|
Welcome to the BOINC and grid computing thread! Were you a cool kid back in the dotcom era, running GIMPS or SETI@Home on your work desktop? Did you crunch Folding@Home on your PS3? Well, that scene is still going, and in fact it has expanded substantially. I want to help fight Covid-19! What do? Here are some projects to join:
Rosetta, TN-Grid, and most other distributed projects use a framework called BOINC, so you'll only need to install one piece of software to contribute to them (then you attach your BOINC client to as many projects as you want). F@H does their own thing, and have their own client, so you'll need to install that if you want to contribute to their efforts. This isn't good or bad, it's just the way it is, and setting expectations is always good. And here are some videos on how to set things up: And of course, ask questions in this thread if you need any help. Thanks for volunteering your cycles! What's all this, then? Grid computing is a way to divide up huge problems and/or datasets among many computers for distributed analysis. If you're into science, medical research, and/or pure mathematics, there is almost certainly a grid computing project out there that you can contribute CPU and/or GPU time to. Research teams with hugely time-consuming computational tasks (on the order of milennia of compute time on modern hardware) split these tasks into very large numbers of chunks, called workunits (WUs). A WU is an amount of work which should be computable on reasonable hardware in a reasonable amount of time -- anywhere from half an hour to a couple of days. Volunteers then sign up to help with projects, install a client, and their hardware begins crunching WUs. BOINC is a management framework for distributing and aggregating tasks for grid computing applications, and a terrible acronym standing for Berkeley Online Infrastructure for Network Computing. Most grid projects now use BOINC as their backend and network infrastructure, which is good for users who want to contribute. Instead of having a lot of single-purpose applications installed, you can just install a single BOINC client and attach it to various projects. BOINC also provides a lot of control over how/when/how much of your hardware gets used. You can declare what timeslice each project gets in relation to each other. You can suspend computing between certain hours, or when your machine isn't idle. You can only allow a certain number of active threads, so that you have some CPU reserved for yourself -- and a lot of other controls besides. Once your client(s) are set up, contributing to modern grid computing projects is basically zero effort. Unless (like me) you're the sort of nerd who likes obsessing over how much you're crunching. But why? The people I've talked to, tend to fall into two categories:
Lol SETI is bullshit and nobody's PS3 cured cancer I'm not going to weigh in on the merits of any specifc project. What I will say is that the field of projects has grown, and most of the popular ones currently running are dedicated to tangible scientific and medical research and are led by actual scientists from actual universities. Further, while finding a cure for something would be lovely, that's not actually the focus of most projects because that's (usually) not how science works. Most of the projects oriented toward medicine are taking the approach of identifying good targets for further research. A lot of the utility in the current generation of projects is in winnowing away the possibilities which do not work, leaving researchers with a much more focused set of possibilities for attempts to find things which may accomplish their goal. Right tool for the job, and all that. Finally, acedemic publications do come out of community work. The team behind the Help Stop Tuberculosis subproject at World Community Grid, as an example, recently announced that they will be publishing a set of papers informed by WCG data by the end of the year. This is real science, and that means it's almost always a long march rather than a sprint. Sorry bro saving the world is nice but I gotta mine buttcoins I'm sure you will be absolutely shocked to learn that there is now crossover between grid computing and cryptocurrency. I'm not going to link to them, but the big player here is Gridcoin, who claims to pay out for work done on various BOINC projects. Also -- though I'm not sure why, and I don't believe they're paying out for BOINC work -- a team based around the Jesus-oriented coin Biblepay (because of course that's a real thing) has recently started competing in community challenges on the World Community Grid project. Ok, fine. What can I do with my computer machine to make the world a better place? A lot. Since I'm more intimately familiar with them and can provide more information on them, I'll start with the projects I run.
