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.

Except by doing that, he risks confirming that he is Satoshi.

It is telling that satoshi broke his silence to say he was not Dorian.

P2Pfoundation link

The real Satoshi risked exposure to clear Dorian's name.

It only took me 10 months, but I built a 10Ghash/s stealth miner. It draws less than 300W.

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.

The "satoshi premine" theory bugs me. When Bitcoin first started, the difficulty stayed at 1 for 11 months because nobody was interested in mining it.

I think I recall reading about it on Slashdot in 2009 2010. I thought it was a neat idea, but that it would never take off. One year later, Bitcoin is worth actual money. Suddenly, I can't CPU mine anymore.

I was going to guess no internet until you posted the error log.

"0197 > Failure: twisted.internet.error.ConnectBindError: Couldn't bind: 24: Too many open files."
 suggest you should look into raising the number of file descriptors. Have not run into that problem myself yet.

The Arctic Ocean is now ice-free in the summer. My government has literally burnt books in an apparent attempt to cover up historical records that contradict their message that climate change is not a problem. That suggests that even utltra-right-wing Conservatives believe in the science.

Science is all about observing cause an effect. If you destroy the observations, I am going to have to rely on emotions.


The ones I have seen in Canada are limited to 32 km/h, 500W and use lead-acid batteries. Unless you are talking about actual electric motorcycles and not vehicles classified as "electric bicycles". I can beat those things up a (relatively short) hill by brute-force.


Yes. My point is that electric vehicles are only non-competitive if you assume that you need to quickly replace the energy expended. Most cars spend about 95% of their time parked, and are used for commuting less than 100km.

Pure Electric vehicles are not appropriate for continuous commercial use (like a taxi service) or long road-trips. That said, I have seen electric vehicles used all day in an industrial setting (intermittent use, read: forklifts and electric pallet jacks).

If I was going to spend a lot of time on the highway, I would consider a diesel vehicle.

If it takes your node more than 30 seconds to verify a block, you will get 100% stales. Even without 6 different hashing algorithms, I have seen up to 12 seconds of CPU lag on my machine processing P2Pool.

Part of the problem may be that python is an interpreted, not compiled language. A Just-In-Time compiler like PyPy may help. (I think I will test that on my own machine)

Edit: I think actual block processing is handed off to Bitcoind. P2Pool still has its own chain though.
Edit: pypy Does not work:

Update:

I live in Edmonton. We get most of our moisture from the Pacific Ocean, which does not freeze over.

ICE motors only get about 30% efficiency at best. They also have more moving parts and require more maintenance. Lithium-ion batteries may be lighter, but in practice you are only allowed to use about 60% of the capacity (30-90% charge). This negates the weight advantages over NiMH.

Patents don't require science to prove they exist: Patent encumbrance of large automotive NiMH batteries.

I ran into those limitations trying to build a lighting system for my bicycle. You don't see rechargeable D cells in stores: even though "real" NiMH cells have more capacity than the alkaline cells you can buy. The NiMH C and D cells you may see in stores are really AA cells: which is why advertise the same capacity as AA cell.

I got to play with some real 10Amp-hour cells imported from China by a friend. They cost about $3 each (packages of 16 totaling about $50). The "fake" D cells cost $5 each, and only have 2 amp-hour capacity. I asked about buying high-capacity cells locally: and was quoted $50 per cell. Patent fees discouraging automotive use is the only rational explanation.

Battery patents also appear to be why we do not have Electric SUVs pulling generator trailers on road-trips
(Bold is mine)


No, that since the 1996 WIPO "Copyright" and "Performances and Phonograms" Treaties, devices and software have been designed to betray the user in order to enforce Effective Technological Measures.

As an example, HDMI was invented to encrypt the video signal going from video players to displays. The standard sucks: it's only purpose is to require device manufacturers to agree to a patent license. The patent license in turn, requires gives a $1-2/port discount for the use of HDCP, which does the actual encryption. The real "single cable digital audio-visual" standard is DVI, which does not use HDCP. The cabling is actually cheaper and higher quality (and can send signals much longer distances).

You go into a lot of detail, but most of it strikes me as slightly wrong.

Not everything is part of some global conspiracy to control the masses. As you point out: delaying problems inevitably leads to collapse. Any secret cabal in power must realize this.

Global warning is real, and appears to be happening in my life-time. We are seeing more and more snow-free days. When I was a kid, snow-free Decembers were rare in Edmonton. In the last 15 years, we have come close to something like 5 "brown Christmases".

Ironically, I believe the conspiracy that electric cars are being held back by battery patents on NiMH batteries. As a result, they remain expensive and in limited supply. This conflicts with the theory that smart meters are control measures. (Electric cars can make very interesting use of smart-meters).

You also seem to be under the delusion that "Intellectual Property" rights should trump physical property rights. This leads to devices designed to betray the users. The rabbit hole goes so deep that the thought of setting up a VPN service users may rely on for their safety sickens me. Because modern computers are inherently insecure, sometimes actively made that way: there is no easy way to prove your infrastructure is not back-doored.

The think is, we need to risk setting up privacy-enhancing infrastructure. We need to start auditing our routers, NICs, keyboard and storage firmware. We need to start proving software (especially embedded software) is written correctly.

Well I think I found out where the extra 6 seconds of lag came from:
I may be misinterpreting things, but I appears to take my machine 1-7 seconds to put work together.

