Hype or not, it's fairly accurate. Spondoolies has always mentioned in the past that their margins were not great. In addition to ASICs, each board in the SP20 contains 16 VRM phases (twice as many as an S1) compared to the S5's zero VRMs. The controller was designed entirely in-house instead of leveraging an off-the-shelf devboard. By a pure components count I bet the SP20 is over the S5 by a full order of magnitude. The PCB very likely contains quite a bit more copper as well, given the current density required for their ASIC topology. That they sold SP20s at the same price point as an S5, I'd believe the last of the machines they actually took a loss on.


I'm going to try as hard as I can to make a solid miner design in the next several months, and if I can do what I want to do you'll eat those words.

Anyway, MBP's miner buyback.

Bitmain, maybe. Spondoolies is inbetween chip generations and probably won't have a deployable miner for a couple months.

If you can fetch a box of broken gear for little or nothing, fix some of it, sell most of the working ones and mine on the rest (or the ones that you can just get limping). This requires practical skill in electronics repair, of course, but it works.

Poke the chips with your finger and see which ones are slightly cooler. Also, check voltage outputs on VRMs with a multimeter; a bank with down chips will read a very slightly higher voltage because of less current sag. This will be more noticeable if you check the voltage at the bottom of the board.

I can confirm this. The PSUs are 4x DPS-2000BB with our interface boards and cabling. Whether Cheeseater releases his hosting space or not is up to him, but if the buyer and he agree to leave them here, hosting is a flat $0.10/KWh. The one-time new customer fee is $25, and VPN remote access is free.

I'm also working on an undervolt study on AM BE200 gear, including Tubes, and will probably offer undervolting as a free service to hosting customers when I release the results.

No, I purchased one new from Bitmain last July. I had sold a bunch of S1 and was gathering coin for some Habaneros but their last batch sold out about three hours before all my funds were consolidated. It was also the last day to cash in a $400 S2 coupon so I went for it. It's been running one of our prototype-batch DPS2K boards pretty much continuously since then.
That kit's the only S2 I'd owned before about Wednesday. I just fetched a used stock S2 (without PSU), and took another one in trade from a friend for fronting him a loan for some S4. I'll keep one stock for posterity and seek an upgrade kit for the other, for the museum shelves.

I bought the kit without a PSU and strapped it to a DPS2000. Not sure how much power the backplane is rated to handle.

I would also be quite surprised if they required a PSU upgrade for the new kit. It makes the most sense to limit it to 1000W, 100W per board.  Possibly something like a 3x20 string clocked at 6.4GH/chip per blade, which would give 3.8TH per machine (and, with adequate power and cooling, an overclock potential around 4.8TH). That sort of chip density they might keep for their S6 proper, so an S2 kit would more likely be a 2x18 string clocked around 10GH/chip per blade at 3.6TH per machine, overclock potential about 3.9TH.

Unless they're going to be lazy and use a 1x16 chip string clocked at 16GH/chip per blade, so about 2.6TH per machine.

The answer is yes for a strict-semantics parsing of the question. Provided the miner is mining, you will get a return on your investment. Whether that return is positive when initial investment and operating costs are factored in, however, is a much different question.

And no, I won't apologize for correctness. I cringe every time I see the term "ROI" misused. It's this community's "their/there/they're".

You got any chips running around?

Gleb, aren't you the guy that busts scams? Sounds like you're promoting one there. I'd refuse to do business with anyone who relied on what you just described. Here in Missouri we're breakeven-cost hosting 50KW of gear and paying the bills fair and square.



Or if you can build your own for cheaper - or at least a better one for the same cost.

So we know there are issues going on with ASICMiner at present. What's the word on the existence of these chips?

I received the miners I purchased today. They're non-PSU so I haven't had a chance to fire them up and test. The dragon was an LCD-less model.

I'm told they're quite a bit more stable, and that AM improved the BE Controller software to suck less as well. It probably still wouldn't hurt to use a USB adapter and run it on cgminer directly instead though.

We've got four on the shelf, and I think we'd be willing to sell two. Probably best to talk to novak, he knows the USB hardware better than I do.

Talk to CrazyGuy. He probably has some available.

I believe load balancing means, if you configure multiple pools it'll spend approximately equal amounts of time hashing on each pool.

Oh. Then I offer that.

Is offering 0.6BTC and a FedEx label considered an insult?

If they sold off their gear to recoup expenses, there'd be a slight dip between when they shut down and when buyers turned those machines back on. Overall hashrate is affected by retired gear, not relocated gear.


That's a good point. I wasn't around for the last block halving so I don't really know what folks were doing then. It'd sure be nice if the coin price went up this time, though really I wouldn't be too upset if a lot of big farms had to shut down. It's fun when high rollers lose to the house.

Why don't I think that? Because the price right now is better than it's been in months and there's still profit to be made. Once the block reward halves, the payout from mining will be cut in half while the costs will remain about the same, but the production cost of those 12.5 bitcoins won't have much effect on the perceived value of the other 16 million bitcoins already available. So no, we don't expect a drop in difficulty right now because approximately everyone is still actually making money from mining. "Using the same argument" doesn't apply to anything that happens before the halving unless the price of BTC were to halve overnight. When that did occur (or nearly so, second week of January) what happened to the diff? Pretty sure it was negative two changes in a row while people turned off gear that wasn't profitable anymore. And then the price went back up, and so did the diff. The diff increases on average if mining bitcoins is profitable on average. The diff decreases on average when profitability decreases. When the block halving occurs, the total supply of coins won't be changed substantially but the dollar value of solving a block will be halved so profitability will take a huge hit, masses of gear will be turned off and the diff will drop. Unless somehow the price also goes up quite a bit, which is not likely since, as mentioned, the total supply of coins won't be changed substantially so the supply/demand curves won't move very far. Unless bitcoin's perceived utility increases, the price will not.