Yes I had previously done that calculation as well. The only thing I can think of is if you are (or maybe your employer is) already paying for a rack that is underutilized power-wise but has a little space left you can add a few of the miners. I couldn't give a rats ass about nanometers or "sexy" miners (seriously you guys need to get out more). All I care about is price, delivery, reliability, hash rate, and power usage.
Obviously, I don't have to agree with you 100% (except for the part that I should get out more).. but you're right on the money with that fact that the ASIC doesn't need to look "pretty". To get the job done is the right attitude and it's not like we'd look like poster boys ourselves either. Furthermore, it left me thinking about the 1.25U competitor.. If they consume between 1.2 - 1.35kW power, why bother with such a small factor (and risk with not being able to dissipate the heat effectively, hence more noise for fans spinning at max)? Yes, it is the coolest looking kid on the blockchain, but as most Data Centers can normally only facilitate up to between 8 to 12 kW of power per rack (including the equivalent in cooling), where lies the benefit in 1.25U when you can only fit 10 of those into a 42U rack maximum, leaving 70% of the rack unpopulated from a density point of view? Data Centers that could facilitate more power are few and far between. In this sense, BITMAIN seems to be spot on with the sizing - but it does come down to the price and how well they can position themselves on the market. No doubt more competition is coming and the big names of the past are looking to regain their position. What sort of servers are used that have higher densities of power than the usual rack? (fitting with the scenario you suggested above). there are examples of server racks that are 3 antminers/4U = >30 antminers or 12kW. I assume with enough airflow the high power demand can be handled. Most racks though are not often equipped with more than 1 or 2 6" PDU units, generally capable of 4-6kW each depending on the outlet style. However, adding another PDU or simply employing multiple shorter but equally powerful PDUs would at least deliver power to the rack. Its also quite possible not many datacenters expect such power density, and may not have enough available outlets per rack even if airflow/cooling are not a limitation The 2 DC that I'm in are completely modular, I can order multiple 60A 3-phase connections if I wanted, so power density isn't an issue. I currently have 2 30A 208v single phase per rack, which is enought for 27 S1's overclocked, and the plan will be for 10 S2's. Depending on the DC and their rules, you may be able to go above the 80%, at mine, the only thing I lose is the uptime guarantee if I trip my own breaker. 2 of these gives you a 10KW rack http://www.newegg.com/Product/Product.aspx?Item=N82E16812120346That PDU doesn't take up very much space. If density required it, 4, 6 or even 8 is completely feasible, however if you're running a 40KW rack, you'll probably want to think about 3-phase.
|
|
|
^^^ Look awesome...
What is the power consumption?
Each rack consumes 48A @ 208V Single Phase. (24A each PDU) Each PSU powers 6 Ants, each bank of 3 ants is fed with a single 4awg wire (single + and single -) to the distribution block.
|
|
|
Love the modified server PSU, I've considered using them, but concerned over a potential increase in fire hazard. I know those dells have pre-configured kits, but didn't know if it can cause potential issues as I read a couple people saying they had hot wires to the touch.
If wires are hot, they are probably not thick enough. Hmm, third hit says 18Gauge is good for 16 amps.. would not want to run more than 5 though that. This pages seems to agree, but qualifies that with a maximum length. I would put the practical limits (based on experience) around: 18AWG (Quality product such as PSU cables) - 10A maximum, 8A is safe 18AWG (chinese mlex adapters and other flimsy wire) - 6A maximum, 5A is safe 16AWG - 15A max, 12A is safe 14AWG - 22A max, 17A is safe generally my antminers use the 14AWG wire in order ot reduce clutter and simplify things. The wires get slightly warm when using only 2 pairs of wire for an S1, but probably only 30C or so which is quite safe I use a pair of 16awg on the ants (2 + and 2 -), no problems at all with temperature. Ants overclocked to 375.
|
|
|
Filling out nicely ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2Fr6P8s2d.jpg&t=663&c=NpXDEFsyoywfcw)
|
|
|
Still have room for some S2's! ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2Fr6P8s2d.jpg&t=663&c=NpXDEFsyoywfcw)
|
|
|
We will know when they ship, as their data center will have a huge spike in hashing power. Once they've maxed out their data center, I'm sure they'll start to ship customers units.
