They say time is money, but if that’s true, money must also be time. It’s all figurative, of course, but in the case of this NTP server heater powered by Bitcoin mining dongles, money actually does become time.
This is an example of the lengths to which Network Time Protocol aficionados will go in search of slightly better performance from their NTP servers. [Folkert van Heusden], having heard that thermal stability keeps NTP servers happy, used a picnic cooler as an environmental chamber for his Pi- and GPS-based NTP rig. Heat is added to the chamber thanks to seven Block Erupter ASIC miner dongles, which are turned on by a Python script when a microcontroller sends an MQTT message that the temperature has dropped below the setpoint.
Each dongle produces about 2.5 Watts of heat when it’s working, making them pretty effective heaters. Alas, heat is all they produce at the moment — [Folkert] just has them working on the same hash over and over. He does say that he has plans to let the miners do useful work at some point, not so much for profit but to at least help out the network a bit.
This seems like a bit of a long way around to solve this problem, but since the mining dongles are basically obsolete now — we talked about them way back in 2013 — it has a nice hacky feeling to it that we appreciate.
Idea for doing useful work – replace the ASIC miners with one or more RasPis running BOINC. Old Android phones or Android TV boxes could also do this, but the automation will be more difficult.
On the topic of NTP accuracy, ordinary ntpd software with automatic sync and clock drift adjustment has been good enough for me so I haven’t gone too deep into it, but wouldn’t something like this be more helpful than temperature-stabilizing some form of ordinary computer clock? https://www.electronics-lab.com/meet-the-modern-and-sleek-usb-real-time-clock-usb-rtc-designed-for-accurate-time-keeping/
I found that they do not produce enough heat for this purpose (I tried).
Damn, I got way too excited that someone finally found a use for the handful of Block Erupters I still have kicking around :(
Heaters doesn’t do it for you?
Oh the dirty kludge…
If stability is such an issue, why not simply invest in a rubidium frequency reference ?
Rubidium is radioactive. It’s illegal to buy stuff like that unless you got a permission and proper containment structures in place to keep it from endangering people around you.
Rubidium clocks do not need any license.
Nonsense–the amount of radioactive material is small and a license is not required. Anyone can buy ionizing smoke detectors using radioactive americium from your local hardware store or online unrestricted.
What do people use these amazing high spec NTP servers for?
high speed distributed computing has a huge demand for this, high precision network time allows for packets to be synced to the cluster of machines so that if one machine gets out of sync with the rest, you dont get a “packet from the future” scenario, this in turn reduces computation cycles on the hardware and efficiency and speed of the system is raised
Theoretically yes, 2442 Bq/gram (66 nanoCuries) for Rb-87 (Natural Rb mix has about 20% that), and a rubidium clock has a small quantity of it (milligrams) so we end up at less only a few Bq per clock, which is comparable to a banana(~15 Bq). Hence, as far as I am aware these are not regulated, and are certainly ok to have in The Netherlands (i have one).
https://www.ebay.com/itm/175063606877?hash=item28c29aa65d:g:nAAAAOSwsoZhkwY7
What’s the fun in that?
+1
this warms the cockles of my heart. and the server, apparently
Wouldn’t it be nice if crypto mining ASICs would have some features in them to make them useful for “anything” when they get obsolete as far as crypto goes.
Though, I can see how that would interfere with mining performance, after all, an ASIC for mining only really has 1 job, and that is to be a waste of resources by the end of the year.
I don’t know of any other chip that is as “useless” as an antiquated mining ASIC. Though, the exact same thing can be said about video encoders, some specialty audio effects chips, etc, but these are at least still useful for their original application, while a crypto mining ASIC isn’t useful for its intended application when it has gotten “old”.
What can a crypto miner ASIC contain?
A few math functions dedicated to very large numbers, probably a bit of memory, maybe even a core for administrating the work. Though, I wouldn’t be the slightest surprised if the actual “mining” part is a large state machine built to do its one job of mining with no reconfigurability to speak of.
Just imagine what one could do with a reconfigurable state machine handling variables in the hundreds of bits. Even if it were rather limited in how one can configure that state machine, one could likely get it set on new interesting applications. Maybe have it as a general encryption accelerator or the like, or file compression/decompression or other similar tasks. (might still have a limited scope of applications, but if its too slow for crypto it might still be fast enough for compressing files to saturate its interface)
All they do is a very special double SHA-256 hash that’s only useful for bitcoin mining. You can’t even use them for generic SHA-256 hashing, because there’s no interface for generic data.
I know that these ASICs are indeed very application specific.
But it would be interesting to see what “end of life” applications they can have if they were a little bit less application specific.
I for one wouldn’t mind a file compression accelerator.
The loss in mining performance would be too big. The whole ASIC is hyper aggressively optimized for the single purpose of producing as many very specific hashes per Joule as they can.
Isn’t this exactly what op said in their first comment?
Rehashing is a humourous thing to do when discussing crypto crud I guess lol
So like… Can these do anything else? Like dang that is the most complicated way to make a heater… ever.
If the miners are not doing useful work, then they are also not helping the network.
The useful work is their heat production. [Folkert] could have used some resistors to the same effect and also benefited the network.
A resistor wouldn’t be able to sign transactions.
Taking advantage of the fact that all electronic devices operate at 100% efficiency if your local environment is colder than desired.
Indeed. When is electricity for mening free? Whenever the miner replaces a resistive heater.
“But everyone uses heat pumps”, I hear you cry. Well. Yes, sort of. Bit there’s still plenty of places where you need some heat locally, so panel, space or fan heaters are still being sold. If it was easy to turn on and off the miners, putting these heaters to good use would be nice.
Mind you, I’m reluctant to call crypto mining “good use”