FOR SALE: 16AWG PCIe leads, ideal for Server PSUs! Also, 18awg Y-splitters! 24" LEADS6-pin PCI-e connector (black) 30cm (24"), 16AWG wires (yellow and black) stripped ends on wire (0.5cm) These are solid connectors and far superior to the garbage 18AWG connectors found on ebay. I have yet to test them on hardware, but there should be absolutely no problem carrying 250W or more (potentially upwards of 350W) through these leads. (In comparison, I have seen other cheap 18awg wiring melt or burn at over 150W) 6" Y-SPLITTERS(M-F-M) connection, with each branch of the y being 6" (15cm) long. perfect for devices like the antminer with a little bit of slack to spare. 18awg - should handle 150W without issues, and possibly over 200W. Please use discretion though, if the cable gets too hot it may not be suitable to your application SHIPPING FROM TORONTO, CANADA 1-4 units: $3.50 each 5-10 units: $2.80 each 11-19 units: $2.40 each 20 units: $2.10 each 50 units: $1.80 each 100 units: $1.50 each 200 units+ $1.40 each SHIPPING:CANADA1-4 units: $6 5-10 units: $10 11-19 units: $14 20 units:2-3 day tracked expresspost : $25 3-6 days untracked : $16USA1-4 units: $10 5-10 units: $13 11-19 units: $16 20 units2-3 day tracked expresspost : $40 6day tracked expedited: $27SPECIALS: 4 cables and basic US shipping = $25 8 cables and basic US shipping = $40 16 cables and basic US shipping = $60 I might be able to get the shipping rates a bit lower on certain quantities depending what the post office can offer - feel free to ask for final costs on an order BTC and Paypal accepted
|
|
|
Anyone know how many "blades" can run on Rockminers Rockminerweb? I've got a new style R-Box running alongside my RK-Box via Rockminerweb v0.9.3 and was wondering if I could just keep adding more R-Box miners??
at least 8 (I can run 2 rkboxes on both an RPi and on a cubie a10), i imagine more can be handled as its effectively cgminer and thus shouldnt be capped in the software. Bigger issue is how powerful the CPU of your interface (RPi or cubieboard?) is. A cubie would likely handle more units than an RPi due to more powerful core
|
|
|
I'd love to see whatever they're coming out with be a direct drop in / swap for the current systems. As in unscrew the old boards, swap the heatsinks to the new boards, and screw in the new ones.
would be nice, though i expect the housing doesnt cost much to produce, maybe just some shipping expense. anyway to get a better power consumption from these? 1.2w/GH is becoming too much power usage for where i live, and I'd be happy to lose 25% of my hashrate in exchange for <1w/gh Yeah, I was wondering the same...mayhaps we can pencil mod these similar to the S1's or something! depends on the regulator. I tried to pencil mod the rockminer rkbox/r3 hashboards and the regulator setup on it did not allow for pencil modding in any of my attempts. The best I did was to determine a pair of resistors that must be simulateously modified to adjust voltage. Unfortunately, reducing the resistance of both via a pencil mod INCREASED voltage, meaning a decrease involves the more difficult task of increasing the resistance. Havent done any experiements on the BTCgarden board, but schematics or at least identifying the regulator would give me a place to start. The risk to a single board is less than the risk of running 1.2w/GH for much longer
|
|
|
8 units in series? How is that possible without the water being extremely hot by the time it reaches the last few units - what is your outlet temperature? more importantly - what pump and radiator are you using to achieve this, and is it water or a specialized coolant?
