A fool and his not-really-money are soon parted
-
@blakeyrat said in A fool and his not-really-money are soon parted:
I'd love to see the books on how much each of these machines cost Bitmain.
I'd be willing to bet that the profile of costs is exactly the same as with most hardware that uses ASICs. Big one-off cost to make the first batch; pretty cheap to make more after that as running off another wafer is a relatively straight forward industrial process. Those sorts of economics are why ASICs are outside the reach of hobbyists, yet fairly cheap once the first batch is done.
Unless they go for wafer sharing and stuff like that, which alters the cost profile, but that sort of thing is dumb to do when you're going to be shipping thousands of units and hardware makers tend to be very cagey about their IP.
There's also a lot of costs involved in the design stage before manufacture. They're spent on things like simulating the chip (including all the tricky capacitance interactions) and making sure that everything is reliable and testable (since the manufacturing stage has an automated testing sub-stage within it; the rate of failure depends on the exact process being used, but it is very much non-zero and automated tests are vital).
-
Give hope, by destroying the environment for your children.
INB4: Poe or Noe is that way .
-
@dkf said in A fool and his not-really-money are soon parted:
They're spent on things like simulating the chip
(including all the tricky capacitance interactions)
For analog stuff and memories, maybe; not so much for standard digital logic. Digital logic isn't terribly sensitive to noise and crosstalk, and any given process has rules that, if followed (and they're checked automatically as part of the flow), should keep them within limits that won't cause incorrect logic operation.
and making sure that everything is reliable and testable
And the tests are efficient. Making chips is a parallel process — a lot of wafers are processed at the same time, and each wafer has many chips — but testing them is sequential. And time is money; a couple extra seconds matters when you multiply by the number of chips you're making.
(since the manufacturing stage has an automated testing sub-stage within it; the rate of failure depends on the exact process being used, but it is very much non-zero and automated tests are vital).
Typically, two automated testing sub-stages, wafer sort and final packaged parts. Wafer sort, as the name implies, is when you have a big slice of silicon with a bunch of chips on it; the tester has a probe that makes contact with one of the chips, supplies voltage, drives inputs, and measures outputs, marks whether it's good or bad with a little ink dot, then moves to the next chip. Assembly and (final) test is expensive, and you don't want to waste money putting defective chips through it, so they're found and discarded before the wafer is cut into individual chips. The tester also keeps statistics of things like whether chips from one area of the wafers are more likely to be defective than others, which can indicate a process problem. (There are also a few test structures on each wafer that can give you a lot of information about how well the process is adhering to its spec.) Getting a wafer with no bad chips is a rare occurrence, so much so that sometimes they don't even use the chips from the wafer, but frame it and hang it on the wall as a trophy.
Final test is, of course, testing the chips after they've been packaged and tests the connections between the chip and the package, as well as the chip itself. Sometimes, depending on the value of the chips and the market you're selling to, some or all of them may be subjected to burn-in, which means operating them under stressful conditions for some period of time to eliminate infant mortality. Even so, some undetected defects will slip through; this is qualified as DPM, and for some customers, even 1 DPM is too many.
-
@hardwaregeek said in A fool and his not-really-money are soon parted:
some or all of them may be subjected to burn-in, which means operating them under stressful conditions for some period of time to eliminate infant mortality.
Is it really eliminating infant mortality if you still kill off all the weak children, but make sure to hide the bodies where nobody will ever find them?
-
@scarlet_manuka As long as customers never see the weak babies, no. The infant mortality that matters most is chips that die in the hands of customers (although, of course, if you're getting a lot of failures during burn-in, you have a manufacturing problem that's costing you money, so you really want to fix it). But no matter how long you test, there's always the possibility of a baby that was right on the verge of dieing when the test ended. And of course burn-in is really expensive, so you don't want to do any more of it than you really need to.
-
@hardwaregeek said in A fool and his not-really-money are soon parted:
Wafer sort, as the name implies, is when you have a big slice of silicon with a bunch of chips on it; the tester has a probe that makes contact with one of the chips, supplies voltage, drives inputs, and measures outputs, marks whether it's good or bad with a little ink dot, then moves to the next chip.