OK, I finally got P2Pool running and learned the truth. One thing that confused me is that the +1 option does not appear to be documented anywhere. What it does is tell the blade to report diff-1 shares. Apparently, the blades do not support higher difficulty.

Test results are in: no longpoll support. 21% stale for me, I guess. (confirmed with netstat -- only transient connections show up on port 9332) The reported DOA rate varies between 20-40% for some reason.

PS: Slush's proxy does not work because P2Pool does not appear to support the stratum protocol. I may try a different proxy if I think it will result in less CPU usage (for example, using compiled, rather than interpreted code). Better power savings would result from replacing the CPU though.

Edit: P2Pool does support Stratum. Slush's proxy not working is more complex. P2Pool tries to automatically detect whether the worker is using stratum or not. While at the same time, the slush proxy tries to automatically detect if the pool is using Stratum or not (using the getwork protocol). I tried removing that detection from the slush code, and it still didn't work.

Is this miner running a stratum proxy locally? each of the 32 chips seem to mostly ask for work independently.
My blade gets work 150 times per minute, meaning that the latency should be at most 12.8 seconds (150/32*60).

I don't actually have P2Pool working yet, but moved the midstate calculations to my Stratum proxy on the assumption it can calculate mid-state faster than the blade.

Edit: Since AFAIK, the getwork protocol does not allow workers to be interrupted with new work, we should be able to estimate the expected stales given a specific block frequency. If we assume a 13s worst case latency, that works out to at most 43% stales with a 30 second target. If we assume a 6.5s average, that works out to 21.7% stale. -- that does seem high.

Edit: Apparently Longpolling works around HTTP limitations by having the miner request new work immediately. The server then does not respond until new work is ready. Testing time.

FreeBSD uses a log rotator called newsyslog.

The following line in /etc/newsyslog.conf appears to do the right thing:

The fields are separated by white-space. They are: 'logfile_name', 'owner:group', 'mode', 'count' (number of archives to keep), 'size' (in kb), 'when' (chose anytime), 'flags' (J->Bzip2, R-> Treat next field as as shell command to run), 'path_to_pid_cmd_file' (filename containing daemon's PID, must start with /), 'signal_number' (optional, omitted in example).

I first tried appending "killall -HUP bitcoind" directly, but the arguments were interpreted as an invalid signal number. With the size limited to 100 kB, the logs get rotated every hour (because they they grow faster than that).

sighup.sh is just a simple shell-script:

I was not able to find any 'copytruncate', setting, but it appears to do the right thing.

The initial responses seem to be alluding to wallet-stealing malware.

The thing to remember is that proof of ownership with Bitcoin is knowledge of a private key. For large amounts, the private key should not even be stored on an Internet-connected computer.

Because transactions are irreversible, computer security suddenly has money on the line.

I recommend paper wallets. Of course, you have to keep in mind that if anybody is able to read the keys without your knowledge, they would be able to spend the coins at any time.

I also recommend keeping more than one copy in at least 2 geographically separate locations. While this will increase the risk of theft, it will also reduce the chances of loosing your Bitcoin in some disaster that destroys one location.

Finding a "fully open source" computer is hard enough to find, never-mind a smartphone.

Here is one example that is never coming back now that Nokia went Windows phone exclusive: Nokia N900. I heard there was some mis-management as well: firmware updates were adding features, not fixing basic bugs.

Wikipedia list of open-source mobile phones

Android is more open than Apple phones. If you buy it from Google, it is not even carrier-locked. That list says it excludes Android due to proprietary drivers. However, it is common for GNU/linux PC users to install proprietary drivers as well.

Recently, the nouveau project benefited from Tegra driver source code written for Android. Things like register names and functions were disclosed. PS: AFAIK, the PowerVR graphics used by the N900 probably also required proprietary drivers.

The difficulty is that nobody cares. If they tried, they would quickly burn themselves out.

What is the Monkeysphere?

Even if we did have the capacity to care, it would be difficult (but probably not impossible) to vet all of the sob stories. Bitcoin is a game-changer. Suddenly, you can irreversibly transfer value in minutes. People who don't react with a "this is probably a scam" attitude are setting themselves up to lose a lot of money.

I gather that many Bitcoin enthusiasts are not very rich at all. They spend them almost as fast as they acquire them. I myself made low thousands because I was not able to sell as the price climbed 1000% in November (paper wallets FTW).

I can tell you that a few thousand dollars won't last long if you have to pay rent. One rule of thumb I heard is that rent should not exceed about 30% if income. The corollary is that people spend a lot of their income on rent (or transportation).

In the past month volume has bee relatively low.

I suspect early adopters are artificially keeping the price below $1000USD/coin.

This will cost them Bitcoin, but may not be a bad thing in the long-run. One of the criticisms of Bitcoin is that the early adopters control too much of the wealth.

By keeping the price stable; they can encourage Bitcoin adoption and spread coins around (reducing coin concentration) at the same time. In FIAT terms, how much they can buy with their Bitcoins may end up being about the same anyway.

Offline storage is a very good idea for Savings wallets. The time delay saved me from getting scammed out of 1.75BTC (the scamer was promising a high price If I could sell more than I wanted to).

With credit cards, you have relatively low credit and daily transaction limits. with Bitcoin you can send millions of dollars of value around the world in minutes, while paying pennies in fees.

Credit cards have to be reversible because they are vulnerable to the replay attack. Bitcoin has to be irreversible because it is decentralized.