pfft, nobody reads a 323 page thread before posting? (I used the search feature to save time.)

I have my console for accessing my miner within 1m. Sometimes it is convenient to cluster your equipment.

That said, I should be getting that miner working, rather than posting here

PS: I eventually want to use an RS-232 cable to do initial setup on a second router: with tor and IPv6 support.

My actual rig is is written up here. Though the blade has been replaced by 2 rockminers. Those rockminers have been sitting idle for 4 months

TL;DR: The RasPi has only 256MB or RAM. That is not really enough to run a full Bitcoin/P2Pool/namecoin node. Together those use at least 1.2GB of RAM on my machine. I am also not sure of the CPU capabilities of the RasPi. Lately P2Pool has been using 40% of one core. I am interested in replacing that machine with something a little less power-hungry though.

I am also going to need moar disk-space soon too. I have only 10GiB left on a 50GiB partition (left 4GiB for swap, 4GiB slack (because: SSD is "not for server use"))

The documentation for one of my sister's (Toshiba) laptops actually said: "Do not run for more than 30 hours at a time."

Actually (re-reading the specs), it is 16 amps per phase: which would only draw 32 amps in the degenerate case where two phases are loaded with one unloaded. But that is still less than 20 Amps per phase..

If they are brushless motors, you can not just reverse the polarity without letting the magic smoke out. Brushless motors are essentially AC motors with an integrated controller that handles the conversion (and starting) for you.

Took me a while to figure out exactly what you were complaining about.

The yellow wires are ground, the breakers indicate the brown is hot, and the blue would be neutral.

The terminals will be live. The terminal block is probably designed to go into some kind of enclosure. The point of entry into such an enclosure would be the logical place for strain-relief.

I guess they can still pull a single machine out if they unplug at the box.

Since that appears to be 3 phase power, each phase would have a draw of 3*1600=4800W/phase.

At 230 volts, that is 20.1 Amps per phase.

So something like this 32 Amp PDU should work. (Don't know about price; probably not cheap.)

Apparently I only knew about 2 of the 3 data centers in my area. Bandwidth prices are around $1/GB no matter who you go with. I blame the "supernet" Public-private-partnership. Wholesale bandwidth costs $50/Mbit.month.endpoint. At those prices, power is included (within limits).

(Population is also sparse)

PS: if anybody is interested in data-by-mail, PM me. (Not useful for Crypotcurrencies, except for the initial block-chain download.)

Mining at a small loss may be useful sometimes. For example it you want to mitigate hash-power concentration. Of course mining at a loss makes more sense with negligible power draw, so I doubt the very large hashers would be able to employ that strategy for long.

With my 1 10 Ghash/s blade, Internet connectivity is the dominant cost. Currently, I have 100% of the mining revenue ear-marked for paying power costs. A faster hasher may pay for it's Internet for a few months: doubt it would pay for the hardware cost though.

That appears to be a P4 or PIII machine, which I believe pre-dates the alienware brand Nope, Alienware has been around since 1996.

Edit: PIII by no P4 power connector.
P4 because large wire bundle appears to have at least 2 12 volt lines (yellow), but the 20 pin ATX connector has only 1. I checked: The proprietary Dell connector also had only 1 12V wire

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.

My Search-engine-fu failed me. "CCD" commonly stands for "Charge Coupled Device". Adding "bitcoin -camera" to the search terms brought up Sony console game .torrents (apparently .ccd is a common extension).

A search of this forum for "CCD-600" only brought up the above post.

I would like to second the complaint about people quoting images. My "good" computer (below) runs bitcoind, so the machine I am using to browsing the web is 10 years old.