Also, some chips can tolerate a certain level of defects in them without being declared failures. For example, our current production chips can lose one CPU core per chip (out of 18) and still be good enough; that boosts our yield enough to make a big difference to our profitability. The software we run on them copes with whatever actually worked (except for a small number of annoying failure modes that it's awkward to test for before the package gets to us).
-
-
@blakeyrat said in A fool and his not-really-money are soon parted:
They sell based on the machine's income potential
Business venture: Virtual Bitcoin Mining.
"Can't afford the bitcoin mining machine? Well, virtualization is here to help you! For only $999.00, I will sell you a virtual bitcoin mining machine. We've virtualized and emulated all the major bitcoin computers. Now you can run the hardware of your dreams without ever owning the hardware! Just run it on your own, existing computer, and watch the bitcoins flow in!"
-
-
Status: Foolin'
-
@blakeyrat said in A fool and his not-really-money are soon parted:
Status: Foolin'
How's your new space heater so far?
-
@tsaukpaetra
I mean, just the fact that it actually got delivered puts it several standard deviations ahead of most ASICS...
-
@tsaukpaetra said in A fool and his not-really-money are soon parted:
How's your new space heater so far?
People were not kidding when they talked about how fucking LOUD Antminer fans are-- to make it worse, this machine shipped with no fan controller, so it was even super mega-louder than that because the fans always ran at 100% (while reporting 0 RPM to the dashboard.)
So I've already probably voided the warranty: replaced the Antminer fans with 2 coolermaster fans, one of which I had sitting in a box as a spare and another I yanked out of my PC. Ah. My ears thank me. The hottest ASIC chip temp appears to be hovering around 80, which is bad compared to the stock fans, but well below their tolerance level.
So far this machine's earned over $6 and it's only been plugged in about an hour. It's reporting a large number of discarded shares, which kind of worries me... not sure what's up with that. But if it earns $6/hour while I'm sitting on my ass I can't really get upset about that.
Note to self: order more Coolermaster fans.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
But if it earns $6/hour while I'm sitting on my ass I can't really get upset about that.
And what's the cost of the electricity it uses per hour?
-
@masonwheeler said in A fool and his not-really-money are soon parted:
@blakeyrat said in A fool and his not-really-money are soon parted:
But if it earns $6/hour while I'm sitting on my ass I can't really get upset about that.
And what's the cost of the electricity it uses per hour?
Plus depreciation of the antminer?
-
@masonwheeler said in A fool and his not-really-money are soon parted:
And what's the cost of the electricity it uses per hour?
It consumes 360 watts and I pay $0.11 per kwh, so you do the math. No, seriously.
At this rate of coin earning, it's not even worth calculating the power. It'll be like maybe $25-50 a month more, and this machine will earn that in like a day. A gaming PC going full-out will consume more than 360 watts, generally.
@pleegwat said in A fool and his not-really-money are soon parted:
Plus depreciation of the antminer?
You run them as long as they live, then throw them away. It's not something that has resale value.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
@pleegwat said in A fool and his not-really-money are soon parted:
Plus depreciation of the antminer?
You run them as long as they live, then throw them away. It's not something that has resale value.
Depreciation isn't really about resale value for a business asset. It's about amortizing the expense of the asset over its expected lifetime.
-
@boomzilla Well the cost of it after Bitmain's refund was about $1950, plus another $160 for the PSU. I'll do the math later, but I am starving and have to grab chow now. Spent all morning dicking with this new toy, I need to do some real work or I'll get fired.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
I need to do some real work or I'll get fired.
Sucker. Sounds like you need a better job.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
I pay $0.11 per kwh,
That's cheap.
Still I guess all of this is most economical in either China, where dirty energy is super cheap, or way up north where you need heating anyways.What are you mining with this, bitcoins or some altcoins?
-
@topspin said in A fool and his not-really-money are soon parted:
What are you mining with this, bitcoins or some altcoins?
It's an Antminer B3, it mines Bytom.
-
Conversion from BYTOM (BTM) to US Dollar (USD) 17.25216815 BTM = 14.56 USD
This stupid block of aluminum is making more per hour than my first like 3 jobs.
Is it possible that I am... not a fool?