|
|
|
Thanks to everyone who contributed to this discussion I totally forgot to take some close ups of the Dell 2100w PSU. In the end I soldered the cables onto the terminals, this was instead of using spades. The terminals are slightly tapered so spades are not secure enough. The connecter blocks worked a treat. The only thing is the PSU is quite loud, but thats not a problem because its in my Dads loft which is 22 miles from my house lol. Currently there all OC’d at 250 and combined its hashing between 1900 – 2000. One thing I have noticed is one miner is running at 430GH not sure why quess ill have to research that one. I’m back over there in two weeks so I’ll take voltage readings and provide and average Watt/GH. Buying some more in a few weeks. BTCBTCBTCDon't you need to plug in the other two PCI-E connections if you overclock? When I first started this tread I was under the assumption that the card required extra power though each connecter, when in reality this is not true. Yes the card does require more power when OC’d but this can be drawn from one connection providing the cable can take it. I believe the reason its recommend to plug in all four when over clocking, is because a standard PCIE with 18 awg cable will get hot and therefore burn out. I have made my own cables using 16 awg and then from the connector block that connects to the PSU terminals I used 14 awg, even after 4 hours the cables and PSU were stone cold. ^dead on. 1 connection per blade is sufficient when using a proper cable for the job. one additional thought though - by using the second connector as well, it better distributes current on the PCB. This would MARGINALLY reduce resistance in the PCB and ensure the area around the single connector doesnt get warm. However, no issues or PCB failures have been seen yet, and any power savings would be on the order of 1-2w/PCB - so its not really necessary to plug in all 4 connectors unless you actually happen to have enough cables to begin with (which most PSUs wont - such as the EVGA 1300 having 6 cables only (2 of which are double connectors though)
|
|
|
it wouldnt be surprising i think - the S3 prices are unreasonable for anyone who pays >$0.10/kwh and havent been dropping much, almost as though they either have no issues selling at the current price, or simply have a limited supply thats dwindling at just the right speed.
Its known that 0.3w/GH gear isnt far off, and that a well-designed 28nm chip is capable of 0.3-0.4w/GH, or even that the current 28nm chip might offer ~0.4w/GH if aggressively undervolted.
the biggest clue in my mind is that the S3 units cant be selling very fast right now - as anyone outside china will have a hard time getting ROI with the shipping prices. I havent bought hardware in about 2 months for this reason
|
|
|
2.5BTC invested. Would invest more if there was a more fixed form of insurance/assurance for the fund, as even 50BTC of investor funds could become tempting to dissapear in exchange for the loss of a <1yr old forum account.
not sure how to do that though, without effectively removing those bitcoins from the arbitragable balance. maybe some partial escrow holding of a 50% float or similar?
|
|
|
I never got my sig campaign money for October. Did everyone else get paid? I'm supposed to get paid on the 5th and no replies to the PMs I sent. I'm wondering what is going on?
maybe because everyone is getting sick of users whoring out themselves for pennies in advertising money? Everyone knows about the prisma - If anything ill avoid buying it because of the 5+ ads for it i see in every page of every thread recently. not to mention actual reviews indicate an unreliable unit with RPI issues
|
|
|
But how do they keep the moisture/humidity out from damaging the equipment??
hot surfaces = evaporation. cold surfaces = condensation miners and heatsinks are hot surfaces, so theres no water on them to cause shorts
|
|
|
I have no clue why the pre-fire picture couldn't have the ethernet cabling nicely clustered and numbered, running in straight lines with the shelving rack - it would be tidier, safer, and easier to service.
