Looking for loose Bitfury chips. Offering $5 for each in BTC. Need 15 or so. Prefer the ones the have Bioinfobank etched on the front as they match
the others on the boards I am repairing. Chips must be new.
Thanks!
|
|
|
ROI on these? at .75 Btc at avg 20% increase and 10 cent KWh
do these ever make the btc back?
.015 BTC per day should ROI in under a 100 days. I already had the PSU's from S1's so no additional costs there. 0.75 BTC is the hairy edge IMO. NOW where are the COUPONS!! Id check that math again...They will NOT roi if difficulty increases @ 20%. Well Christ I do not believe Diff will continue to rise @ 20%, as of this increase network has fallen on it's face and is running under 6 blocks per hour and has been for almost a day. That 25% increase KILLED the last of the FPGA/GPU miners. Anyone running a BFL single is going to be shutting them down SOON. My 2 60's earn approx $2.04 USD per day, but cost me $1.15 USD so they are on eBay as we chat. I believe these will make money for anyone that: 1) pays no import duty 2) has the psu's already 3) pays under 16 cents a kwatt 4) can do the summertime cooling without ac I meet all of those criteria but I think I will still wait.
|
|
|
ROI on these? at .75 Btc at avg 20% increase and 10 cent KWh
do these ever make the btc back?
Most likely they will not. I will wait until they are shipping same/next day. Maybe they will be cheaper then. I thought about buying about 10, but them not shipping until July 10th just doesn't sit well with me.
|
|
|
I see one for 22.5BTC. That's like 30 units!
|
|
|
I purchased dozers boards. I will post in this thread when I receive them. I hope I will have some extra controllers for sale.
|
|
|
I believe the issue OF LRP is being addressed and is entering alpha testing. Hopefully sooner rather than later. I would LOVE to get my ants back leasable again!
|
|
|
I am still fighting hardware/spi errors. Rebuilding all of the cables that link the boards together. Seems some of them are miscrimped and have opens! This can cause crazy high errors. I like to keep errors below 2%. If you don't want to torture yourself and rebuild these things I would suggest buying new ones.
|
|
|
Regarding lack of standoffs, I have found laying the boards down and using two on the bottom to connect them works well. For me, most of the dinky
40mm fans where toast anyway so I removed them and use three 120mm fans per bank of 10 boards. I reversed the cables and run them under
each board and the fans blow in from the cable side. This allowed me to use half the standoffs AND get better cooling. They take up slighty more
I tried this - and the boards that i removed the fans from - seem now to get way too hot. I pointed 4 120mm case fans at the bank of 5 boards to see if it helps How much air do your fans move. I am using fans that move about 60-70C cfm. Three per 10 boards. Most of the heatsinks are just warm to the touch. The fans aren't quiet but where they are this doesn't matter. Don't worry though these boards are not very heat sensitive.
|
|
|
You can order standoffs from some of the dx and aliexpress type sites - or several sellers on amazon have them. search for 3mm standoff or something like that and pick the ones you want.
I've gotten about 11 boards up and running anywhere between 400 - 500 ghash on cgminer.
I'm going to play around with some reconfiguring of the stand offs - as suggested by helipotte. As well as try to reduce the number of boards per chain.
I'm not sure what was meant by the back feeding on pin 10 so I likely won't be doing that.
It's clear that one of my first mishaps was to connect boards with the pins down and boards with the pins up - as well as then not worrying about whether the pins were up or down on the controller card.
Pin 10 is the SPI bus voltage. What I did was run a wire from the control board 1.8Vdd to the empty connector (spi out) on the last board in the chain. I noticed the voltage drop was the greatest on the last board. What connecting a wire does is basically a pull-up of this voltage. The gauge of wire used in the 10 pin cables is, in my opinion, to small for the current required by more than 48 chips. This leads to SPI errors and chips being marked as bad/slow by Cgminer.
|
|
|
How do you do this with only a 30Gb disk. I just checked the blockchain size on my node and it is 30.1Gb. The blockchain has gotten huge!
|
|
|
Got some of the boards working - the ones with the pins facing down - hooked to the controller with pins facing down. Couldn't get chainminer to work for a darn - so I'm using cgminer for now. More HW errors than I'd like but I'm hoping that can be configured away.