|
|
|
Here's a nice picture of Sunbreak's PSU adapters. You can run 6 antminers off each of these PSUs, the setup with the 6awg wire lets you run 3 groups of 2 antminers. 1 6awg per pair of ants is just right. Delta fans keep it nice and cool. It doesnt get any better than this. ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FBH46NMq.jpg&t=663&c=W6_-u9ibof_UOg)
|
|
|
This powers four of them: ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FcaJgJgB.jpg&t=663&c=mLkjFYPmtT4MfQ) Details on the PSU? it's 4, 2kw PSUs from the IBC bladecenter. Absolutely rock bloody solid power supplies. I'm had them running 6 antminers each actually, I keep it at 4 each for better power efficiency. I made more than I need, so I'd imagine I'll be selling the extras, with the power distribution blocks as well. (Can run 6ants from 1 block) Complete plug n play kit if anyone is interested. They only run at 208v-240v single phase.
|
|
|
I call it the AntCube (Cable ties after the pic was taken): ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FwS6AqR1.jpg&t=663&c=Ft6IYj9o9FS7cQ) This powers four of them: ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FcaJgJgB.jpg&t=663&c=mLkjFYPmtT4MfQ) This is what it looks like (without the ethernet tied up nicely yet): ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FRkh9fVL.jpg&t=663&c=LFecniBV_CB-7g)
|
|
|
I call it the AntCube (Cable ties after the pic was taken): ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FwS6AqR1.jpg&t=663&c=Ft6IYj9o9FS7cQ) This powers four of them: ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FcaJgJgB.jpg&t=663&c=mLkjFYPmtT4MfQ) This is what it looks like (without the ethernet tied up nicely yet): ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FRkh9fVL.jpg&t=663&c=LFecniBV_CB-7g)
|
|
|
Welp, I gotsta pay for some bills, I can either sell BTC or get a refund for a Neptune. I chose the latter. Just put in my refund request. We'll see how it goes tomorrow
|
|
|
Will buy MtGox Balances of 100BTC for 1 satoshi.
|
|
|
And why would they then point gox.com to mtgox.com if they were rebranding. Some epic trolling if you ask me.
|
|
|
So where did the BTC refund come from did they buy it on bitstamp, or was it sent from 1A73ExsM2doRwTLp82rv5U36QHbBFmHD1X or from 1Nbq2XZaRsKknf5fcT2wTXvBS31PaUWSeX
|
|
|
All of this after we gave them piles of money to fund their business a year ago, and again on their 2nd product offering, this is how they repay us.
it seems the shovel makers realized that shoveling makes money too that should be new motto on their neptune boards.. "Shoveling makes our dreams get found" "All your Jupiters are belong to us"
|
|
|
This all is a big smoke screen it seems. I'm trying to make sense out of it and am having av difficult time.
I have a batch 1 order. Wtf does this mean? I don't see how this info helps me make an intelligent decision... Do I hodl or bail and reinvest elsewhere?
I'll rewrite what I wrote earlier, to make it a little clearer. KNC will already be hashing with all the machines they will need to give their 3600 neptune customers 3TH on 1st July. In fact, as I said earlier, it's probably a drop in the ocean in terms of what they plan to have at that time. Come on! They aren't going to leave millions of dollars worth of machine powered off until 30th June 2014 and then turn them on at midnight! Which means the customer appreciation batch will be just about as worthless as the batch 2(3), as all these machines will already be hashing, the bitcoin difficulty will have already adjusted/risen to account for these 3600 machines. So instead of having a few weeks of mining and then seeing earnings decline (as with jupiter), you'll start out with the declined earnings and knc will have cleaned up any profits there was to be made beforehand! Essentially, your neptunes are going to be pre-mined. You'll get it delivered at the highest difficulty ever seen. You won't be adding a 3TH machine to the global bitcoin hashrate, you'll just be re-attaching it when it's probably already been there for a couple of months. In other words, knc will reap the benefits of your investment and all the bitcoins they can mine, then pass the leftovers at massive difficulty to people on 1st July. You are paying for that. The physical delivery of the product will just be to end any obligation to you and will most certainly be utterly useless hardware by the time it physically arrives. The cost to power it will probably be more than it can mine back. And to think at one point I suspected you were a crazy conspiracy theorists. Tuens out you were right on. Explains why the network difficulty has had such major increases while KNC and Cointerra weren't shipping. I knew antminers were selling fast, but yeah, there's out network protection plan, or for KNC a Netowk Pock Plan, just pocket their share of the network. Nice way to announce that they are completely screwing their early adopters by hashing well above 5% against them, just sell it to your "plan B" company and now they're a customer. Isn't "plan B" usually scrap the place and get out?
|
|
|
|