its because we are at a low point in the bubble cycle. If bitcoin goes back into a frenzy again expect coin prices to rise 20-50% under the higher demand.


right now there are still a lot of people holding 20+ coins that are unloading them, without many newcomers rushing in to collect them

looking for a particular price?

The 18gauge splitter works fine and works well with the 16awg leads.  I power an Rk-Box using 2 leads and 2 splitters without issues, and thats for a power draw of ~460W which is more than an S3. The 18awg wire i use is very good quality

I previously had an S1 with a cronjob for this - You can insruct to stop mining, or to even load different asic-freq files at different times of day (ie: you could load up a 300MHz file during the times you want less heat/noise, and then another task that switcheds in the overclocked frequency file at a diferent time of day

your site prices have not changed - any plans to drop prices to compete with bitmain?

same - just got the email this morning.   Bitmain is back on a timely schedule

perhaps hashfast boards since they can do 400-700GH in a chip

by lead i mean bare wires on one end and a PCIe plug at the other. What you say above sounds better suited to my 6" M-F-M Y-splitters that would be great for powering all 4 sockets of an S3 using only two pcie connections

My first batch was 1' leads and i got a lot of feedback that they were too short unless used the way you do. I can realistically order a mix of long and short cables in the next batch, and a 6" lead would likely be about 50-60% the price and cheaper to ship

can confirm that not all DC-powered fans can opperate on reverse polarity. Not all will vaporize into forever-sleep, but some simply wont spin, or will spin strangely.

The BE200 (asicminer) is 40nm. That it is borderline-competitive with 28nm designs demonstrates the effect of a good design

1) yes and no. if you dont adjust and reapply some paste it would likely be worse than it started out. might still function fine but after all the work of remounting it you would have poorer contact than you started with
2) small amount. maybe the size of a plastic BB. dont put much effort spreading it since the pressure of the heatsink should do that for you
3) give it at least 12hrs to cure a little before running the antminer, it may take longer to reach 100% cured but will work anyways

if 110nm is shrunk down to 28nm (1/4 the size) it could theoretically be as much as 16x the efficiency, but other factors may mean something not as ideal. however, 1/10 the power usage could make a very competitive product, so I look forwards to what comes out in rev. 2

definitely hoping to snag a review/testing spot for this

no worries - youve bought cables from me before to modify to suit more specialized orders
Im open to custom requests if the MOQ is >50 or if I beleive that I can sell >50 of the design in a month. By comparison I sold more than 250 of the PCIe leads this past month and very few splitters, which indicates many of my customers are dealing with server supplies

ps: If you are looking for 1-10 splitters I can likely ship them for slightly less than the listed cost since they are a bit lighter than the 24" leads. I am open to custom-quotes based on what is needed on a per-customer basis

^This. I have a pair of these and they run great.

POWER
- PCIE connections, and possibly screw terminals. No 24-pin cable though since its extremely bulky/rigid, takes up space, and can be replaced with a paperclip for on/off. (perhaps include a seperate 24-pin peice that can plug to a mobo cable for on/off function (like this: http://i.ebayimg.com/00/s/NzExWDk0OA==/z/o6YAAOxymnFSF8M1/$(KGrHqNHJEwFIH03lp5JBSF8M06jm!~~60_35.JPG))

SIZE
- stackable with little or no extra effort and hardware (slots together or include zipties)
- preferably two types: 1) small-scale which is less than 1 cubic foot and ~3kg  &       2) large-scale that is 2U-4U rackmount
- packaged in a way it wont bend or snap if thrown down a staircase by UPS/DHL
- able to be placed and run on a metal surface (ie: components electrically protected on the bottomside by a chassis or mounting feet)

EASE of USE
- MUST MUST MUST have its own ethernet connection. no new hardware should require a host computer - it should be self-sufficient with internet connection
- plug and play is ideal - see ethernet connection
- a Raspberri Pi is not a built-in solution, it is a host computer dangling off the side and should be avoided. build in a proper controller and network chip.
- include 3 not-too-bright LEDs for power status, network status, and hashing status. should be visible from 3-4ft away if looking at the device, without keeping the room lit up even at night time

FUNCTIONALITY
- access to settings, ability to reduce frequency, change network settings, and both DHCP and static IP options
- ability to adjust voltage (how difficult would it be to control the voltage of all the regulators with a single pot resistor?) This ensures your hardware can be undervolted slightly late in its lifecycle to make running it profitable for longer
- display stats (perhaps a basic and advanced view modes) for everything from individual chip status to reject/error rates and %

HASHING
- make the chips capable of individual adjustments and optimizations. For a good example, look at spondoolies-tech where each one is independantly driven to achieve ideal performance for each chip. For the BEST example look at bitfury, where the system had a file that saved and allowed users to modify the ideal settings for each chip. (like the SP-Tech method, but saves the best configuration to use after a reboot or miner restart, then adjusts automatically from there if needed)

please build units that are designed at the 2TH/1.5kW level with seperation of the 12V between sections of the system so that it could be easily powered by a pair of 850W+ power supplies. Preferably seperated into eight 250GH/180W sections with an individual PCIe connector for each with no 12V crossovers. Ideally a unit like this would be 4U rackmount

running well for me - no visual differences though.

cant seem to OC though - 300-350 entry in the run.sh file has no effect

even if you could assemble for free you would lose money on this - 850W PSU for $160? You can get 1000W+ for that price brand new


they need to reduce all those prices by 20% to actually liquidate anything while it still has value

if this is 110/130nm technology efficiency doesnt matter much - consider it as a small-batch test of thier design. If they go and replicate into 28nm they could catch up to the market quicly, and the review indicates a stable unit once mining. For a company like biostar playing catchup will be easier than it was for bitcoin manufacturers to oranise and fund thier first die-shrink


who knows - maybe a BTC-240GH is on the horizon at the same power draw (would require 28nm assuming this was 110nm)

realistically i expect to see 9-12% for the next 2 jumps - difficulty is slowing but i think the last two weeks were a nice low hop from variance and the massive 20% jump before that. another jump under 9% would be amazing though, and almost surely lock in a profit for mining