-
@blakeyrat said in A fool and his not-really-money are soon parted:
Conversion from BYTOM (BTM) to US Dollar (USD) 17.25216815 BTM = 14.56 USD
This stupid block of aluminum is making more per hour than my first like 3 jobs.
Did you find a way to convert it to real cash yet?
-
@timebandit said in A fool and his not-really-money are soon parted:
Did you find a way to convert it to real cash yet?
Nope!
-
@blakeyrat said in A fool and his not-really-money are soon parted:
@timebandit said in A fool and his not-really-money are soon parted:
Did you find a way to convert it to real cash yet?
Nope!
Do you still want an answer for your question?
Is it possible that I am... not a fool?
-
Fans swapped for less deafeningly-loud versions: check
Maximum ASIC temperature reached with quiet fans: 73 C
Wifi repeater installed so I can place it in my utility closet where it won't be in the way: check
Amount earned so far: 143.88395326 Bytom
-
@blakeyrat said in A fool and his not-really-money are soon parted:
place it in my utility closet
If it is a closed space, the fans will not help very much?
Bytom
I always think first about Bytom, a city in Poland.
-
@adynathos said in A fool and his not-really-money are soon parted:
I always think first about Byto
nm, a city in Poland.Which is known for mines, btw.
-
@adynathos said in A fool and his not-really-money are soon parted:
If it is a closed space, the fans will not help very much?
- It's not that closed (it has one of those slatted doors)
- This machine only pumps out 360 watts of heat. It's not like an S9 which is spitting out 1200 watts all the time.
-
If this thing is specialised for mining a single type of coin, does that mean you're gambling on the value of said coin staying at least somewhat high until you've paid off the mining rig?
-
@jaloopa said in A fool and his not-really-money are soon parted:
If this thing is specialised for mining a single type of coin, does that mean you're gambling on the value of said coin staying at least somewhat high until you've paid off the mining rig?
Kinda.
It mines Bytom, which is a new(ish) coin that's specifically designed to be mined by ASICs. Bitmain and the company that made Bytom teamed-up on it, since Bitmain's business model is at risk due to:
- An increasing number of coins that are forking themselves specifically to prevent ASIC mining of them (Bitmain has an ASIC coming out in a few weeks that supports 4 coins-- 3 of those 4 coins have announced that they're going to fork to sabotage Bitmain's ASIC miners.)
- The old standby SHA-256 coins quickly becoming a losing proposition due to difficulty increases. (It'd be hard to justify running an Antminer S9 in the US unless you literally got your power for free.)
Bitmain's such a big player in the first I'm not worried about the coin value nose diving before the machine's paid for itself. At the current rate, I only need a month worth of mining.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
An increasing number of coins that are forking themselves specifically to prevent ASIC mining of them
Maybe this way the cryptocurrency people will sponsor research on more efficient alternatives to FPGAs, or better memory throughput for them.
It would be nice to have this kind of JIT-hardware.
-
@adynathos They do is because they're worried that the few companies capable of creating power efficient ASICs will just put them all in a giant data center and get close to 51% control of the currency. (Since the currencies are concensus-based, if you can get a majority of mining power you can arbitrarily approve transactions that should have been rejected.)
I think it's kind of a wacky belief, but there you go.
There's also an argument to keeping the barrier of entry low. For example, I think Amazon badly needs to split their stock at least 10:1 because their current price is over $1500. That's completely pricing small investors out of the market; unless you have tons of disposable income (by normal standards, not 1%-er Amazon employee standards) it's literally impossible for you to buy a share of Amazon. IMO that's just unethical on Amazon's part.
But in the crypto world, GPU miners aren't really much cheaper than ASIC miners, so that argument doesn't really hole water. Bitmain has some models that are $600 or less.
-
@blakeyrat
All it really is is bog standard free market cartelism. The existing miners are using GPU rigs that would get turfed out by ASICs, so they’re using their existing market power to block the new entrants.Unregulated capitalism leads to monopolist behavior, news at 11.
-
@izzion said in A fool and his not-really-money are soon parted:
Unregulated capitalism leads to monopolist behavior, news at 11.
Of a stupid "currency" that's only worth anything because some people are bigger idiots than others.