The answer is simple: amateurs don't know how it should be done, they think the way they did is just OK. If they knew better, they would do it properly. Amateur hour ended when they set up and operated (safely and at good temperatures) 450+ units for over two months. -yes, the web of wiring is ugly, but the obstruction to airflow is pretty minimal, especially as its hard to tell in the pictures but there appears to be a 6" or greater gap between the wires and units. -power wiring on the back side was cleaner it seems. -area is clean, freshly painted, and had at least a few fire extinguishers and a shutoff switch. This was not a novice operation, and IMO/IME there is no problem with using crappy basic shelving - it does the job and doesnt obstruct airflow. I dont think the setup was in any way the problem. the problem was: 1) whatever component started the fire (a unit, a fan, a power bar?) 2) The fact that 2 IT workers were sleeping. If one was awake as they should have been the shutoff switch at the transformer would have been hit and the fire stopped dead in its tracks (possibly without even needing an extinguisher). Apparently it took over an hour for the fire to spread according to the dropping hashrate of the farm. 2b) smoke alarms or other automated systems must have been absent or insufficient. It's not just the obstruction to airflow issue with that ethernet wiring, it shows the general sloppiness together with absolute absence of attention to detail. Anyone who wired 450+ ethernet devices like that most probably never wired anything more complex than a home router, it was bound to cause at least some network glitches. Network cables should not carry it's own weight, and certainly not the weight of hundred or more cables meshed into a spider net, with the whole combined weight of all of them carried by whichever is the shortest one of them, certainly giving some tension on the motherboard it was connected to. It sure was not the cause of the fire, but is obviously wrong to anyone who wired a single rack of equipment. We can only assume how they did the power distribution, but looking at the front side it would be surprising if there were not some very elementary errors which directly contributed to what happened. the post-fire images indicate that all the power cabling on the backside was bundled far more cleanly. The ethernet was a mess but honestly i dont think its the issue being made out to be. They appear overlapping, but not tangled, so its not like 10lbs of cable is pulling on the shortest link. furthermore, to run wires all the way to the floor, then to the networking rack and back up again would probably double or triple the length of some cables used, which really just leads to more mess of its own, and being unable to follow individual cables when 50 white cords are ziptied together. A lot could be improved, but until the actual cause of fire is revealed its just a lot of guessing - with most being accusations of the wiring or hardware, when the fire could have been initiated by anything.
|
|
|
An Inert gas fire suppression system would ruin ROI for any Bitcoin mine but surely if you spent a few million on SP30's you'd buy a few $5 fire alarms and have a security guard?
I think it's an insurance scam but they forgot to buy any first.
You could fry an egg on top of an SP10 I certainty wouldn't put 2 on top of each other.
SP10 had the heatsinks bolted to the bottom of the frame, heance the case being so hot. the SP30 'hovers' so its case doesnt get remotely as hot to the touch. and most fire suppression systems would either damage the miners, or are designed for enclosed data centers (you cant just sray inert gas into a windtunnel - it would be pushed out of the building in a heartbeat by all those fans
|
|
|
bump to the top! This subforum buries things quickly
|
|
|
Hi, any sort of "refrigerator" or "air-conditioning" system will not be enough to cool your average Bitcoin miner. The most important thing is bring in fresh, outside air into the unit and taking the hot air away as far as possible, so high RPM large fans are the best way to go. The most important thing is the amount of cubic feet of air per minute the fan can move.
Source; Running a mine in Australia, Summer :/
Ossy guys I hear a lot about evaporative free cooling over there in Australia. That kind technology is easy to implement and cost a lot to build and operate. Would that be solution for bitcoin mining. Their will be 1 pump and I think we can use the miner fan to pull in the air and a exhaust fan to take out the heat. What do you think of the idea? i think you dont know whether to use refrigeration systems, immersion cooling, or evaporative cooling, yet claim to have thorough knowledge of each. The point is that cooling can be different for every location, but refrigeration tech is 100% the least cost-efficient If you are an expert, please send me your CV and we can talk about a job offer for you over here to Thailand to help me a perfect cooling bitcoin mining. FYI. This cabinet doesn't use refrigeration, it use precision cooling data center with airflow optimization front to back. build an ope-air system with high airflow. use evaporative cooling. Look at cowboyminer's farm (pre-burning to the ground) for an idea of how to operate high density without cabinets. If you want quiet high density, look at datatank. and no, im not an expert with a CV in HVAC- only an engineer happy working in canada.
|
|
|
Hi, any sort of "refrigerator" or "air-conditioning" system will not be enough to cool your average Bitcoin miner. The most important thing is bring in fresh, outside air into the unit and taking the hot air away as far as possible, so high RPM large fans are the best way to go. The most important thing is the amount of cubic feet of air per minute the fan can move.
