Bitcoin mining on an Apollo Guidance Computer
righto.com
righto.com
And now it's the Apollo Guidance Computer (0.097 H/s)! I think it's particularly hilarious, as the slogan says, Bitcoin to the Moon!
[0] https://www.youtube.com/watch?v=y3dqhixzGVo
[1] https://www.righto.com/2015/05/bitcoin-mining-on-55-year-old...
[2] https://www.righto.com/2017/07/bitcoin-mining-on-vintage-xer...
Edit: fixed all incorrect H/s numbers, thanks for spotting that.
I wonder if you can do these on that old HP calculator everybody loved (the 48, I think?) and how fast.
Repeated hashing is notoriously difficult to parallelize as it's naturally a serial operation, that's why we use repeated hashing as a Key-Derivative Function to protect passwords. For humans it's more challenging, since SHA-256 has 64 rounds internally, which means: to calculate a single hash, you need to iterate over the data 64 times, each time using the output of the previous round as your input.
What you can do, is:
(1) Optimizing the process of calculating a single round. I guess you can identify all the needed arithmetic steps (mostly logical opeartions and bitshifts) and divide them to multiple units of task-forces, assembly-line style, like a human-powered ALU. Ideally, workers that doesn't care about "which round it is", but only care about the computation itself should be as many as possible, i.e. using combinational logic instead of sequential logic.
I think many computations were done in this way before electronic computers, I believe even numerical simulations of fluid dynamics were done by rooms of female workers with mechanical computers running the same algorithm again and again over a matrix.
(2) Create as many copies of your human-ALU as possible. Although iterations within a single-hashing operation is not parallelized. But mining only requires you to do independent hashing, so it should scale linearly.
I don't have the needed skills to get an actual number, if you - the reader, thinks you can. Please do it and submit your blogpost to Hacker News.
And I wasn't interested so much in the absolute throughput, but how the figure would change in terms of hashes per second per human, so duplicating a full ALU module wouldn't be relevant; the concern is how much an additional human improves the rate after adjusting for the fact that you added one more human-resource.
Edit: I'm new to queuing theory, but I may just grind through the numbers there. Could do it as a "Show HN" if there is a tool (preferably as a web app) where you can model a process and tweak the numbers like in Factorio.
> Right, I was mainly thinking in terms of speedups from not having to switch tasks and the gains from specializing in one kind of computation.
Switching tasks is going to be a relatively small amount of time over a 24 hr period. Given that one person can only hash 0.67 hashes per day assuming full undivided attention I'm really not sure what you could count on saving.
The biggest advantage adding more people would do would make it actually feasible to calculate continuously as I doubt any human has the focus/patience for doing mundane number crunching for 24 hours without breaks.
I totally get that this is all hypothetical-for-fun math, obviously nobody is going to actually try to mine bitcoin by hand. But that 0.67 hashes per day is terribly low, a 50% speed improvement would only barely put you over 1h/d. Not to mention we normally compare in h/s not h/d so it's actually 7.75e-6 H/s.
Speaking of AGCs, he just recently uploaded a video where they used a restored AGC to dump prototype core rope memory modules from the Computer History Museum's (non-functional) AGC.
Although I do like the sound of your idea...
I suppose computers really are fancy clockwork. So are humans.
We don't really understand how the brain works, but it does seem that neuronal integration is an analog and probabilistic process, quite different from deterministic responses in computers (or clockwork). Some people have even speculated about quantum mechanical effects, though those hypotheses are very controversial and I don't put much stock in them personally.
It's overly reductive to compare the brain to a computer, and the analogy tends to obscure more than it illuminates.
https://mindmatters.ai/2018/08/the-brain-is-not-a-meat-compu...
https://aeon.co/essays/your-brain-does-not-process-informati...
We as humans might be missing the right equations/right measurements/precision to actually figure out it's exact state.
We are just accustomed to computers being electrical. In the end it's all binary and we just happen to all use digital input and output.