-
Speaking of fools:
What the HOLY SHIT are these guys selling? Other than the word "blockchain" I can't make heads or tails out of anything on this site.
If you can figure it out they're apparently giving away $5,000,000. But it's unclear if it's actually $5,000,000 or just that amount in NAS tokens. Who knows.
JavaScript's involved somehow.
-
-
@blakeyrat Ok I swapped the fans again to Coolermaster Jetflo 120 fans, THESE ARE AMAZING FANS I AM NOT KIDDING. They're louder than the lower-end Coolermaster fans I had in that last photo, but still orders of magnitude quieter than the ones Bitmain put on the machine and they're 4-pin so they work with the machine's fan controller, should they ever update the firmware to give it some ;)
(Bitmain puts in some weird keying on their fan connectors to prevent the use of off-the-shelf fans-- I just shaved the keys off the fan plugs and they fit and work fine.)
Moved it into my little utility closet where it can buzz away forever and I'll barely notice it.
-
-
@blakeyrat said in A fool and his not-really-money are soon parted:
What the HOLY SHIT are these guys selling? Other than the word "blockchain" I can't make heads or tails out of anything on this site.
That does not increase their credibility:
Then again, it is not worse than a currency whose main property is that it can be mined on ASICs and its purpose is allowing Bitmain to sell more hardware.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
If you can figure it out they're apparently giving away $5,000,000. But it's unclear if it's actually $5,000,000 or just that amount in NAS tokens. Who knows.
JavaScript's involved somehow.Yeah. Way down in the weeds:
About the Nebula 1.0 Eagle Nebula Mainnet
The Nebula 1.0 Eagle Nebula mainnet has all the features of Ethereum and surpasses the second-generation blockchain in several ways:
✦ Nebulas is developer-friendly and supports the use of JavaScript to write smart contracts and DApps, making it easier for anyone to get started;
✦ Nebulas is the only blockchain that implements the world-renown LLVM compiler, features superior performance through concurrent technology, with a transaction processing capability of 2000TPS;
✦ Nebulas is more secure, stable, and has strong expandability. It also provides novel measures to invoke smart contracts and upgrade protocols.But everything they talk about is paying out in "NAS" which I guess is their own coin or whatever. It's all about writing DApps (decentralized apps) which appear to be apps that...do blockchain operations on their blockchain?
-
@blakeyrat said in A fool and his not-really-money are soon parted:
So I've already probably voided the warranty: replaced the Antminer fans with 2 coolermaster fans
I'm sure you've probably already voided the warranty by bitmining.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
Telegram’s billion-dollar ICO has become a mess
Hmm...
Looking beyond the messy sales process, the project has been heavily criticized for ambiguity. Pantera Capital’s Charles Noyes called it “opportunistic” with a white paper that is “essentially a wishlist of things they want to have, and how it will work assuming that their wishlist doesn’t crash and burn.” MIT Technology Review called it “bold but short on ideas,” while others have pointed out that much of the technical theory laid out in the document is recycled from other projects.
This sounds familiar.... ah yes, here we go, ~2 months ago:
https://what.thedailywtf.com/post/1313141
"Telegram ICO: Scam Among Cryptocurrency Scams?"
http://cryptocafe.tech/crypto-news/telegram-ico-scam-among-cryptocurrency-scams
-
Speaking of revisiting, I wonder how those "top 4 Japan Crypto exchanges" are doing?
https://what.thedailywtf.com/topic/24242/a-fool-and-his-not-really-money-are-soon-parted/664
Number 4: SBI Virtual Currencies
The supposedly bank-backed exchange that attracted investors is-- delayed. Again. Until summer 2018. Why have they been delaying since December until summer, which means anywhere from a 6-9 month delay? "To tighten up security". Speculation is because the Japan SEC is scrutinizing harder these days.
Which means one of a few things:
- They were originally going to launch in a state where security was 6-9 months behind being "acceptable"
- They are so incompetent that they require 6-9 months to implement acceptable security.
Either way, not a confidence giver.
And how is "Ripple", the coin they were laying their hopes on, doing? Last we checked it was lauded as a miracle by surging to $3USD/coin-- if you ignore the immediate plunge back to < $1USD per coin.
https://i.imgur.com/iBgfxmV.png
The plunge continued until it bottomed out at $0.47USD / coin, where it stayed for an ENTIRE FUCKING MONTH before limping back up to $0.80USD/coin. THE WAY OF THE FUTURE!
