Topic: rpietila Altcoin Observer - page 222. (Read 387493 times)

hard to believe they are selling on poloniex if you watch that book

Christian the GPU miner is mining 7K+ each day. At market price that is 14 btc a day. It is hard to compete with... BBR dev is working on a solution though, but plan B is a fork.

Why don't they just issue a bounty to have a GPU *public* miner done for them?

DRK did that before even private GPU miners came to X11 (a cpu whale scared everyone into thinking a GPU miner was on the loose - but as it proved later on, GPU miners weren't that fast with X11) so as to avoid the selfish GPU miner scenario. I think it was ~3500 DRK as a bounty back in mid Feb...

Definitely not dumping.

I think memristor blocks are probably already available - definitely in 2015 - and would make a keccak/cryptonight asic much better at scale -- no dram required, no pin bw bottleneck.

I should have explained my message better. BBR has a GPU mining problem as its GPU isnt open source. The one using is has a way better advantage against the CPU miner since its using a different algorythm that crytonight (wild kekak i think). It has the consequences of centralizing most coins in one single entity. 95K BBR is HUGE comparatively of the current supply. And if the the guy isnt dumping, according to his posts on the GPU thread, he has 7k more BBR each day.

Bitcoin must have had this problem during the mining arm races. I dont know how they resolved it though as I wasnt there! My idea: patience is all.

Interesting, we are at over .006 BTC now and if you look at the Poloniex orderbook there just are not many sells. https://www.poloniex.com/exchange/btc_xmr
The first notable wall is at .007 and 35 or so BTC, then a "wall" of 9 BTC at .008.

There are 57,000 XMR for sale on Poloniex and just giving a rough price of .007, that is only like 400 BTC worth.
This has explosive written all over it.

But something doesn't make sense. If Bots are large miners of XMR, why are we not seeing large dumps (or are we)?
I guess the buy volume is meeting the demand...

And you think this is not true with XMR or BTC? I think you know the answer.

This is the problem with BBR.. 85K coins! At anytime you can be dumped on!



It's not there anymore though.

What about fpga implementations? Would power efficiency be enough to justify the development? If yes, what would be the right dev board for cryptonight?

Partly agreed re ASIC.  But it'll just move the bottleneck to the memory controller(s) on the ASIC.  You can probably pack quite a bit of memory bandwidth in there, but off-chip b/w will be the limit.  The power efficiency will, of course, be much better -- perhaps 10-20W for ASIC+DRAM vs 90-100 for a CPU, once the CPU miners are fully optimized (the current miner uses only 55W on my 85W TDP i7).

If all of the implementations are pin-bandwidth limited, the big advantage of an ASIC will be that you can pack the asics and dram densely at low power, close together, but that kind of advantage is numerically relatively weak compared to the asic advantage in bitcoin.  If we really want to be nitty detailed, current DRAM designs may also present the bottleneck:  The 2Kbit row buffer is inefficiently used by random 256 bit accesses.

The more interesting long-term question is whether the linear scratchpad growth is an appropriate match for whatever happens in the (very uncertain) moore's law space.  That's a toughie.  Right now, the linear growth will probably take the scratchpad beyond CPU L3 sizes pretty quickly, but in the longer term, my crystal ball fails.

I don't really have an expert opinion regarding the GPU dragon beyond the efficiency gains and what I've observed from cloud mining - I'll leave the economic issues to others.

It just means you need a PCIe block on your ASIC.  Or enough memory to store the blockchain.  Not in any way insurmountable.  At 2-3x speedup (existing example being the best estimator) it is also much less of a threat than it is for bitcoin.


cb said he was mining 5k BBR daily at one point.  Now it could be more (more hardware) or less (more difficulty).  100k would not be a high bar for him.
I see a lot of latent supply risk in holding.

. How did that zero get in there? I meant to write 8. My bad.

I find that when I divide .00536199 by .000719 I get 7.46. Is my calculator broke?

I did. I was trying to accumulate between .0002 and .0005 roughly, though not many went through below .0003. Presently it's about .0007 which makes it roughly 80 times cheaper than XMR. The developer is trying to find improvements where he can, so I think there is some value in that for more than just QCN. https://bitcointalksearch.org/topic/m.7380771
The only thing IMHO that could make it rise signifigantly, is a good development team.
But then again, after many years of trading various instruments I find that what I think should happen, doesn't always correlate to what does happen. 

Hey, folks - since my comment was cited in this thread, I figured I'd chime in.

Let me first admit my bias:  I now own 6700 BBR, so I'm no longer an unbiased observer.  I'm actively contributing code back to the standalone BBR miner in order to try to increase the value of those holdings by making sure that people with small setups can fairly mine compared to the huge operators and cloud miners.

I stand by my analysis of the GPU efficiency gap for this one.  As I stated in another post, my _hunch_ on these coins is that XMR's algorithm will prove better for GPU/CPU efficiency.  Once all of the bytecoin slowdowns were stripped out, it's one of the most elegant PoW algorithms I've ever seen for achieving this goal.  (yvg1900 tells me I need to look more at MMC before I make up my mind, and he's probably right.)  It's brilliant.  I wish I'd designed it.  It has the drawback of slow validation, though.  It's painful on both GPU and ASIC, unless I've missed some tricks.

BBR's blockchain scratchpad idea will be a pain for ASICs without slowing down validation.  It doesn't quite have the elegance against GPUs, but it will be better than things like scrypt or Momentum.

(I stated in that earlier post that I thought BBR might be better against ASICs.  I've changed my mind.  I think both will actually be a pain.)

I'm also still pretty certain about my analysis that cbuchner1's GPU miner is only 2-3x faster than a CPU for the same cost.  There are a lot of caveats attached to that, and I'd encourage people to read my full post before trusting this too much ( https://bitcointalksearch.org/topic/m.7362060 ).  Part of the reason that I'm more confident that there's not a GPU-equipped dragon sitting in a cave with 100k BBR it is that I and several others (some publicly, even) continue to mine BBR on EC2 from time to time.  I've posted about a few of those opportunities ... after the fact.  Now, it could be that this is because of the diff adjustment, and the GPU-dragon is simply annoyed during these times because he can't get all the coins, but I'm a little skeptical.   But I may also be naive.

I think the more fundamental question with BBR is whether the blockchain changes (improved anonymity and reduced blockchain size) are technically sound and effective.  I've told Zoidberg that I'm willing to help with a user-friendly presentation of these, because I want to understand them better myself.

The problem of Nxt clones is that it is impossible to change the code without breaking everything if you didn't participate as a Nxt core dev.
All Nxt clones tend to die quick. You need to hire someone like Come_From_Beyond is you want a Nxt clone to succeed.



Most altcoins are definitely useless but we need some of them for innovation.

People are too focused on Price for altcoins. If they don't have anything you can do with them but trade for bitcoin. They are useless.

what do you think about nxt clones? Will any of these succed?

There are only two miners concerned (Christian & Christian dev of ccminer which have developped a private miner and will probably incorporate in ccminer later): they do not operate from official pool...