I have 11 of the 18 boards hooked up now. Based on space, power and the lack of standoffs - I have to get some more parts in to get the rest working
Regarding lack of standoffs, I have found laying the boards down and using two on the bottom to connect them works well. For me, most of the dinky 40mm fans where toast anyway so I removed them and use three 120mm fans per bank of 10 boards. I reversed the cables and run them under each board and the fans blow in from the cable side. This allowed me to use half the standoffs AND get better cooling. They take up slighty more space but I have plenty of that. I will post a pic soon. Works really well actually. Edit: Finally got the errors under control. Seems more than four boards drops the SPI bus voltage too much. Back fed this into the last board on pin 10 and no more errors! I noticed that with 5 boards, the voltage was only 1.55 at the last one. some chips apparently don't like it this low and send out a lot of bad data even though they are hashing with low errors.
|
|
|
is it worth making available a arch linux image with compiled cgminer for these?
Raspbian is working well for me. Just imaged it to a class 10 SD card and complied cgminer on it. Getting about 785Gh out of 20 cards. They do seem to glitch up sometimes and chips get set to bad status. A Cgminer restart clears them. Working on fixing this. I tried Minepeon with updated cgminer but it seemed to require a lot more of the CPU. It would always max it out and slow down Cgminer.
|
|
|
Has anyone got the blue boards to run yet? I cant get any of them up.
Most of my boards are blue. They all run OK. Just check your controller. Blue cards need the SPI connector to be on the underside for the pins to match. If in doubt, check with a multimeter in continuity mode. Be sure to check pin 10, this is the 1.8V bus feed. If you connect it backwards, you risk burning up the controller card! I am not where the documentation I made is but I figured out the pinout. I think it is: 1 MISO 2 ground 3 MOSI 4 ground 5 SCLK 6 ground 7 NC/external clock 8 ground 9 ground 10 1.8Vdd bus This is from memory, will verify later, but I am sure about 9 and 10.
|
|
|
To update, I have been told this would be very difficult to do. This is mostly due to the extra GPIO pins used to interface with the control board. Also the a beaglebone might be less suited for the task. Still hoping the some one can help me with the driver.
|
|
|
I'm only trying to get a chain of six to work - i get 100% HW errors. I tried chainminer - couldn't figure that out - so I installed cgminer using the kanos directions.
It recognizes the boards - sometimes up to 93 chips - never all of them of course.
I'm powering the Pi using a usb cable. I am not powering the controller board as there is no way to do so - it has an odd 4 pin connector on it that looks close to a molex - but the molex from the power supply wouldn't fit into it. Not to sound like a n00b but I'm stumped on that one.
Some one referenced overclocking the pi itself to 900mhz - is there a quick and easy way to do that?
As far as overclocking to Pi, the easy way to do it is to connect the SD card to another computer and edit config.txt. Or do it from the shell. Example: http://elinux.org/R-Pi_ConfigurationFile
|
|
|
I did use chainminer, briefly to compare against Cgminer. I prefer Cgminer because I monitor it's API remotely. I will try to post pics of my modifications
to the control board, when my second Raspberry Pi gets here.
|
|
|
Do these boards run slower in larger setups? I have had a problem with boards running much slower when 15+ are connected to one controller. I am using raspbian/Cgminer4.3.4. I tested ALL of the boards individually and have 20 of them that do 40Gh+ by themselves. This is 2.5Gh per chip. When they are chained 5 boards to a bank, some them only manage 1.0-1.2Gh. The errors look good, rejects look good but speed is way low. They are cooled by 120MM fans running at full speed and 1500W of clean power(wired together by 14 gauge wire). Seems that it is a bus issue. Maybe a CGminer issue? I posted earlier some suggestions, these worked to cut errors/rejects but now speed issues. I am starting to think these boards are POSSESED! Thanks for any advice/feedback.
|
|
|
Is anyone else having trouble with Eligius and disconnects using LRP? I have been having a time with this. I am now using another pool and all is working great. Eligius and P2Pool just won't work with LRP. There appears to be some hard-coded buffer size in LRP that drops the end off the work requests, which is why it works for a bit (at the start of a block) and then suddenly starts failing. I've had a ticket open about this since before it was sold. That's why I don't use LRP any more for AusMiner. Just the answer I was looking for. Thanks. I guess my Ants aren't rentable until this is patched. Is this issue ever going to be patched? Thanks. Why would you want to use LRP vs direct for an ANT? It is not one, but four. It is also WAY easier that dealing with four separate leases. Also my IP is not static, and changes with out warning. The proxy simply provides better service in this particular configuration. My other gear is direct, but has a staic IP.
|
|
|
|