OK... to be clear: I didn't write this review, but as no-one seems to believe it's real, here's it is pasted from the site. To be fair, it's from a site that pokes fun at BTC
so might be a tad biased on the negative side.
Edit: Indeed, BFL says
some photos in the review were of the wrong machine (they've edited the review now, I've included their change below). Read for entertainment purposes only, don't take too seriously at least until reading the comments in the thread, BFL deserve right of reply.
The $22,484.00 Butterfly Labs Mini Rig bitcoin miner is a huge, broken, unstable piece of shit.
Butterfly Labs has a long and horrible history with their mining rigs. They started taking pre-orders over a year ago, with a ship time sometime in late July. After numerous delays in production, shipping problems and general incompetence
, the only thing they’ve managed to get out the door are some of their tiniest miners, the Jalapenos. And those mainly ended up in the hands of reviewers and blogs in order to keep pumping the Butterfly Labs hype train
and securing millions of dollars of pre-orders still in limbo
Lucky BFL forums user Luke-JR however scored a sweet Mini Rig
from Butterfly Labs (it’s just a coincidence he’s a driver developer for them I’m sure). This rig was originally promised
to produce 1500 GH/s hashing power at 1500 watts for $30,000, but has since seen it’s hashing power slashed to a third of what was promised
and it’s power consumption increased 75%, now just offer 500 GH/s at 2400 watts. They’ve promised to make good on pre-order buy sending out 3 rigs to match the initial hashing rate, so now it’s only 1500 GH/s at 6900 watts, a reduction in GH/Watt by a factor of 5.
So what does $22,484 buy you? Take a look!
Minirig is here!
Today, my Minirig arrived.
FedEx apparently dropped it somewhere along the way, and the weakest part of the case, the thin metal part around the back of the PSU, broke.
I’m not sure how sturdy the back side was supposed to be, but its two pieces aren’t quite together either.
The power supplies (EVGA 1500W) also created havoc interfering with the neutral on the power line. This disrupted X10 communication significantly enough that the pool overflowed because the system controlling it was unable to turn off the pump. Workaround: This PSU supports 240V, so we rewired the outlet. 240V does not use neutral, so now all should be okay.
Edit: 240V workaround is only partial. Still having problems
But the good news is, it all seems to be working for the most part.
Next up, installing it in the window so the heat goes outside
A twenty two thousand dollar box of electronics that is broken out of the box, that required the guy to do a sketchy electrical workaround to get partially working, that he is going to install in a window… and he’s happy about it?
In case you didn’t notice it, the delivered unit is different than the picture on the website. They had to install 2 power supplies instead of 1 and had to modify the case to fit. Also, if you didn’t notice, the LCD/Phone thingy in the front has been replaced by … a piece of cardboard spray painted black. Wonderful.
You could maybe chalk this up to a careless Fedex postman, but when you’re shipping something that costs as much as a mid-sized sedan, how bought putting a little more effort into packing? Dell and HP can ship bigger and heavier servers across the world without this kind of problem.
The unit had to hit its huge power draw increase by putting dual EVGA consumer grade power supplies in the unit. We’re talking almost a 75 amp load (6*1500/120), disregarding power factor. He could very well overload the circuit panel and trip the main breaker for the house.
Let’s take a look inside this guy.
This is from an earlier version of the Minirig (note the single power supply)
This is apparently from an earlier FPGA but it will give you a good glimpse at what kind of craftsmanship you can expect from a computer that is half the average household income in the United States.
Consumer grade PSU and cheap USB hubs glued to the inside case.
Electrical tape and random velcro glued to the insides
A closer look at the USB hubs. Plugs are hot glued to stay secured.
Electrical tape everywhere, splices and voided hardware are the theme.
You can view the entire album here: http://imgur.com/a/Uanjr#23
Despite all that, this thing can still mine bitcoins and it should be profitable. How much money can we expect from from a rig like this? If the difficulty didn’t change, they would make 37 bitcoins a day and recoup the initial investment in 124 days. Difficulty is jumping pretty much 20% every 12 days or so, so in the next week before adjustment, they’ll make 259, the next 12 days 369, the next 12 days 312, then 256, then 213, etc.
So by day 127, they’ll be halfway to breaking even, but by day 151 they’ll be making less than 5 bitcoins a day, and even if difficulty stopped rising at that point(which it won’t), it would take another 435 days for a total of 586 days to break even. If difficulty kept rising at the same pace, by day 200 they’d be making 2.4 bitcoins per day, and it would take 1024 days to break even with no difficulty increase. Assuming 25 cents per kw/h, and $100 a bitcoin, it would cost 0.43 of a bitcoin per day in electricity which means the unit would no longer be profitable on a power usage basis by day 307, at which point it will have produced 2620 bitcoins.
Bear in mind this is only for the first few units, and that’s running 24/7 pumping out around 24,000 BTU, so yes, medical bills from heat stroke will be on top of that.
But Alas, the chips don’t run nearly as well as they’re supposed to, frequently running too hot and giving multiple hardware failures. Coindesk noted
in one of the first ever runs of the Minirig by hosting provide gigavps that it was running much too hot and erroring out.
At the time of posting, gigavps warned that the unit would be repeatedly shut down while ckolivas, who was assisting, modified the machine’s software to optimise performance. After some tweaking, the device was said to have been left to run continuously for two hours, and was shown to have an average hash rate of 478.1 GH/s. As you can see in the table below, ASIC number four (of a total of eight hashing chips) ran significantly hotter (86 degrees) and consequently gave the highest hardware (HW) error rate.
So, what happens if you just decide you don’t want this, you don’t want to wait over a year to get a $22,000 broken piece of shit? Nothing, because BFL won’t let you cancel your preorder because they’re now “shipping”, i.e. they sent out one unit to their own company shill.
Which is of course illegal
regardless of what Butterfly Labs may say.
So in summary: Don’t buy anything from Butterfly Labs … ever.
Edit: there's one comment on the site so far, which is worrying reading. Is there any truth to this?
This is incredibly dangerous since it is doing neutral line loading. If that is fucking with a dudes pool then it is sourcing current from neutral which is a big no no. It is a really easy way to start an electrical fire. I am actually wondering if this has to pass UL or CE testing to ship legally in the US or Canada because when it starts a fire in some dudes house and he finds out he can’t claim insurance because he was running this he will be unhappy as hell.