Not random, sure, but certainly chaotic.

I haven't gotten notification that the PCBs ordered on the 14th have shipped yet so there's been no progress.

It's a GekkoScience 2Pac. There's a thread on the first page, with posts from today (instead of a year and a half ago), with "GekkoScience 2Pac" and "support" in the title. Probably a good place to start.

Are you a collector specifically looking to get an EU version, or just someone in the US who can't locate one of the dozen-odd resellers of the regular stick?

The config issue is a known issue that, from what I remember having been stated at least once already in this thread, is a carryover from something broken in cgminer core and not related to VH's gekko driver.

Or the apparently very difficult to acquire Hotmine stuff, that turns a S1 chassis into about a 900W miner.

Maybe. Gotta remember it's got almost 200 16nm ASICs in there, and those aren't exactly cheap to make.

The ready-made downloadable Windows binaries whose download links are available in the first post of this thread? Those didn't detect your 2Pac?

I'm gonna go out on a limb and hope English is not your native language.

Ah, okay. So you're trying to find a way to predict the output of double SHA256 given a set of random inputs. Good luck.

Also I don't think that has anything to do with this thread's topic, or with "overclocking".

I'll be honest, I really have zero idea what you're trying to do.

The statistical incidence of a hash capable of meeting difficulty criteria is approximately 1 every (entire world's hashrate)*(600 seconds). The difficulty criteria is self-adjusting to make sure this remains true. You can't just wave a magic wand over a Block Erupter and coax it into calculating more golden nonces. And you can't make it run infinitely fast before it catches on fire. And even if you could solve the meltdown limitation you'd be paying out the butt for power compared to new gear. New gear sells because it is quantitatively better.

Please spend a little bit more time reading and learning before trying to explain yourself, that way you can get a better idea of whether or not what you're trying to explain is remotely possible or if the words you're using mean what you are trying to say.

Overclocking works great until physics happens and your chips start to burst into flames. Also consider that MOS power consumption pretty much increases with the square of switching speed so your efficiency will fall off really fast. And then your chips will burst into flames.

CPUs suck for mining because they have a very few very generic cores so there's a lot of extra stuff running and very little of the right kind of math going on. Think about CPUs like idling your car to run a 12V inverter into a USB adapter to recharge the battery on your phone.

GPUs are better for mining because they have many less generic cores; there's still a lot of extra stuff but it's a much lower percentage of the total amount of stuff and there's more of the right kind of math going on. That's your small standalone gas generator into a USB adapter to recharge your phone.

ASICs are the best for mining because they have almost literally nothing except incredibly specific cores. There's a small controller built in that handles data IO, clocking and basic maintenance tasks, and the other 99.7% of the chip is a hardwired piece of circuitry custom-built to do exactly the kind of math bitcoin mining needs and nothing else. That's the little solar panel with a USB charge cable that plugs directly into your phone.

I think some of you guys have some fundamental misunderstandings about how mining chips work, or how the design and manufacturing process works. And then stuff like "generic Intel i7" contradicts itself - you can't go into a store and buy a non-name-brand i7 for your computer. You get the one Intel designed and manufactured, or you get nothing. Nobody else makes them, only Intel. Just like only Bitmain makes the BM1387, and only BW makes the LK1403. And by "makes" of course I mean "spends months and millions of dollars with a team of experts designing, simulating and redesigning, then sends the design to one of the exactly two open factories in the world capable of manufacturing". Intel has their own factory so they literally do design and manufacture the entire CPU. They just also sell them to third-party integrators. Most of the bitcoin mining ASIC developers don't, probably because they can profit enough already from making the entire miner and if someone else made a better miner with their chips it'd make them look pretty bad.

The foundries, in the case of Bitmain's chips TSMC, only know "how" to make the chips because Bitmain's designers figured out the design. TSMC just has the tools to manufacture. Bitmain's the architect, TSMC is the carpenter. TSMC isn't going to sell Bitmain's design to anyone else because they'd get sued up the butt and their lack of integrity would cost them a lot of business from the rest of the world.

Also why did you build the .exe from scratch? Did the ready-made downloadable Windows binaries not work for you?

Bitmain spent millions of dollars to custom-design and have fabricated the chips inside the S9. Canaan, EBang and BitFury each did the same thing for their respective designs. They hold the rights and they hold them tightly and it's only fair that they do so.

So yeah, if you wanted your own comparable chip you'd also have to spend millions of dollars to design and have fabricated; the other option is, you could go to Bitmain, Canaan, EBang or Bitfury and see about buying their chips for your own integration. Bitfury would probably be most likely to talk to you, and it'd only cost you about $1M upfront instead of $10M plus about six months R&D time.

If I remember the "about us" on their page (which won't load a darn thing without javascript, screw them and their web designer) they claim to have been around in the LA area but moved operations to China sometime last year. So who were they back then?

I myself don't mine all that much. I started out with a 10GH AM Blade back in September 2013 and, with reinvestment and ingenuity, turned it into about a 2TH operation by the following April. After proving I could make a whole lot of something from practically nothing, I shifted more to collecting - buying one of something new just so I could play with it, not necessarily to mine.

Right now I'm running a 100KW datacenter entirely used for hosting other people's stuff. I have I think two of my own miners running right now, but that's only because they're being tested out after a voltage hack (for the record, they're NotFuzzy's retirees). I did all the datacenter's electric install on my own, built the network infrastructure and shelving and whatever else. If I do the work myself I don't end up paying someone else a heck of a lot more than I think it's worth which saves a lot on initial costs, and I don't have to second-guess anyone that it's getting done right.

Most of my business (hosting just pays the shop's rent) is actually designing and manufacturing small-scale miners. I've put out a couple USB sticks, am inches away from a ~60W quiet home miner and an updated version of that (with better chips sourced from Bitfury) should also turn into a 4TH/450W retrofit kit for an older Bitmain series chassis.

With a well-made miner, and note that very few are up to my standards these days, you can adjust the chips' operating point for increased efficiency (at a decreased hashrate of course). This can extend the viable life of a miner substantially - for reference, the two miners mentioned above are 45-chip S7s, which stock at about 4.7TH for 1400W or about 0.3J/GH; I have them running at 3.7TH at 880W off 120VAC (not as efficient as 208/240V), a shade below 0.24J/GH and a heck of a lot quieter. A lot of machines don't let you make adjustments like that, which is basically total BS.

Late arrival of a rig is a consequence of ordering from a shady manufacturer. Never ever pay into a company without a proven design. If you're buying a thing "six months out" you're more than likely buying from someone who will pocket your money and vanish into the wind.

A lot of profitability will depend on your electric cost. That's likely going to be the single largest factor for financial feasibility.

If you're wanting to build your own miners for BTC, you're either gonna have to go to an existing manufacturer like Bitfury with at least $1M for an order of their ASICs, or come to an existing IC developer with about $10M (and around six months) to run you out your own custom design. Unfortunately the only "realistic" course - unless you have a whole lot of money you're willing to risk losing entirely - is to buy existing miners from someone else.

That would only be a problem if the miner isn't mining but the fans are still running. The heatsinks get hot enough to almost boil water during regular operation.

3.5nm is a pipe dream so remote this is the first time I've heard anyone even mention it. 7nm is a pipe dream at least two years away. Even 10nm for practical/reliable use is at least a year away. Remember there are only about two places on the planet one can order up 14/16nm designs, and only about two places on the planet currently even attempting 10nm fabrication.

If you check out Mining Speculation, you'll find more discussion on how foolish it is to expect anyone to put up 7nm miners in any kind of timeframe to affect current utility of S9s.