Minting Money with Monero and CPU Vector Intrinsics

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.

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.

There is a huge difference between unoptimized code and maliciously slow code. This case definitely falls in the second category.

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. :)

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.

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/

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.

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...