This is my response to to Project: Stealth Mining Rig thread. Only took about 10 months

Specs

• Antec Sonata Case (not sure of specific model)
• Antec EarthWatts 380W "green" PSU (see? it is painted green...)
• 3Ghz dual core Pentium D processor (power hungry)
• 4GB RAM (3306MB available)
• 60GB SSD (expect to need to replace it by the end of the year)
• 80GB SATA disk (used for unrelated encrypted jail)
• 100Mbps RTL8139 NIC
• 9600 baud serial console
• FreeBSD 9.2 (apparently I have to figure out how to migrate to 10.0 now)
• ASICminer new-style blade (10.7 GHash/s)

The grey cable is a serial cable (serial console over a null-modem cable). Removing the 2MB video card freed 124MB of addressable memory. I had to install a RTL8139 NIC because the on-board LAN would lock up when I tried to transfer bootstrap.dat over using SSH. There is no performance loss, since it is only plugged into a 100Mbps router/switch anyway.

Besides being discrete, the metal case blocks Electromagnetic Interference (EMI). I actually had to move the machine almost directly below the TV antenna (out of picture) because the front of the case does not block EMI as well as it should. The machine is also relatively silent (case is designed that way). With the fan on high, I can easily hear it, but the controlling machine (Pentium 166) is still louder.



As you can see, I crammed an ASICminer new-style blade into the case. The only fans are the PSU, the CPU, and the one to the upper left (120mm IIRC). I tested that the air-intake on the CPU stayed below 38°C with the fan on low; with no miner in the case. Did not re-test with the fan on high and the blade in the case.

Cable management is key. It helps that the drives are partitioned from the rest of the case. I wanted to use a PCI-E 6 Pin -> dual molex adapter , but was not able to find one locally. Ended up using a 2 molex -> PCI-E 6 Pin adapter I got from a local computer shop free-of-charge. Apparently when they are building systems they always get an extra one. They come with video cards, but sane builds will use a PSU that includes the 6 pin connector. I made sure that each connector uses its' own wire pair going back to the PSU. This meant I had to run a different cable to the drives: one of the SATA power chains had a molex connector at the end, while another did not. I do share blade power with the vent fan (about 3W, 0.25A) and front LED though.

I was disappointed by the lack of mounting holes. The blade sits on a boxy plastic tube with a slot for the fingers cut into it (possibly not needed, but did not want to risk it). I have a zip-tie wrapped around the fan connector (left), fastened to the case with a series of half-hitches. I have insulated wire running through the two small holes on the right side of the board. That is looped around internal case structure.



I did not want any add-on cards at all. However, I think the NIC is low-profile enough to not cause problems. Because the spinning disk is SATA, and only has data used by its' jail on it: it is actually hot-pluggable. I learned the hard way that the ASIC miner blade is not hot-pluggable. It worked no problem once (the one wire gets hot until you plug in the second one). However, the second time I tried it, I got a big blue spark, and the PSU shut-down. Toggling the hard power switch got things working again.



I was surprised how much extra power P2Pool uses. With Bitcoind and Slush's stratum proxy running (pointed at Eligius), the CPU idled between 700-350Mhz (0.07 load average). Running P2Pool, the CPU stays between 2-3 Ghz (0.30 load average). This results in a power draw of an extra 20 Watts or so. I saw one report that pypy (a just-in time python compiler) works with P2Pool, but have not been able to get it working yet.

Update: Got pypy working. Results are inconclusive so far, but looks like about a 0.20 load average (33% CPU usage reduction) the same CPU usage at the cost of more than double triple the memory usage (from about just under 0.5GB 400MB to over 1.0 GB 1200MB). CPU still stays at high clock speeds.
Edit: It appears the initial drop in CPU usage was simply reflecting the reduced network traffic the node was seeing after restart.

Power Usage

• Pentium D Machine = 94 Watts (idle, with fan on full)
• ASICminer blade = 80W (28 W idle)
• Pentium 166 = 36W
• VDSL modem = 4W
• Netgear Router = 2W
• Subtotal = 216W

PS: I am aware that newer boards are more power efficient. The Pentium D was probably the most power-hungry design ever made by Intel. They went back to Pentium II based designs for their next chip (the Core 2). At 92Watts of idle power draw (fan low), I am convinced something is wrong with the machine. I don't think the board supports a calibrated power supply output resistance of 1.6 milliohms required by the processor. Stopped short of thermal imaging because I could not find an appropriate camera. I disabled everything not in use on the board, with no effect. Most my old computers draw about 40W with non-essential components stripped.