As far as using core rope with relays, you'd probably need to wrap each wire around the core many times (transformer-style) to get enough voltage to trip a relay. The raw signals from cores are really small and hard to detect even with ICs.
Relatedly, Jacquard Looms were some of the earliest punch card-based "computers" as well.
There's probably some set of interesting bootstrap paths to general computing entirely from the textile world.
> it would take the AGC 4×10^23 seconds on average to find a block. Since the universe is only 4.3×10^17 seconds old, it would take the AGC about a billion times the age of the universe to successfully mine a block.
Pretty mind boggling stuff.
According to Google, one block requires 2,440,643 * 2^32 hashes. Multiply that by 10 seconds per hash, and you get 3.32176901e9 years. The Universe has been around for longer than that, about 13e9 years.
Brilliant.
"The AGC that we're restoring belongs to a private owner who picked it up at a scrap yard in the 1970s after NASA scrapped it. For simplicity, I refer to the AGC we're restoring as "our AGC"."
How much other fantastic equipment of historical significance have we lost, I wonder?
Crazy thing was, I must have walked by where it was stored hundreds of times over the years since I first started to go to this place back in 1991 (Apache Reclamation and Electronics, here in Phoenix). Apparently it was kept in a trailer that was outside. I never saw the Altair in there, but it was stuffed full of junk and hard to "wander" around inside (the few times I went in, I was lucky not to be bit or stung by any number of things, to be honest).
So one day I was out in the yard wandering around; I turned around - and saw it sitting on a chair (I must have walked right past it). I walked over to it, thinking "No way - just, no way" - but yes, there it was. Very dirty, missing it's top cover, but otherwise all there with a ton of peripheral cards slotted in place.
I went over to one of the owners of the place; he was up on a forklift - I pointed over at it, and said "I'll give you $100 for that thing over there". Yeah, I knew what it was worth, but this same guy had sold me 100 meters of multimodal fiber cable for the cost of copper, back when that stuff ran you 25 dollars a foot - so I thought "maybe he has no clue what it is".
He looked, turned to me and said "Nope - can't do that. Some guy a week ago pulled it out of that trailer there and told me it was an antique. Can't go any lower than $250.00". I dropped my smile, hemmed and hawed, grumbled a bit - you know, put on a show. Then I looked up, and said "Danny (that's his name) - you drive a hard bargain, but that's a deal."
15 minutes later I had it in the back of my pickup and was driving home as quick as I could before anyone changed their minds. A sale's a sale! I went back the next week, dug around some more, and came up with a ton of S-100 bus cards - and got those for nothing as well. Every now and then I go back and look for other cards, but other than a stash of NOS S-100 edge connectors - I couldn't find anything else.
I later went online on to an Altair message group, posted what I had, and was told mine was rather unique, as mine had flat handle toggles, and the later versions all used round handled toggles. I was offered close to a grand for it in as-is sight unseen condition, but I turned it down. My ultimate goal is to restore it to operating condition. I have not plugged it in (beer can capacitor would probably explode and catch fire if I did). It needs a ton of TLC.
It has a full amount of RAM, and the 8080 card was replaced with a Z-80 card (common upgrade of the time); there's enough RAM to run CP/M if I wanted to. Various serial and parallel I/O cards, plus a fixed-sector floppy drive card (good luck finding a drive, let alone media, for anything short of a fortune). I later had others donate to me other cards and parts (including an interesting unpopulated 6502 processor card - but I don't know if the pinout is the same as what the Altair used, or if it's meant for some other S-100 bus computer - despite the S-100 bus supposedly being a standard, there were a few variations of it - as it goes).
I figure if I can get it working to some extent, and get the test program from the manual toggled in and working, I'll consider that a success. But what I'd like to do is take it further, getting my VT-100 terminal hooked up to it, and using some kind of method to boot either BASIC or CP/M from a flash drive or something (I know this can be done).
