I have a concrete suggestion. There are several combinatorics-related sequences I can think of that are well suited for highly parallelized distributed computing, but for which only a few terms have been explicitly calculated due to exponential computational complexity. Here are two examples: OEIS A088672 and OEIS A028420. I have already written fully parallelized and optimized code for both of these; other sequences which I have in mind would be similarly easy to code.
Actually, this summer I had intended to create my own BOINC project along these lines. One of my friends was supposed to take care of the server side of things, but he fell through on his commitment. So the idea hasn't gone anywhere yet.
Here's an idea though. If someone wants to work with me to get a project up and running (I manage code for workunits, other people manage servers), then that project could be used as a foundation for more ambitious objectives, like the 'MicroGrid' concept I've heard mentioned before. The key point would be that the combinatorics problems I proposed would provide a steady and inexhaustible supply of workunits.
@h202 Interesting - please get in touch with me on [email protected].
@pomegranatepool @hotbit
If you're interested, we should try to make it happen. It's about time I follow through all the way with one of my ideas. I can dust off my old code and make sure it's BOINC-worthy. Shouldn't take too long in principle.
What would be a good way to communicate? Creating a new Discord server? It looks like that's free.
@cm-steem You're right that getting a basic test server up and running is easy. On the other hand, getting a fully fledged project site up -- complete with dedicated servers and robust workunit pipeline --seems like a lot more work.
Last weekend I was just checking out how to set up a server... nice timing. But I have close to zero sys admin experience. What about hardware? I could be interested to join if possible.
Running a BOINC project is pretty simple using boinc server docker, you can distribute work through the supported boinc-to-docker tool. "A 10min task"
its not hard to compile the damn boinc server from source code located here @ https://boinc.berkeley.edu/trac/wiki/SourceCodeGit vs using some cheesy docker bullshit... Yes lets run a visualization on virtual server GO RESOURCES! Did you notice Win10 incorporated docker native and then dev dropped , gee go wonder why... There are project admins like yoyo whom are helpful and you can ask to help setup your server and even help admin.. But if you cant compile something yourself you shouldn't be running it , because who the hells going to maintain it if you cant?
You do realize the linked source code is for the BOINC CLIENT, right? @cm-steem is talking about using Marius' docker container for the BOINC SERVER.