Ask HN: What are some worthy non-cryto uses of excess home compute nowadays?
When I was a kid, I remember participating in Folding@Home. Unfortunately, the project doesn't appear nearly as active as when I was younger. What do people do with excess home compute (if anything) nowadays?
I'm not interested in crypto. I originally built my home server for academic use (computational physics), but after changing in security I've come to use it as a fuzzing playground. Oftentimes I kick off a fuzzer on some OSS, e.g., GCC, v8, standard *nix binaries, and let it run for days/weeks/months. If I find anything I submit it to the appropriate people or make a PR, if Im motivated enough that week This sounds very interesting and potentially quite fun. How would one get into this? Is there a specific tool chain you use? Learning how to is half the fun! There's a bunch of good tutorials out there on [dumb] fuzzing (presumably where you'll start). One starting point I'd recommend is taking a binary that accepts input from stdin and making some proof-of-concepts with AFL (https://lcamtuf.coredump.cx/afl/). If you'd rather start from a code/library perspective (and not CLI), I'd recommend libfuzzer (https://github.com/Dor1s/libfuzzer-workshop/). There's a lot of other fuzzers, techniques, and depth to the field, but I'd recommend inch worming through (speed up as you gain more comfort). The Fuzzing Book is good to help you understand the logic behind techniques and strategies (https://www.fuzzingbook.org/) As for some management, there's a few decent "monitoring" systems out there; personally I just SSH in and check the fuzzer manually (I leave it running in a tmux pane), but if that's not your cup of tea I've heard good things about OneFuzz (https://github.com/microsoft/onefuzz) and LuckyCat (https://github.com/fkie-cad/LuckyCAT). Happy to answer any specifics of the sort :) https://en.wikipedia.org/wiki/Rosetta@home is popular with over 1 million users (more projects at https://en.wikipedia.org/wiki/List_of_distributed_computing_...) This wiki article is great, and is basically the answer to OPs question. I run Boinc connected to Science United which allows me to pick/prioritize domain of interest and it handles of the heavy lifting of getting tasks, running and uploading the results. It has been a install and forget experience with currently a small 3300 days of CPU contributed. This is a great use of extra compute overhead! If you don't know how to get into doing computational research yourself, this is definitely the next best thing Please consider running a Tor Relay, relays are the backbone of the Tor network https://blog.torproject.org/new-guide-running-tor-relay/ Doesn't this have the potential to cause issues for your home network? If you run an exit node, definitely. If not, you should be fine. Just make sure you understand that the IP will be publicly listed even if you run a relay and most likely banned by some organizations. Wasn’t a big deal in 2012, today I would be hesitant to run this at home. One could support I2P network (https://geti2p.net) or distributed P2P web search (https://yacy.net). Lichess has a distributed network for analyzing chess games with Stockfish: https://lichess.org/get-fishnet https://github.com/niklasf/fishnet There's also the Mersenne prime search which passed some milestones recently: https://en.wikipedia.org/wiki/Great_Internet_Mersenne_Prime_... If you want to support indie Blender artists, sheep-it[0] has been around for years and lets you use your free CPU/GPU to render user-submitted 3D projects. You get points for rendering, which you can then use to submit your own jobs for distributed rendering. > When I was a kid, I remember participating in Folding@Home. Unfortunately, the project doesn't appear nearly as active as when I was younger. Hmm, I don't follow it that closely, but I have given a clock cycle or two to the project. And from taking a quick glance at their website[0] it seems they are still pretty involved. The most recent posting is called "Markov State Models (MSMs) of SARS-CoV-2 proteins" (although it may be from June 2020) I'm also curious as to why the disinterest in crypto. I am not hugely swayed one way or the other, so its nice to hear peoples reasoning. The energy use of proof of work crypto is terrible for the environment, that's at least the best argument I hear. Better to use the energy for research imo Modern hardware, unlike hardware back then, idles at much lower power. Just let it sleep/idle The OP says excess home compute, he might as well be running a blog on his server/pc directly without Kubernetes BS, so I don’t think this might be an option since wants the unused resources to be utilised. WorldCommunityGrid sort of replaces folding@home