Computational power

[Anonymous No.933773]

Hi /biz/, I do data crunching for a very big insurance company (we're talking millions of customers here).

That means that, whenever the systems I manage are idle, I have access to an enormous processing power.

Think several servers with 150-200 GB of RAM and 128 processors.


How can I profit from this?

[Anonymous No.933781]

Mine coins.
Run maths algorithms.
Brute force attack some cryptographic funtion that's practically infeasible to it.

[Anonymous No.933785]

Mining bitcoins seems the obvious thing to do.

[Anonymous No.933834]

>>933785
>Mining bitcoins

Mining bitcoins requires access to the Internet.
These systems are, quite obviously, behind closed doors.

>>933781
>Run maths algorithms.

This is the kind of stuff I was talking about.
What kind of math algorithm could I run in a highly parallelized environment, that requires reasonably easy operations but just needs tons of computing power?

I was thinking some prediction algorithm for sport bets or even the stock market, but I have no idea where to start.

Yesterday I gave a shot at an extremely crude prediction algorithm, and it couldn't crack "result = (a+b)/c", lol.

[Anonymous No.933864]

>>933834
Finding prime numbers.
Solve Partial differential equations(seriously fuck those)

[Anonymous No.933879]

>>933864
how to profit off this?

>>933834
It sounds like you don't have the high-level maths to do sports/stock betting. Seriously, the algorithms that the market run on have hundreds of cumulative hours of expert mathmaticians' proprietary work in them. Even if you were the best on the planet at maths, you would not be able to compete. Those algos are built on a foundation that has been developed for years, and each revision has ten guys working on it. Those ten guys understood shit like pic related in highschool and fuck around with stuff like that for fun.

I've got a lowly finance degree, but pragmatically, I think your best bet is to calculate average free server time per month, then figure out the curve of electrical+other overhead cost between 50% and 100% load. mark up that price 10% and see how that compares to other datafarm pricing, then present that to your superiors like "hey we can undercut these other guys, lets form a subsidary front company to sell this extra server time." Obviously they should put you in charge of the new company/department and give you a raise along with it.

[Anonymous No.933887]

>>933773
Brute-force crack the user database password.
After that demand your director a bribe in exchange of the data back. Instead of giving him sell it to the closest competitor.

[Anonymous No.933888]

>>933773
Progressive?

I work for em too

[Anonymous No.933953]

>>933773

>>933879
I like this Anon's plan. If you don't have access to the internet on those machines, I don't think you can do any under-the-table data crunching to generate profit.

Scientists always have a need for gigantic data crunching, but Idk how much they'd pay (if at all). Physicists have equations, astonomers have calculations, proteins need to be folded, etc. They crowdsouce that though so idk if you could get any money from them. See: SETI@home, Folding@home, etc.

[Anonymous No.934672]

This sounds to me like the perfect opportunity to train some neural networks. Maybe even on your companies data. Teach some RNN to detect interesting markers in your data, or maybe stock data, or whatever else you can get your hands on.

[Anonymous No.934738]

Pretty sure this was in an episode of Halt and Catch Fire. The guy leases the servers to a startup company that badly needs them

[Anonymous No.934747]

You're a data cruncher but not the sysadmin who will catch your ass as soon as you start doing something shady.