Sadly, I contacted the case manufacturer about getting a top cover, and had no luck. The company (I don't recall their name), used to make that case as an enclosure that tons of companies used for their equipment. They made it for decades, long after the Altair ceased to be a thing, long after MITS went under. I told them what I needed, and they told me they no longer made that enclosure since a couple of years prior, and no longer had the engineering drawings, either.
So - if I want that part fixed, I'll either have to custom make something myself, try to find the part other ways (maybe visit and talk to people at a vintage computer fest?), or pray I run across one somewhere. Another possibility, though it wouldn't be cheap, would be to have a machine shop make me a one-off of a top that'd fit. Honestly, something tells me the first option might be the best, and in spirit with the machine itself. I doubt it would hurt it's value any.
I am the curator of a small collection of retro computers here in Vienna [1], and have managed to get a few systems up and running again - nothing as cool as an Altair, but you've now gotten me motivated to have another look at it as a platform. There's a clone kit out there [2], but I'm not sure its available .. but for sure its something I'm going to check out in an emulator.
Got a place we might be able to follow you for future updates about your restoration project?
[1] - http://primitur.at/TIMETRON2019_Opening/
[2] - http://altairclone.com
Unfortunately, I didn't find an answer to something I suspect, but I don't know for certain? Maybe someone here could shed light on it?
He mentions that the AGC was the first computer to use integrated circuits, in the form of 5600 NOR gates, built into custom "logic modules", each set then wired together using a backplane or something like that. But take a look at those ICs - are they thru-hole, or SMT?
I seem to recall seeing pictures of the AGC - of other parts of it - and that it used a form of SMT in its construction; does anyone know if this is true? If so, would that also make it the first time SMT was used in a computer? Or were there other prior examples.
It seems also curious that if it was SMT being used - that such a thing was chosen for this critical piece of hardware, mounted on a rocket undergoing tremendous strain and vibration. I would think in that case DIP would be the better solution, ideally with extended pins for wire-wrap, then everything embedded in conformal coating after test - but while the DIP vs SMT question is up in the air, I've never seen any pictures that suggest wire wrapping was used, and nothing that showed anything like conformal coating.
So - can anyone shed any light on this?
The ICs were surface-mount flat packs onto boards in the modules. They were welded, not soldered. The modules plugged into the backplane connectors, which were wire-wrapped. The modules and backplane were encapsulated in plastic (epoxy or polyurethane) for flight. (This AGC was for ground testing, so it was not encapsulated.)
The non-logic circuitry was built with cordwood construction, i.e. the components were inserted into holes in the module. Point-to-point wires were welded onto the components on either side. These were also encapsulated in plastic.
I don't know if DIP integrated circuits would have been more reliable, but the DIP wasn't invented until 1964, too late for the AGC.
The Saturn V rocket contained a Launch Vehicle Digital Computer (LVDC), which was completely different and built by IBM. It also used surface-mount modules, but IBM's hybrid modules, not integrated circuits. So the AGC wasn't the only surface mount computer at the time. (Does anyone have a good history of surface mount?)
> if there were anyone who could do it, it's him
Honestly, Mike Stewart is the AGC super-expert who made the restoration possible. He knows way, way more about the AGC than I do.
> I managed to get everything to fit in one bank by reusing these 16 words for multiple purposes, but I spent a lot of time debugging problems when a variable clobbered a location still in use.
It could be fun to make a slightly higher level ad-hoc assembly language for solving these problems. For example SSA with basic blocks.
The estimate of the amount of electricity Bitcoin consumes (from a footnote) is distressing -- on the order of a small country...
This is one of the big reasons why I personally will never buy into a cryptocurrency (and especially bitcoin) -- too much wasted energy. If I want to engage in speculative trading, I'll do it in the stock market.
But what are you comparing the cost of running Bitcoin to, the entire world's banking infrastructure? Does this take into account all of money that has slipped through the cracks over the years?
i.e. https://www.thenation.com/article/pentagon-audit-budget-frau...
