Location was the problem for me(USA based), otherwise I'd have bid on things.

I'm running 10 5830s across two power supplies and the system draws just over 1500 watts. Have a damn hard time achieving greater than 2 MH/W with these inefficient cards

Interesting, never really thought of nailing boards to wood to secure them. I myself use enamel coated wire shelves from Target with a rubber foam layer to keep things sliding between shelves and messing up the works.

The discussion on the Cyclone V launch got me thinking. Besides the fact that the Spartan LX150 is cheap on a per unit basis, is it the best option on a MH/$ basis? For example, could we be better served by a gigantic $1500-$2500 FPGA? I understand the first FPGAs to be used had to be cheap on a per unit basis as there was no established market. However, the BFL Single thread in the marketplace describing the number of orders just shows how much of a market we have now. This all stems from my interest in seeing something in the 1-5 GH/s range that bridges the gap between Icarus/X6500/Single and the BFL mini rig.

If it is a case that there are no economies of scale in purchasing bigger chips then that's a pretty quick and easy answer.

So, now it is no longer a maybe but a fact that this unicorn board has a two year warranty? I'm done arguing with you as clearly your board is the superior choice.

I think you are compromised on this due to personal interests in said device. Its competitors are FPGAs and GPUs and at this point it has longer pay-off times than most GPUs and some FPGAs. I suppose in the end the critical matter is whether you view bitcoin mining as a hobby or as a business. As a hobby, less than prudent financial decisions are acceptable.

You've said that about a dozen times now. We get it. It may be two years. BFL may also ship in 4 to 6 weeks.

BFL's Single, at current rates and ignoring the reward drop, will take me 9.3 months to pay off at $0.11kWh. Your board, at current rates and ignoring the reward drop would take me 16 months to pay off. So, either box will be out of warranty by the time they pay themselves off. More importantly, for a few bucks more I could get a pair of BFL Singles.

Also, the jump between GPU to general FPGA efficiency is massive. After that we are squabbling over a few bucks a month which at this point are a tiny fraction of the device's overall cost. It takes 98 months for the BFL Single to consume its original price in electricity. For comparison it takes 10 months for a 5830 to consume its original price in electricity. Until you start talking another jump like that then the differences in power between FPGAs are inconsequential for some of us.

Very interesting. Also interested in the actual load as I don't believe the input power for the memory, while tiny, is included. Might be time to pick up a used 7970 myself.

Icarus is no longer available. Who knows what Lancelot will be(maybe 4x Spartan lx150?) but I believe it will be announced in May. ZTEX is expensive per MH, same efficiency as Icarus, but has a warranty. The X6500 is a fair compromise between Icarus(in that it is available) and ZTEX(MH/$). All these are based on the same FPGAs so they achieve the same MH/W efficiencies.

BFL's Single is a great option in the MH/$ range and it bridges the gap between Spartan LX150 based boards and GPUs(at 10MH/W vs 20MH/W and 2MH/W for LX150 & GPU). The issue is who knows, if you order a single now, when you'll actually receive it and which revision you'll receive.

Lastly, there is the issue of resale/reuse value. I think everyone agrees that the BFL Single is a bitcoin only box at the moment. Icarus could possibly be used as a dev board(people tout Icarus' IO pin count). I have no clue on ZTEX and it sounds like X6500 has a very low IO pin count.

I might have missed a FPGA and the above is all my opinion on FPGAs. Some of it could be wrong.

I'm surprised and giddy. It is awesome that someone stepped up for this, going to keep tabs on this as I'm curious how it plays out.

One of my cards is really finicky on the memory clocks and I got tired of fiddling with what is otherwise a solid rig. I'm going to play with it again in a few weeks and isolate the finicky card/core when I feel like I'm not stressing enough about my GPU farm.

It isn't quite the same but I run 4 5970s on a Corsair AX1200. Right now it is pulling around 1150w from the wall with the GPUs at 800/500. The GPUs are watercooled(~40C core temps) and I'm not sure how much the cooler core/vrm temps make a difference on power consumption but that'll at least give you a ballpark figure.

For a reference point I did a trade with ragingazn for some 5870s and they arrived today. Will test them tomorrow but they look undamaged.

Interesting. If someone can get over the financial hurdle of making bulk ASICs their $/MH will in theory be so low and their MH/W so high that they could force out the vast majority of current users given enough capital. Could make for a very interesting future IF this were to take off. If they were the only ones with access to very cheap and efficient hashing power it could in theory break the network.


Would be fun if the next currency that cropped up rendered their ASICs useless 

I wasn't going to mention a Halon system just due to the unrealistic costs for a mining operation. They are very neat systems though  I wonder though, a CO2 suppression system could be built for maybe $1-2k. It would require some coordination between ventilation  and solenoids but it would not be a difficult system to implement. Ideally a small room/enclosure, a intake/outlet ventilation system with a system of shutters to block the intake/outtake, pressure regulation manifold to link and meter multiple CO2 cylinders into the chamber.

The other option would be liquid nitrogen. I have some experience with LN2 solenoids and I could easily/quickly vent a 180/230L dewar in to said room. From the wiki "Since the liquid to gas expansion ratio of nitrogen is 1:694 at 20 °C (68 °F), a tremendous amount of force can be generated if liquid nitrogen is rapidly vaporized." You won't get quite that high of an expansion ratio as it'll be injected in to the room in a liquid state and quickly drop the ambient temperatures but if done correctly you could gradually expand the LN2 in the delivery system. Perhaps have a finned heat exchanger like used to generate dry gaseous nitrogen from LN2. Place that in the room and just vent the nitrogen in to the room.

Get a normal fire alarm. Check to see the voltage and current that goes through the buzzer and find a solidstate relay that fits within those limits. Use this relay to trigger larger mechanical relays to cut out your power across your breakers. A better option might be a small arduino controller that monitors these pins, sends out a SOS when SHTF and then triggers relays to cut power. All just ideas thought up while reading this but there are many different approaches to this. In the end something battery backed so in the event of an electrical fire causing a breaker to trip you don't lose your watchdog.

The chemical temperature triggered extinguishers, if placed correctly, will work also. They'll save the building but effectively destroy your gear.

Awesome, thanks for the response. I'll wait about patiently 

Not worth the trouble. Who knows if another card will exhibit the same behavior but it was masked due to the more finicky card. 300 MHz on the ram has been working fine for about 2 weeks now and I got tired of fiddling with it.

Any chance that there are more invites being sent out? Would like to play around with this as it looks rather slick.