Here's some other popular projects:
And then there's Wikipedia's list of distributed computing projects, and the BOINC list of projects which includes some more esoteric projects involving things like seismology, underwater acoustics, and cognitive science. Last but not least, there is Folding@Home, which is something of an odd duck. It's a very legit project, but it does not use the BOINC framework. Tell me more about the incremental game aspect Most projects are heavily gamified and track/rank contributors based on:
Several projects also award badges for hitting various milestones of work done. And then there's the meta-stats provided by BOINCstats, which show you how you're doing across all your projects, compared to everyone else who is doing any BOINC work, worldwide, and against yourself via historical charts and tracking of performance over the past 40 days. This poo poo is like digital crack to me. I currently have Goon Teams In addition to your solo achievements, BOINC projects also have teams, which are collections of users whose contributions are counted together for bragging rights. * When you're part of a team, all your points/runtime/etc are counted toward your personal stats, and also toward that team's stats. * You can belong to only one team at a time. * Points earned while on a team stay with that team if you leave; you can't move points from one team to another here are the known Something Awful BOINC teams:
And also the SA Folding@Home team. Tooting my own horn If you find yourself getting addicted to this poo poo and falling down a hole of building machines which are completely dedicated to crunching WUs, then maybe you'll also find yourself wanting a lazier way to keep all those machines up to date and synced in configuration. I've been working on a suite of tools for that very thing, called Homefarm. What now If you're already crunching, hang out and talk. Keep us up to date on news in your projects and talk about your hardware plans. If you're new or interested, ask questions and get terrible opinions! EDITS
mdxi fucked around with this message at 21:30 on Aug 8, 2022 |
#
?
Oct 13, 2018 21:47
|
|
|
|
| # ? Apr 18, 2024 21:43 |
|
|
I've been running BOINC for about a year now, and I've been joining the Something Awful teams for projects such as Asteroids@Home, mainly as a way to show that I have stairs and I'm proud of it. Is there any incentive to not doing that, and joining some other team instead? BOINC can be ran on something as small as a Raspberry Pi, although not every project will support the architecture.
|
|
#
?
Oct 13, 2018 23:14
|
|
|
I started running F@H about a year ago. Any reason to switch or donated compute power can help anyone equally? I'm medically retired from MS and losing my ability to walk, F@H doesn't have a specific MS project, but if there's one out there I'd switch.
|
|
#
?
Oct 13, 2018 23:56
|
|
|
Tuxide posted:I've been running BOINC for about a year now, and I've been joining the Something Awful teams for projects such as Asteroids@Home, mainly as a way to show that I have stairs and I'm proud of it. Is there any incentive to not doing that, and joining some other team instead? Not to my knowledge, no. Teams just let people aggregate their work, for fun. And you don't take your points with you, if you do switch teams. Whichever team you're on gets credit for the work you do while you're part of that team; nothing from before or after. There's even a kind of a meta-team on the WCG forums, who expressly jump around and lend CPU time to other teams as they see fit.
|
|
#
?
Oct 14, 2018 04:45
|
|
|
craig588 posted:I started running F@H about a year ago. Any reason to switch or donated compute power can help anyone equally? I'm medically retired from MS and losing my ability to walk, F@H doesn't have a specific MS project, but if there's one out there I'd switch. From some quick googling, I couldn't find a MS-specific project. I found some posts on the Rosetta forums which talked about the possibility of such a thing, but apparently they've never had a researcher who was focusing on MS. Also noted was that the protein folding/molecular dynamics type projects are what an MS-specific project would likely look like; you'd just need an appropriate dataset and a conjecture to investigate. I did find an MS team on WGC, who have been crunching since 2004 (wow!), and have over half a millennium of CPU time. I also found their forum thread. WCG doesn't actually have any subprojects working on MS though.
|
|
#
?
Oct 14, 2018 05:00
|
|
|
If you have the bandwidth for it https://archiveteam.org/index.php?title=ArchiveTeam_Warrior wants volunteers. It uses virtually no CPU time and about 5GB a day. If you want to download the internet it's light enough to run at the same time as another project as long as you're leaving your PC on.
|
|
#
?
Dec 17, 2018 12:03
|
|
|
I've been running BOINC on my media server for the past 2 years, contributing to various projects that also award Gridcoin. Not that it's very much money at all - quite literally the same amount of money I'd probably have found between couch cushions in the same timespan. Since the machine is on all the time already, and BOINC seems to be smart enough not to impede transcoding operations, it's not been a problem at all - and at least contributes back to the various projects I'm in. WCG, Rosetta@H, Einstein@H, etc.
|
|
#
?