> "there were no ledger entries or receipts to back up how that $6.5 trillion supposedly was spent"
> "In all, at least a mind-boggling $21 trillion of Pentagon financial transactions between 1998 and 2015 could not be traced, documented, or explained"
I seem to remember that proof-of-stake has been coming 'next year' for about 3 years now.
Is it notable that proof-of-stake moves power from the means of production to the owners?
Just looked that last one up - yep.
http://www.righto.com/2018/09/the-printer-that-wouldnt-print...
The other two:
http://www.righto.com/2018/03/a-1970s-disk-drive-that-wouldn...
But I seem to be wrong on the Lisp machine? I couldn't find anything on it, but I could have sworn somebody has been working on getting one into operating condition for a while now (it's been a slow process)...
Anybody know?
A Waterdhavian gold dragon is a fictional possession (though, in concept, tangible).
A Bitcoin is intangible, but nonfictional.
This isn't an honest conversation though, because people are getting overly defensive about BTC.
Person A saying "this thing is totally a waste" and Person B responding "actually, no it's not, for these reasons" is not an example of Person B being "overly defensive". Especially when B provides rationalizations of their position and A does not.
And in no context I've ever seen does "fictional" mean "intangible". Even if it did, this would not be one of those times, as the context of the OC clearly indicates being dismissive of the currency as a whole, not just pointing out that it is intangible.
The original comment is obviously not using "fictional" to mean "intangible", as it is dismissive.
Since you used the word "fictional" as a continuation of the original usage without clarification at the time, one must assume, given the context, that you meant the same thing. Anything else would not be reasonable in context.
But again, nobody outside of this very discussion has ever used "fictional" when they really mean "intangible". "Imaginary" maybe, but not "fictional". I think we both know that this was a post-facto rationalization of your own unnecessarily dismissive comment.
I do appreciate you proving just how defensive Bitcoin people are, that is nice.
> If they're not fictional, show me a Bitcoin.
p.s. I'm not a Bitcoin person.
A TF2 hat, like other digital collectibles, is a nonfictional intangible object (which happens to represent a tangible object in the fiction), not an intangible fictional object.
For example, I don't know if you own any old jewelry with diamonds in it, but if you do, you can probably assume that it is worth less than you paid for it, given that diamonds can now be artificially produced with clarity / hardness / etc that parallels "natural" diamonds, and a big part of a diamond's price has historically been it's relative scarcity in nature.
This is a cool project that shows how far we've come. The Bitcoin network (wasteful or not) is currently capable of calculating 74.5 exahashes per second. You would need 10x74.5x10^18 Apollo Guidance Computers for that, you would probably need to build the equivalent of the Death Star to fit them all in the same place.
I think the ecoin concept by itself is a step forward for the progress of humanity, and without mining, the blockchain needs a lot less resources. How would you fairly distribute ecoins if mining is excised? No idea.
Without mining, you would have an even bigger problem than distributing coins, which is making the system secure.
People are working on solutions to that (I have no idea if they will prove fruitful).
I can argue that every form of money is fictional, it's only in the minds of people, it's a social construct. In fact, I can go further and say that there is absolutely nothing "physical" in this world, what is physical? just some sensations you're perceiving in your brain? Your brain is making up all the meaning.
There's another (weak) argument: somehow, building all that crazy tech and going to the moon is more important than taking care of humans who're suffering here on Earth. Why is going to the moon more productive?? who benefited from it?
I suspect this is just an intellectual circle jerk.
Compute? No. Energy? Yes, because today energy comes at the expense of environment which is running thin on the ground.
> I suspect this is just an intellectual circle jerk.
Come on, was that really necessary?
Someone just as judgemental could criticize HNers for all the energy they waste using high powered devices to sit on HN. Or air conditioning their entire house when they could simply sweat a little bit or use a fan in the one room they are in. Or driving to the movies when they could twiddle their thumbs at home for entertainment.