BTW, that crash is described thusly: "RIPPLE (XRP) continues to crash overnight suffering heavy losses over a 24-hour period to remain the worst performing major cryptocurrency on the market." That crash, along with a dip in some other coins, wiped out "$40B Billion With A B" from the market.
The reason given boils down to "uh-oh, regulators are looking our way, quick run away from the scam!" (https://www.express.co.uk/finance/city/939305/Ripple-price-crash-why-is-ripple-going-down-why-is-ripple-crashing)
Number 3: Zaif
Last we checked, the thread had covered their "oopsie" glitch that allowed people to buy coins for $0. That's pretty much polluted their search results. But they seem to have been keeping their nose clean since.
Number 2: bitFlyer
This was the one that actually had details on their security measures, and it was rated as
So this one is "hasn't fucked up yet". Let's revisit it in, say, a month or two?
Looks like they've been hiring former Wall Street bigshots to bring insight and expertise to their market, have also Tom Love, co-creator of the programming language Objective-C to help improve their blockchain and make it dev friendly, and have put in much stricter user-verification process to meet upcoming regulation.
So... this may actually be the first ever non-wtf, developer-friendly, consumer-focused, security-minded and regulation-observing exchange yet. We'll revisit them to see if they've fucked up yet.
And finally.
Number 1: Coincheck
Last we checked, they'd just been hacked for $400M USD in "the biggest coin heist" ever.
How are they doing now?
They're being sued by 132 investors who want their money back plus 5% annual monies for damages + erroding trust in their marked, and they're being sued by other groups of customers who couldn't withdraw, and ... well, just google "coincheck lawsuit", there's too many to list.
So an investment firm stepped in, bought everyone's shares and took over the company. The current execs have been demoted, and a new President is in place. They hope one day to tighten security and offer ICOs. They dropped $34M on that acquisition. Have fun.
Keep in mind, this is the exchange that promised to "pay back all the people who lost money" in the hack. Yes, normally when someone robs a bank, the bank covers the lost money. But the Japan SEC heard that statement and basically said "bullshit". Because, y'see, the FSA had ALREADY told Coincheck to fix their shit a long time ago, which is why Coincheck was never approved as an exchange. So the FSA has demanded Coincheck to report how they've fixed their security, and will be doing an ONSITE inspection of Coincheck to see if they are even financially capable of fulfilling that "yo we promize" promise.
It took them until March 12, 2018 to announce HOW they would do that. I can't find many articles confirming it, but it appears they actually went through with paying out 88.549 JPY / token, which is way more than the coins are currently worth (because Coincheck crashed the market). That money had to come out of their own capital. Which probably put them in the position to be bought out.
I LOL'd when I saw a couple articles title things like "Coincheck an example of how to bounce back after a hack". Yes, losing all trust in your platform, passing out half a billion dollars of your own money, narrowly avoiding bankruptcy, then being bought out and having the execs lose their jobs-- PERFECT example of a bounce back. Good jorb.
-
@lorne-kates said in A fool and his not-really-money are soon parted:
So... this may actually be the first ever non-wtf, developer-friendly, consumer-focused, security-minded and regulation-observing exchange yet. We'll revisit them to see if they've fucked up yet.
You don't think Coinbase is that?
Or are you only evaluating Japanese exchanges for some reason?
-
@blakeyrat said in A fool and his not-really-money are soon parted:
Or are you only evaluating Japanese exchanges for some reason?
Yes. There was an article about "the top 4 Japan exchanges" that the post was about. So I've been evaluating how "the best of the best" do.
-
@blakeyrat said in A fool and his not-really-money are soon parted:
NAS tokens
You need to be in a New York state of mind to use them.
-
Tools
-
I should add more buzzwords to my CV. I only have 'IoT' out of the word-salad in that press release.
“Blockchain 3.0: the intersection of Blockchain, IoT, big data, AI and the Intelligent Edge,”
On the other hand...
-
@cursorkeys had a onebox in A fool and his not-really-money are soon parted that said:
blockchain-based data platform