Feb 16, 2019 23:37
|
|
|
After way too much time, because I am terrible at arts and crafts, I have completed my multi-machine chassis. These machines do nothing but crunch WGC workunits. More info and pics here.
|
|
#
?
Feb 18, 2019 08:08
|
|
|
Version 0.16.0 of my Homefarm project is out. This is the first release candidate, which means there will be no work done except bugfixes between here and v1.0.0. The last big addition is a local repository, which compute nodes will use for installation and update of their OS packages. In more purely BOINC news, the 13th BOINC workshop is being held in Chicago, in July.
|
|
#
?
Feb 21, 2019 08:11
|
|
|
This is super neat and I remember folding@home from awhile ago so thanks for bringing grid stuff up again. I appreciate your dedication to these projects.
|
|
#
?
Feb 24, 2019 15:41
|
|
|
I run BOINC purely from a hobbyist aspect; I'm in nobody's top 10 (or even 100) but I like seeing what machines tossed by other people are still capable of doing. I have a few machines running Einstein@Home and hope maybe someday I'll get a pulsar named after me. Once I got an AMD GPU card out of a MicroCenter bargain bin, re-thermaled it, and it's currently the top producer of WUs out of all the machines I run currently.
|
|
#
?
Feb 24, 2019 21:03
|
|
|
Speaking of GPUs, I just got a GTX 1650, and wrote up my Primegrid results here. Now that some benchmarking is done, I'm about to let it start crunching einstein@home WUs. And GPUGrid, once they un-break their Linux client. mdxi fucked around with this message at 05:26 on May 22, 2019 |
|
#
?
May 20, 2019 03:11
|
|
|
mdxi posted:Speaking of GPUs, I just got a GTX 1650, and wrote up my Primegrid results here. This doesn't load for me.
|
|
#
?
May 21, 2019 21:47
|
|
|
Oops. Thank you. Fixed above, and for your convenience: https://firepear.net/grid/gtx_1650_vs_gtx_750_ti/
|
|
#
?
May 22, 2019 05:31
|
|
|
Asteroids@Home has been a bit sporadic with available WUs lately, sometimes going days with no tasks to send out, but it's the only project that I've been contributing to for a couple reasons. First, it's one of the few rare BOINC projects that actually offers WUs for the Raspberry Pi. Now, you're probably thinking the Raspberry Pi isn't that computationally impressive. It isn't. It takes two entire days just to complete a batch of four. My desktop computer takes four hours to do the same thing. What that means though is that when there are zero tasks for the server to send out, the Raspberry Pi is still doing WUs for days until the server has a new batch of tasks to send out. This plays a small part if you're trying to do something like manipulate team standings for fun. The second is to see the Something Awful team climb in ranks, which I just do solely for fun. We don't really have an impressive size, but I just think it's funny that goons can outperform other teams like Microsoft and video game groups that goons would normally be competing against in other areas.
|
|
#
?
May 23, 2019 10:47
|
|
|
Tuxide posted:The second is to see the Something Awful team climb in ranks, which I just do solely for fun. We don't really have an impressive size, but I just think it's funny that goons can outperform other teams like Microsoft and video game groups that goons would normally be competing against in other areas. I didn't know there were any goon teams out there  It's kind of shocking and/or depressing how little effort and computational grunt you need to throw at these projects to climb the rankings. An enormous number of people obviously show up, maybe run the client for a week (or a day) and then disappear forever. I've been into this stuff for close to two years, and have only made it a kind of goal for the past year. I have four machines, all of which are completely ordinary PCs. A year ago I had one. The fourth only got built on 06 Feb of this year. Only two of them have ever had GPUs slotted, and for most of the past two years, none of them have had GPUs. And still my "team" (which is really just me, though two of my machines run under an account made for my partner, because she loves science but doesn't care to be a computer janitor like myself) is #1259 on World Community Grid by WUs returned. We're #3377 globally on BOINCStats. That shouldn't be the case. It's depressing, because so many people throw so much effort into buttcoins of various flavors. Last week, someone brought a 6 GPU mining chassis into my office and were like "LOL this doesn't even pay its electricity bills anybody wanna play with it?" When I mentioned running a GPGPU grid computing project on it the response was "Or we could do some hash cracking!" Sigh.
|
|
#
?