I went and got new PC how would i put linux on it?
[/quote]

I would recommend just putting Linux on the old machine instead of throwing it away. If you are paranoid, you may want to make it an offline machine with no network access.

I think my advice would be to sell between 1/3 to 1/4 to replace the FIAT taken from savings.

While his would be at a loss, the woman complaining in the reddit post clearly sees it as a very speculative investment (at best). You would then negotiate an allowance to do "dollar cost averaging".

Oh and try to diversify your interests a bit. People tire of hearing about Bitcoin. (Ask me how I know.)

Turing-completeness is a non-starter.

The Bitcoin scripts were deliberately designed to not be Turing complete.

I have to admit, the 28 minute (thorough) block propagation time (for 20MB block) mentioned earlier does seem realistic to me. That works out to about 14 hops with each node spending 2 minutes on block verification. Edit: Forgot to add: Turing completeness, even with cycle limits, will obviously increase block verification time.

This lends credence to LaudaM's contention that the blocks will not fill up immediately. Though, I have seen it argued that the large miners will have an incentive to push smaller miners out with large blocks (filled with garbage if need be).


Because a change like this has to be planned months in advance. During the next bubble it will be too late to meaningfully accommodate the increased transaction volume. I suppose that may be the point: to temper speculation with high fees. The difficulty is that the institutions pushing the next bubble have the ability to print their own money.

For merged mining, the relevant coin has to accept that having a parent header is "just as good" as being the parent. I doubt either MPcoin or Gavincoin will be modified to be merge-mineable.

Edit: it would be interesting if they were both modified to be merge-minable.

If you buy up Bitcoin before the fork, then don't move it (keeping it in cold-storage (ie: off-exchange)), you get to "bet" on both forks at once.

The risk comes when you take up mircea_popescu on his "buy gavincoin at 25% off" offer. Edit: no takers so far.

Presumably, if you need to spend Bitcoin while waiting on the sidelines, You can buy some at an exchange, then spend immediately: assuming both your exchanger and retailer are using the same fork.

Edit: tvbcof has been bringing up an interesting question: how will SPV wallets react to all this. I think it is worth investigation. I suspect SPV clients will try to look at the block the transaction is included in, which means size is important. If a forking block is buried and not examined, the SPV client will probably side with the fork with the most hash-power.

Incidentally, that is how consensus is achieved in Bitcoin: the fork with the most hash-power wins.

For my calculation above, I was assuming the blocks would be about 20x larger. Since the current blocks are about 35% full, we are talking about 5-7MB blocks in the above scenario. I don't claim to know exactly how long it would take for blocks to get to that size. However, a rule of thumb is that information expands to fill the available space. That is to say, making the block-size larger does encourage more transaction. If I had to guess, I would say we can expect the first full 20MB block within 5 years.

I also would not be surprised if Bitcoin transaction growth grows faster than common Internet bandwidth: at least in the near term.

• About 15 years ago, I was on 56Kbps down, 30kbps up dial-up. (~10GB bandwidth cap, with less than 1-3GB of usage expected) (not really comparable).
• About 10 years ago I was on 3Mbps down, 640kbps up ADSL. 70GB cap.
• About 5 years ago, I was on 6Mbps down, 1Mbps up ADSL/VDSL. 100GB cap
• Now I am on 25Mbps down, 2.5Mbps up Cable. 250GB cap.
• If the rough trend continues, I will be on 50Mbps down and 5Mbps up in 5 years. 500-750GB cap.

Disclaimer: those numbers are not terribly consistent. The underlying technology and even price has changed over the years.