This is why it seems like such a belabored circlejerk.
Of course there is an ecosystem around regular money. There's no way it costs 1,000 times as much per dollar as printing the money, and, the major, universal complaint about bitcoin has been that it doesn't provide an alternative to all of the additional functions of the conventional banking system, so your point seems doubly void to me.
How can you consider my point "void" if it does provide some of the functions?
I can't understand how you can acknowledge the "ecosystem around regular money" and ignore it in the very next sentence and resort to your narrative of "1000 times the cost of printing a dollar", lol!
And you complain about hand-waving? italicizing "doesn't" isn't really a way to convince someone about what's missing.
And I don't understand what you don't understand about the factor of a thousand difference. Do you think that less than $100 in energy is dissipated in order to produce $100 worth of bitcoin, or do you think that the regular financial system costs more than a few pennies to produce and maintain $100? It's true that the cost of printing a $100 bill does not cover all of the functions of the banking system, but it's also true that dollars don't even have to be printed to be used, and those which exist only as ledger entries are far cheaper. So I think 0.1% is more like an upper bound than a lower one.
If it was so easy and cheap to print a 100$, why isn't everyone doing it? Obviously, it's not easy or cheap to do, and it's also illegal.
The US govt enforces the legality. So, you should factor in the cost of running the US govt and the cost of enforcing the law through the judicial system.
Suppose the US military disappeared overnight, what do you think will happen to the value of the dollar? you most certainly should factor in the cost of running the US army as well.
And then the banking ecosystem itself in the US, which is well over 10 trillion $ as of now.
The bitcoin ecosystem rolls all of these features into a single concise, self-policed network. I'd much rather have a permission-less SoV that is not controlled/backed by a nation state or a set of corporations (like Libra).
'perl4ever' should've tipped me off, I never learn :(. I don't think I have anymore time to argue with you.
I don't use perl any more - I discovered the joy of Visual Basic. :)
Anyway, see my other comment about the Federal Reserve. It's not that there are no other costs to maintaining dollars as a currency besides printing the paper money, it's that I think the printing costs obviously dwarf the rest.
Other way around.
Bitcoin is valued according to the market.
Mining is rewarded two ways: The inflationary (and reducing) coinbase reward for coin issuance, and transaction (txn) fees. The value of mining a block depends on these two factors, which both depend on the price of Bitcoin.
So long as the marginal cost of increasing hash rate is lower than the marginal gain of Bitcoins mined, hash rate will increase. When hash rates increase, mining difficulty is adjusted by the network to keep average block times near a 10 minute target.
With these costs and rewards, Bitcoin mining costs will reach an equilibrium near the value of mined Bitcoins, with two caveats:
* Risk from price variability and capital investment. Mining infrastructure takes time to deploy, so there is considerable lag between price and mining costs. And capital expenditures have opportunity costs, so deployed resources will reliably reflect predicted returns less these opportunity costs.
* Profit. The mining market will not mine Bitcoins for no profit (excepting brief periods, potentially) as there is no point to do so.
> it's a thousand times more expensive than conventional money
This is a comparison based on transactions, but why is that a suitable comparison? Money is both a medium of exchange and a store of value.
It's also absurd to consider that the costs of the US dollar are in printing alone; the value of US currency is underpinned by the entire US government. If they US did not have a powerful military, trade agreements reflecting that military, taxes that must be paid in dollars, etc., then dollars would not have the value they do.
The global monetary system - banks, sovereign nations, etc. - is extraordinarily expensive. But Bitcoin is a global currency, so this is the suitable comparison, even if it is very young and tremendously inflationary at the moment.
The only costs that are reasonable and appropriate to include, it seems to me, in an apples-to-apples comparison, are the costs of actually creating the money and regulating the supply.
So, ok, over and above the negligible amount to create money, it's plausible to include the cost of the Federal Reserve system, since that is what keeps the money supply stable, analogous to the mining algorithm.