Source; Running a mine in Australia, Summer :/
Ossy guys I hear a lot about evaporative free cooling over there in Australia. That kind technology is easy to implement and cost a lot to build and operate. Would that be solution for bitcoin mining. Their will be 1 pump and I think we can use the miner fan to pull in the air and a exhaust fan to take out the heat. What do you think of the idea? i think you dont know whether to use refrigeration systems, immersion cooling, or evaporative cooling, yet claim to have thorough knowledge of each. The point is that cooling can be different for every location, but refrigeration tech is 100% the least cost-efficient
|
|
|
I have no clue why the pre-fire picture couldn't have the ethernet cabling nicely clustered and numbered, running in straight lines with the shelving rack - it would be tidier, safer, and easier to service.
The answer is simple: amateurs don't know how it should be done, they think the way they did is just OK. If they knew better, they would do it properly. Amateur hour ended when they set up and operated (safely and at good temperatures) 450+ units for over two months. -yes, the web of wiring is ugly, but the obstruction to airflow is pretty minimal, especially as its hard to tell in the pictures but there appears to be a 6" or greater gap between the wires and units. -power wiring on the back side was cleaner it seems. -area is clean, freshly painted, and had at least a few fire extinguishers and a shutoff switch. This was not a novice operation, and IMO/IME there is no problem with using crappy basic shelving - it does the job and doesnt obstruct airflow. I dont think the setup was in any way the problem. the problem was: 1) whatever component started the fire (a unit, a fan, a power bar?) 2) The fact that 2 IT workers were sleeping. If one was awake as they should have been the shutoff switch at the transformer would have been hit and the fire stopped dead in its tracks (possibly without even needing an extinguisher). Apparently it took over an hour for the fire to spread according to the dropping hashrate of the farm. 2b) smoke alarms or other automated systems must have been absent or insufficient.
|
|
|
start counting!
approx 8*3*3*6 (plus another 8*3) = ~450-460 units. This is one building, apparently 3 were in the compound
|
|
|
True when they called about the fire. I can see in the miner stats that the fire was already 1 hour going. So they where all sleeping ;-). and stupid me trusted them with all this equipment. That's my big mistake. But again, I can't turn back time. No one hurt and a lesson learned.
Man, you are VERY passive for losing a ton of money and profit. You must be some rich kid playing with mom's money or never earned the money yourself? Usually when you earn money and lose it, it hurts a lot more then when just given to you. Thats a ballsy statement that I extremely doubt is the truth. An operation of this scale and design is certainly not the first iteration of mining farm, and its likely built entirely from the profits of previous mining (maybe as far back as GPU?) You win some, you lose some. Its unfortunate that this was such a loss this time
|
|
|
Are there any further plans with this project? Further expansion of the hashingpower? Any news or prospectives? There has not been any expansion in over a year - and there wont be any more expansion.
|
|
|
positive roi? nothing!!!
Don't think so. How would big data center make their profit in this crytocurrency mining business then? 1) by not spending $5000 per specialized rack that isnt necessary 2) by not spending aan extra 30% on power for each of these specialised racks, then still having heat to cool yet again. its not complicated. Mining equipment blows the heat out the back of a rack. Your equipment contains this heat, cools it at PUE 1.3 or worse, and then dumps that 130%+ heat OUT THE BACK OF THE RACK. This idea would cost more, weigh more, draw more power, and create more heat. I do not think there is a single benefit other than perhaps a slight noise reduction
|
|
|
If you have ever been to Thailand and seen the electrical wiring system there, you would shake your head. 2 wire system with no earthing is very common. You look at the power lines on the posts, and you cannot count the number of cables that are strung there (often tv, telephone transmission cables) but there are cases where electrical lines have been low enough that passers by have been zapped in close proximity. Also, MANY people have been zapped merely walking on sidewalks for improperly earthed street lamps beside them. And many kids have been killed by electrocution in pools, school yards etc. Terribly poor electrical work done there very frequently.
residentially thats not surprising, but for a compound drawing 5MW it requires significant infrastructure in place that is likely free of these flaws, at least for the most part. and it wouldnt be difficult to install a sufficient grounding system on-site if the power delivery poles did not include a grounding. Drive a few heavy stakes deep into the ground at strategic locations, and wire them all as the grounding system. fairly common
|
|
|
|