The heatsinks from KnC miners are pretty high quality. You should be able to use them as CPU heatsinks if you can fabricate the mounting hardware.

Also don't forget PCB coasters: https://www.google.com/search?q=pcb+coaster&biw=1366&bih=620&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjipsH788DJAhVKOj4KHXhaA3YQ_AUIBigB

Update: So the file copy method works better. Both servers are 100 Mbps down/up, but with a single SSH session I was only able to get 7 Mbps throughput.  I ended up having to write a script to split the files up into 10 groups and run parallel copy operations over 10 different SSH sessions to get close to the max. 100 Mbps throughput and was able to initialize a new node in under 3 hours compared with 2+ days the normal way.

As a test, I also set up a temp web server on the source node and tested copying the files over using the HTTP protocol. It turned out to be a lot faster than SSH for the single threaded copy, but I still needed to run 2 parallel copy operations to max. out my 100 Mbps bandwidth. This is understandable as SSH/SCP is a chattier protocol than HTTP and generates more "dead air" over a WAN.

So yes, copying the files over is a lot quicker than just starting up a blank node from scratch. but you still need to some work on your part to get the most of your available bandwidth.

Trying this out now. I had to stop the source node for 20 minutes to make a local copy. Now I am copying 60GB worth of data to the new server. Unfortunately this is a single threaded SSH/SCP operation and the servers are 6000 miles apart (high latency), so it remains to be seen if I save any time at all.

Can someone tell me what is the best practice for initializing a new bitcoind node quickly? Do I really need to wait 2 days for the blockchain to be downloaded and validated before the node is operational? As far as I can tell most of the time is spent not on network resources but on CPU resources, at least for the hardware I am running (Azure & Amazon EC).

If I already have another fully initalized node that I trust, is there anyway I can just copy the blockchain database from a trusted node over to the new node (bypassing all the time consuming validation)?

No I don't. However, off the top of my head I know Pi to 100 decimal places so I think I am capable of memorizing a couple of my addresses if I really need to.

Just run them at 400MHz or less and they'll be fine and reliable. If you run them at 350 MHz you can even replace the 140mm fans with 1000 RPM silent models and use it to warm your room this winter.

Here is an image of my silent 500W space heater. I removed the fan grills and used 1000 RPM fans and clock them at 375 MHz, resulting in slightly over 1 TH/s. I used extra controllers from old KnC Jupiters so that I can spread them around the house in clusters of 2 cubes.

If we went back 12 years ago when the best Intel CPU was the single-core 3.0 GHz Pentium Northwood, would you have imagined that the best Intel CPUs of 2015 were still only running in the 3.x GHz range?

To me BIP101 is analogous to the Intel product planners back in 2003 declaring that in 2015 Intel flagship CPUs would be single-core Pentium Northwoods and running at 300 GHz. Don't worry about heat --- the process node improvement from 130nm to 22nm should take care of everything!

Instead over the years Intel has added multiple-cores, multiple execution pipelines per core, larger caches, more caches, improved branch prediction logic, and so on and so forth, but  kept the max core clock frequency the same at around 3 GHz.

Saying that BIP101 is the best path forward is similar to those self-proclaimed experts posting in overclocking forums back in 2003 that Intel's product strategy to go multi-core was retarded and they should just be focused on increasing GHz by a fixed percentage for each product cycle.

Indeed. The news of the executions came out overnight over CNN and other networks but no mention was made about Bitcoin.

Eligius. 

Well, OK, that wasn't one of the questions but I think it is not complete without it.

I did not vote because I feel that any scaling strategy should involve more that one dimension, i.e. the blocksize cannot be determined on its own in isolation of everything else.

It was super annoying to see big block pushers who kept screaming "Blocksize must go up because Bitcoin needs to scale!" whereas six months ago they probably had no experience and no clue as to what scaling a distributed system involves. They seem to have shut up a bit now.

I would sell all of my bitcoins if Bitcoin were to become mainstream. By mainstream I mean controlled by big companies and regulated by the government. I have no need for such a coin, even if the price were to climb to $1 million per coin.

Not what I had in mine. The FiatCoin would be unmineable and instead backed by an equivalent amount of fiat. The issuance would be done by a centralized authority (e.g. bank or brokerage, possibly for a small fee) but once issued they may be exchanged freely on a distributed blockchain that is compatible with cross-chain transactions. Mining will be fee based only without block rewards.

It would be great is someone could implement FiatCoin (pegged to USD, EUR, GBP, etc.), hopefully without any AML/KYC restrictions (beyond the initial purchase). Then you can truly trade Bitcoins and fiat from your client.

They may be licensed and regulated but they're still structured more like a bucket shop than an exchange (that is order matching, trade execution, clearing and settlement all happen under the same roof).

OP should start CommieCoin. No mining required. Everyday you wake up the number of coins in your wallet resets to a certain amount regardless of how much or how little you've spent/received the previous day. This way wealth is equally redistributed on a daily basis and everyone remains equally rich.

[Hint: You will destroy the value of Bitcoin if you try to redistribute the wealth of its users, even if you think you are picking on just the miners.]

Some of the reversed trades were in the $400-$500 range. Hardly an "unimaginable" price. I bet you won't feel the same if you were one of the counterparties.

If you have a need to run a current Windows O/S consider running Windows Server. I have been running Windows Server 2012R2 instead of 8.1 and I will soon upgrade to Windows Server 2016 (Windows 10 equivalent). They are meant to be run in air-gapped environments and spyware is disabled by default or easy to turn off. The default server install is also way more hardened than the default desktop Windows install.

At what point do you draw the line to stop Bitcoin from becoming fiat by economic majority?

It behaves more like a smart card than a dumb memory device. The difference is that the private keys stored inside are protected with a PIN and cannot be retrieved through the USB interface. All sensitive crytographic computing is done on the TREZOR and not the browser. However, it just uses an ARM Cortex processor so unlike a typical smart card the keys are probably not protected from advanced electronic tampering or snooping were the device to fall into unfriendly hands.

I think they are trying to save on licensing fees. A lot of good movies have been replaced with crap. Since Bitcoin movies are typically low budget productions I would imaging Netflix would put them up if they were aware of them.

If they can have all those 420 and conspiracy theory documentaries I am sure they can have Bitcoin documentaries.