The M2 money supply (a bit arbitrary, but for this comparison it's somewhat generous to the opposing view and anyway, I guess they don't publish M3 any more) is something like $14.5 trillion. The cost of the Federal Reserve is something like $5 billion[1] So the "maintenance" cost for a dollar is something like 3 cents every hundred years. To me, that seems de minimus and it also seems a person with some common sense would expect that without doing an analysis.
If you want to be obstinate about this, I suspect you will argue that bitcoin as it currently exists provides the facility to do transactions which I am not including the cost of with dollars, and my response is that the throughput of bitcoin is negligible and therefore not appropriate to equate with the capacity of the real financial system.
[1]https://www.federalreserve.gov/publications/2017-ar-federal-...
A current bitcoin transaction is equivalent to about 500kwh, which is about as much as a residential customer in the US uses in about two weeks. Or as a twitter user put it, "bitcoin is like leaving your car running over night to solve sudokus that you can trade for heroin."
And even though there is no current shortage of compute, opportunity cost still exists. Those units of compute or energy could be put to use that is a little more sensible.
We could be using the compute equivalent of the nation of Colombia to collectively simulate cancer drug molecules, or provide free resources to startups in the third world or just about anything that is more difficult than sending money from A to B, which we figured out how to do with much less resources decades ago. I think even buying a carrier pigeon might be more energy efficient (and possibly less prone to hacking) than cryptocurrency.
Get off HN and do it! what's stopping you??
> which we figured out how to do with much less resources decades ago. I think even buying a carrier pigeon might be more energy efficient (and possibly less prone to hacking) than cryptocurrency.
Sorry, you don't understand Bitcoin, you should do some research before making such shallow, reductionist arguments.
"Stone is better, why do you need metal? it's so much work to make metal implements"
"We have horses, why do you need cars that pollute the air??"
"Why do you need a smart phone, I have a desktop computer to do all that"
"Why do you need the internet?, we've already figured out how to send messages to each other via mail"
Stop acting like this is a zero sum game.
[1] https://www.reddit.com/r/AskScienceDiscussion/comments/9fax9...
Computing only gets you so far. The AGC was able to be so slow and still get the job done because what it needed to do was not "hard": it calculated orbital equations and monitored sensors. It didn't use AI, it didn't need to hash anything, it didn't work with databases or networks, it didn't render graphics in 3D 4K - all of these, we expect from modern phones.
If we want to advance space travel, the problem is not with computing. The problem is with chemistry (mostly), physics (to a lesser degree), and politics (let's be real, this is 99% of the battle).
The real tragedy of the cryptocurrency gold rush is that by providing a profit motive it decimated these kinds of collaborative computing projects.
It's just such a perfect metaphor for how we acquired enormous power but only used it for {insert your favorite rant subject}.
Essentially a blockchain is a database of time and energy booked to a contrived measurable $particularActivity. This activity correlates to some past opportunity cost expense (and presumably a counterpart defensive loss aversion behavior expected in the future) on the part of the person burning time and resources to support that activity. A kind of rube goldberg electricity meter with a public log, that functions as a record of someone's ante in a super simple long running casino game.
The more you spend on/invest in this activity, the less inclined you are to forfeit your recorded proof of doing it for less than whatever you consider a fair return on that sunk cost.
This mechanism is fairly universal and applies to everything from hobbies to friendships to careers. What bitcoin (any blockchain really) does is just strip the commitment-payoff mechanic present in almost all human endeavors down to a cartoonishly simplified model and tie a bunch of charts and metrics to it, and use these to collectively goose our animal chase instincts into overdrive.
Also funny is the implication that the astronauts would have had to be careful of malware in addition to the occasional unaccounted for floating bolt.
I have to therefore say (will leave out the 'honestly') that if I was 'that smart' I'd use my brains to do something that would put money in my pocket.