Good point. ASIC efficiency is probably about to plateau for a while. One consideration to add to that is S2 vs S1 - a more chip-dense machine with the same ASIC about half the per-hash power (for the current generation, probably about 0.6J/GH machine-level) will have a higher initial cost but a substantially longer viability. In that way I guess Bitmain has done hackers a favor going with a bucked string topology on later S7 and S9. For a regular joe it's actually worse than unregulated string because you get worse efficiency but if it can be hacked to adjustable voltage that potentially doubles (or more) the miner's effective lifetime.

Cheap can refer to price or build quality. Bitmain didn't used to sell either, but started going cheap on build quality with the S5.

Yep. Honest and polite is the best way to go, but if you have to pick one always pick honesty.

I've got about 40 S7 from various batches in hosting, so if I end up with any of them would save shipping cost. I'm working on a trade for a bunch of S5 but if I come up short I'll let you know.

I also will never mine on Antpool.

Well, I have a unique background. My father is a jack-of-all-trades carpenter and the son of a civil engineer; my mother is an English teacher, the daughter of a math teacher and a telephone lineman (farmers both), so I grew up alternating between swinging a hammer and digesting classic Literature. Alongside my engineering and science degrees I took enough English classes in college that I'm one shy of a minor in writing - something I might complete if I can afford to get back in this fall.

Oh yeah, and hardware is overpriced. I'm looking forward to my S7LN getting shipped any day now, so I can play around with optimized undervolting. Might pick up some cheap 135-chip S7s after the halving and crank 'em down to baseline efficiency. If any of Hawkfish's machines in my hosting are still up for grabs by then we might strike a deal.

Probably won't have any 16nm gear of my own until I'm running stuff I've built myself. Don't really have the budget anymore for a $2100 miner, not when that's more than I paid for my old-enough-to-drink car.

Well, I just turned 30 last month so I don't think I qualify yet.

Nothing wrong with that. As old bastard in training, I respect the point.

CK - my question was mostly rhetorical. I know all the business reasons why people deal with China. I also know (and for the most part represent) all the reasons why some people refuse to deal with China.

So why do we continue to do business with them?

As was noted by one of the cgminer devs earlier in this thread, Bitmain has repeatedly failed (or at best, delayed) to release the code for the cgminer implementations on their miners. One of the license requirements for cgminer, given that it's open-source software, is code must be publicly released (to continue its open-source nature) especially for any implementations used in for-profit commercial applications. S9 already exist on the network; they might not be in the hands of us peasants yet but they're already mining and generating returns for someone.

I thought about doing that with a bunch that I had, take them down to about 0.8J/GH, but never got around to it. Space hasn't been an issue for me since we moved operations into a 5000ft^2 shop. Pretty sweet setup though.

But in the case of S5+, S7, S9 I wouldn't consider heatsinks reusable. The controller board would only be if you were able to use the same IO and cabling for the new chip's protocol. If you're going to build new boards for old hardware, the S7 and its twins would be just about the worst thing to pick since, at practical best, you get to reuse a shell with fans. S1/3/5, A1 Dragon, Tube or Prisma, RockMiner, most old Avalons - all better options for reusing mechanical parts with updated electronics.

An S7 won't be of much use to you anyway since the important mechanical parts are pretty much epoxied onto the ASICs themselves. It'd be more effort to salvage them than it would be to just replace 'em outright.

I don't need a PSU that beefy. Honestly a good 650 would do it, so a cheap 1000W is probably about right and I don't want to gut my Dragon.

I'm going to have a lot on my plate starting about Tuesday of this week and going through the forseeable future, but I'll probably still make sure to take a day and jack around with this miner as much as I can. The BM1385 should be capable of going below 0.2J/GH board-level.

So what you're saying is, we should wait to see what coin the porn industry gets behind (giggity)?

He might have misunderstood it as $80/KW-month, which is pretty bad.

Bump, I've still got DPS812 boards available. I'm looking into redesigning the Dell 750W board to take screw terminals or 6-pin jacks like the new DPS812 board and, if the right thing or two comes together, I'll probably be doing a batch soon. One idea is to offer a two-supply kit with all the cabling required to power an S7 or S9 for about $100, and these would do it off 120VAC.

I am looking to buy about 8 or 10 of the AntMiner S5. I'd prefer working units.

I'd be surprised if the website were necessary for function. Isn't it mostly just for easy detection of miners on your network?

Your pleas will fall on deaf ears, as this is primarily my project and I've never cared about scrypt ASICs. Maybe someone will extend it someday, but I'm really only interested in SHA256 miners.