

As mentioned in my original post, I proposed the full bounty for a 40.4X factor in hashrate. So, with my 750 ti card, that would be 280 h/s * 40 = 11200 h/s. Yeah. 11.2 kh/s. Man my rig would fly.

Granted, obviously, that might be ridiculous or impossible.

Smooth mentioned some stuff here https://bitcointalksearch.org/topic/m.11571118 regarding GPU stuff.

For the 290X, as mentioned below, https://bitcointalksearch.org/topic/m.11730818 , this card kicks out 810 h/s. So a 40.4X of that would be.... 32724. 32.7 khs. Yum.

SO:

If this code exists, this deserves the full reward. Thus, the primary bounty is for these numbers. Because these numbers SHOULD be impossible, this bounty should be ridiculous. So far we have:

GingerAle: 200
Jwinterm: "about 3.5 XMR" which I will interpret as anywhere between 3.5 XMR and 3500 XMR.
Chocolatebar: 2000
owlcatz: some yet to be determined amount
Lloydmiller4: 500 - 1000

Thus, at the extreme high, we have 200 + 3.5 + 2000 + 1000 = 3203.5 XMR


Lets use the number from my analysis 137.7 kh/s, and assume that thats the hashrate of our miner-with-awesome-code (MWAC). Yes, this assumption is flawed, because MWAC could have 80% of the hashrate and the other 11 people have the other 20%, but whatever.

Option #1:

Full bounty for code that can make my 750 ti spit out 11.2 kh/s, as calculated by my 750 ti existing hashrate X 40.4. This number actually comes close to a scenario where MWAC has 2 full rigs, with 6 GPUs per rig. So each card is spitting out 11.47 kh/s.

Option #2: (because option #1 is ridiculous)

Yah know what, we can just use 40.4 as ceiling, and any improvement in hashrate below that will get a bounty = that factors percentage of 40.4 X the full bounty...

a 2-fold increase in hashrate would be 2/40.4 * full bounty. So, at the current full bounty, thats 158 XMR.
a 3-fold increase in hashrate would be 3/40.4 * full bounty. So, thats 237.88
4 = 4/40.4 * 3203.5 = 317.18

... you get the idea.


In the case where a gynormous bounty is created, and someone releases 2X mining code, we could continue the bounty. We would just subtract that fold from the original number.... ah shit this is some complicated mafs. All right, ...

ori_hash = original has (280)
target_hash = MWAC theoretical hash (11200)
fold = original max fold (40)
bounty = original total bounty (3203.5)
relfold = released fold (2)

bounty_release = relfold / fold*bounty

new_bounty = bounty - bounty_release
new_hash = ori_hash * relfold
new_fold = target_hash / new_hash

nah, but then the new fold would be 20. And if someone made a 2 fold on that, they'd get 2 / 20 * new_bounty, which would be significantly more than the first release... but I guess that might make sense.

If someone wants to map out the function of what I was getting at above...

ah nevermind, We can cross that bridge if we get to it.

I think option #2 might cover it, because it encourages 2 things - a ridiculously high primary bounty which effects the lower bounties.

STIPULATION / PROPOSAL I propose that we donate 10% of the bounty to core development, because Monero.

I will of course provide a bounty to having an open source AMD miner, I believe we need one. Is the CUDA miner opensource? I know there is an optimized CUDA miner for people with big rigs(at least I heard there was), but we need this stuff open to the public and get it more decentralized.

I would provide 500-1k XMR additionally for a bounty, easily, for helping GPU miners get more involved with Monero.

You can use Claymore's miner but is not open source.

Totally. Never understood why no one but Claymore has one coded yet. I'd add to the bounty even, just to make more xmr.

What is the theoretical up limit of GPU performance in XMR?

For AMD R9 290X, the hash rate is 0.81kH/s using Claymore miner.

Interesting analysis! Thanks for posting this

edited some numbers due to copy and paste error. Numbers of that pool are still extreme, but less so.

I don't think AV software is necessarily very helpful in this case, since the miner is likely installed with the user's permission, either bundled with another program that they just click through too fast (a la utorrent), or installed surreptitiously when another program is installed. This has been a topic of discussion lately in the vertcoin thread, as one pool (ipominer) has apparently had a big botnet mining at it's pool that is a substantial part/majority of the network ( https://bitcointalk.org/index.php?topic=895020.1100 ). The pool in question tries to deny that it's actually a bot net, arguing semantics, since apparently people are installing it just by clicking through the installation options too fast when installing some other program.

Anyway, I think there's a few other options that kind of fall somewhere in between option (1) and option (3). Like, if someone works with/owns a large server farm, and their only running it at full blast in off hours, or maybe even a gaming focused internet cafe or something, where they have a bunch of comps equipped with gpus that they can run in off hours. Who knows...

With that being said, I pledge about 3.50 XMR to the bounty proposed in this thread

After a quick 'n dirty analysis of the network distribution of hash on monerohash.com (divide hashrate of each pool by the number of miners in each pool), the pool monero.net has a per miner hashrate (137.7 kh/s) that is 40.4X that of the average of all other miners (3.4kh/s).

I can think of 3 reasons for this:

1) Someone is investing significant resources into mining monero in that pool

2) Someone has developed an optimized miner (GPU, etc)

3) And, of course, bonnet

If the reason is #1 - great. Whoever you are, thanks for investing 40.4X the average contributors resources to support the network.

if the reason is #2, congrats to whoever you are for tweaking the code. However, being that the security of the network depends upon the network hashrate, I think it would be beneficial to share this code, if it exists. Granted, if this code and individual exists, they are enjoying wonderful profits right now, so we would need to make the release of this code financially rewarding.

The pool reports finding a block on average every 7 minutes, therefore this pool gets a rough estimate of 2k XMR per day. So if this is an optimized miner, it will probably cost a lot.  

Upon further examination, this pool might be the source of the network flucation (I think someone else may have discovered this before), as evident here: http://monero.net/#pool_stats





However, even at the lowest points, this pool still hashes 600 kh/s, which puts the hashrate / miner at 29X the rest of the entire network.


Edited: interesting. With my new numbers (must have missed something in the copy and paste. woops!) when the low end of the wave is used, the average hashrate / user for that pool goes to 46.16, which is still 13X the average for all other users.

Thus, I propose a bounty dependent upon release of mining software that provides 29X increase in hashrate for my mining rig. I don't have much, but I planned on building a new rig soon. Thus, I will pay 500 XMR for this software (that might not exist), or contribute 200 XMR for a bounty for releasing the software (that might not exist) to the entire network (because if that happens, then the whole hashrate goes up and I'm just burning 200 XMR). Actually, to be honest, if I got it privately I would release it anyway because network, so we should probably try to get the public-release bounty going.

If reason #3, well... then.... people should really install antivirus software.

EDITED some numbers due to copy paste error. Edit time 10:45 AM EST.

Best Monero pool on the World. <3

Enfin, ce sont d'excellentes nouvelles !

Fantastic! Thanks for adding this.