Minting Money with Monero and CPU Vector Intrinsics(da-data.blogspot.com)
da-data.blogspot.com
Minting Money with Monero and CPU Vector Intrinsics
http://da-data.blogspot.com/2014/08/minting-money-with-monero-and-cpu.html
6 comments
Love the story, reminds me of the early days of Primecoin where I had my first exposure to cryptocurrency code. I spent the first two months of it fighting to stay ahead of the curve as well and had a blast. Not nearly as successfully as Dave however and my wife wasn't too happy about me spending every waking hour when home from my full time job on it.
Much respect for profiting hand over fist from pure knowledge. But the fact that this "opportunity" existed in the code is really damning for the Monero project. It highlights the staggering amounts of greed and hushed fraud rampant in the Bitcoin space. I don't fault him for selling everything.
Crptocurrency is all about giving economic incentives to do the right thing.
In most fields, fraud/lying/cheating are ethical failures. In cryptocurrency, they are bugs that need to be fixed and giving them moral assignations isn't helpful, IMHO. A cryptocurrency that fails doesn't do so because "bad people ruined it" but only fails because of its own shortcomings.
In most fields, fraud/lying/cheating are ethical failures. In cryptocurrency, they are bugs that need to be fixed and giving them moral assignations isn't helpful, IMHO. A cryptocurrency that fails doesn't do so because "bad people ruined it" but only fails because of its own shortcomings.
I agree there is a lot of fraud and greed in the cryptocoin community which makes it risky for people new to it or the naive. For people who have done some reading or hung around for a while it's usually pretty easy to spot most scams. In cases like Monero it was probably a something the developers missed. The important thing to note is that the community fixed that oversight within a month or two.
New proof-of-work schemes in crypto-currencies should ideally have optimized implementations from day 0, to prevent the situation described in the article where some people can mine much more efficiently than others.
To this end, I offer several bounties for my Cuckoo Cycle proof-of-work scheme, which David has helped refine, and which has yet to be deployed by any coin.
See https://github.com/tromp/cuckoo for details and a whitepaper.
To this end, I offer several bounties for my Cuckoo Cycle proof-of-work scheme, which David has helped refine, and which has yet to be deployed by any coin.
See https://github.com/tromp/cuckoo for details and a whitepaper.
I remember when Bytecoin was released; the claims to have been used for years were clearly wrong, but I couldn't figure out what the real story was skimming the forum posts. I'm glad someone figured it out!
Thanks. I don't deserve too much of the credit there, though -- I think a bunch of people started realizing it at the same time. It was just too weird to go unnoticed forever.
Someone should see if they can dig up transaction volume data to estimate how much the Bytecoin people made. :)
Someone should see if they can dig up transaction volume data to estimate how much the Bytecoin people made. :)
Premining and instamining (where they just start the coin with x coins in the devs wallet) are fairly common practices among new cryptocoins. They are usually frowned upon by the community but not always as the coin developers will sometimes keep a 1-2% for use promoting and developing the coin. Which most people consider fair. An 80% premine like bytecoin is ridiculous.
It is very easy to tell if a coin has been premined by checking the state of the block chain for the number of outstanding coins. The coins with large premines are usually outed withing an hour of their release.
It is very easy to tell if a coin has been premined by checking the state of the block chain for the number of outstanding coins. The coins with large premines are usually outed withing an hour of their release.
I realize that, which is why I didn't believe the Bytecoin claims to have been used for years. (That and I spend a lot of time on Tor so I was skeptical I'd've never heard of it if there really was an active community using it.) But that didn't explain where the huge apparently-PoW-intensive blockchain came from.
Any suggestions/resources on learning how to implement these kinds of optimizations?
Agner Fog's guides and the Intel reference documents are the best resources that I've found:
http://www.agner.org/optimize/
http://www.intel.com/content/www/us/en/processors/architectu...
IACA, perf, pmu-tools, and likwid are very useful tools.
https://software.intel.com/en-us/articles/intel-architecture...
https://perf.wiki.kernel.org/index.php/Main_Page
https://github.com/andikleen/pmu-tools
https://code.google.com/p/likwid/
http://www.agner.org/optimize/
http://www.intel.com/content/www/us/en/processors/architectu...
IACA, perf, pmu-tools, and likwid are very useful tools.
https://software.intel.com/en-us/articles/intel-architecture...
https://perf.wiki.kernel.org/index.php/Main_Page
https://github.com/andikleen/pmu-tools
https://code.google.com/p/likwid/
In addition to that great list, see also "Hacker's Delight", and several of Jon Bentley's books (Programming Pearls, Writing Efficient Programs, etc.). They're dated, but the basic ideas are eternal. The application to x86 requires combining those ideas with the specifics you get from Agner Fog's references.
On sitkack's recommendation (see my comment history) I'm currently reading through Rick Booth's "Inner Loops". It's another dated classic that may be useful when used in combination with other more modern resources. Some of Booth's explanations micro-ops and superscalar dispatch are great, although it would be difficult for a beginner to differentiate what still applies today from what doesn't.
What is the incentive for the open source developers to release such fast miners? Dave says that one guy could beat him and "everyone" would be running his code. Why wouldn't the other guy run his privately, thus effectively splitting the money with other smart folks? You can always release it later to get the fame and ego points for being brilliant.
Style difference? yvg makes money from a built-in 1-2% dev fee on his (binary) miners. They are fast, support many pools and protocols, and are robust enough that you can pretty much start them running in /etc/rc.local and trust that they'll still be running a month later. I don't know what his total crypto income is, but I suspect he's doing OK. In contrast, you'd cringe at the horrible code I cranked out for some of these miners. Quality, high-performance, and delivered fast: pick two. It wouldn't surprise me if his business model was, in the long term, more profitable. He argues that it is. I haven't wanted to lock myself into providing support for software long term, so I picked a very different model.
I believe, though, that there's a mix of both in many situations. I think some of the miner devs release some code openly as an advertisement for their skills and then do some contract work. The shady ones probably have opportunities to collaborate with coin developers to pre-develop a secret GPU miner for a "CPU only" coin so that the devs can instamine it. There's at least one I know of who's customized his code and sold it to a botnet herder.
I love the question someone raised earlier: If someone can get a huge advantage that "kills" the coin, it's a flaw in the coin design. Is there a way to have the coin dynamics incentivize developers to make fast code and release it? That's one to chew on...
I believe, though, that there's a mix of both in many situations. I think some of the miner devs release some code openly as an advertisement for their skills and then do some contract work. The shady ones probably have opportunities to collaborate with coin developers to pre-develop a secret GPU miner for a "CPU only" coin so that the devs can instamine it. There's at least one I know of who's customized his code and sold it to a botnet herder.
I love the question someone raised earlier: If someone can get a huge advantage that "kills" the coin, it's a flaw in the coin design. Is there a way to have the coin dynamics incentivize developers to make fast code and release it? That's one to chew on...
I still remember this: (notice the GPL violation and 2% donation) http://cryptomining-blog.com/2496-new-sgminer-fork-for-x11-m...
Entirely possible they release a generation of code only when they're already on the next generation. Sort of like Quake, maybe.
Of course, even if you're 3 gens ahead, that's still better than updating everyone else to 2 gens behind you.
Of course, even if you're 3 gens ahead, that's still better than updating everyone else to 2 gens behind you.