May 23, 2019 18:17
|
|
|
Earlier today, the first 2000 beta WUs for one of WCG's upcoming climate projects were pushed out. One of my nodes got 2 of them. Here's what they look like while running:code:code:It is interesting that while the number in the `fe` (estimated FLOPS) field is very large (13.7 TFLOPS), it's actually small compared to other subprojects (46.2 TFLOPS for a MCM WU; 24 TFLOPS for a Zika WU; 23.8 TFLOPS for a MIP WU). This makes the very long runtime surprising. Possibly this is always way off for new projects? Staffers also said to expect greater-than-normal memory usage with these WUs. You can see that each of these jobs is using about 734MB of resident memory (that is: memory used exclusively by that process). This is, indeed, more than the WUs of other subprojects. Here's a look all WUs running on that machine, which has 16GB RAM, sorted by memory usage: code:quote:The work for this project will be broken into small geographical regions, and in the end each region will be simulated for one calendar year. Each individual work unit represents 48 hours calendar time for this simulation. Once a result has been validated for the 48 hours, the output will be used to build the input for the next 48 hours of runtime. Exciting times!
|
|
#
?
May 30, 2019 19:53
|
|
|
I do this on a raspberry pi. Not good, obviously. So I also signed up for lots of free trial accounts on cloud servers and set up a huge 12 instance system. I kept it running for a year constantly and my BOINC stats say I’ve done more crunching than 98% of all other users. Feels good. I ended up setting up an embarrassingly preachy project website called Freecrunch to publicise the fact that everyone could do this and contribute to the effort of all these science projects but never finished the content so there’s really just a GitHub repo with some easy install scripts for installing things. I also tweaked boinctui a bit if anyone uses that: https://github.com/mpentler/boinctui-extended
|
|
#
?
Jun 29, 2019 17:56
|
|
|
thehustler posted:I also tweaked boinctui a bit if anyone uses that: https://github.com/mpentler/boinctui-extended You're the updated boinctui person? That's rad! Thank you for your work.
|
|
#
?
Jul 4, 2019 04:06
|
|
|
mdxi posted:You're the updated boinctui person? That's rad! You’re welcome! The original Russian guy seems to have gone MIA so there was a lot of PRs to take. Edit: anyone who knows C++ better than me want to look at the codebase and see if there’s any huge issues with it? thehustler fucked around with this message at 23:54 on Jul 7, 2019 |
|
#
?
Jul 7, 2019 23:51
|
|
|
I was really into WCG for a while, devoted multiple machines full time to it, and there was a WCG specific thread on the forums about the team(that got archived years ago). The "Something Awful Goons" team still technically exists and is currently ranked #121, with only 22 active members anymore: https://www.boincstats.com/stats/15/team/detail/11871 Best team position was once #73, oh how the mighty have fallen... Anyways, lately I've been wasting all my CPU/GPU cycles finding out which numbers aren't prime for GIMPS, but no goon team there afaik: https://mersenne.org So yeah any of you WCG heads should join that and see if you can top the #3 team contributor who basically hasn't submitted a result in 2 years  https://www.boincstats.com/stats/15/user/list/0/0/11871/0 Also, thehustler I would still be interested in a cloud computing guide if you have the energy to pick that up again. I'm kinda confused about all the different options on AWS etc, and would like to get the most bang for my 0 bucks.
|
|
#
?
Aug 31, 2019 01:27
|
|
|
After slacking for some months, I'm working on Homefarm again: v2 is coming in the not-too-distant future. The big change will be the removal of the requirement for the control node to be a Raspberry Pi. In fact, there won't be a requirement for a physical control node at all; those functions are being moved into a docker container. This uses a disgusting amount of drive space (because it's based on Arch, for consistency with the compute nodes), but removes one whole physical computer from the setup. I'm also working on containerizing a compute node, but that's intended to be only for testing purposes -- right now I can't test a lot of scripts without either doing an actual reinstall on my farm, or having spare nodes lying around for testing. I'm also also streamlining a bunch of stuff, making QoL improvements, improving the docs, etc., etc. Of course, I think I'm the only one actually using it, but I'm still excited about making things better.
|
|
#
?
Sep 3, 2019 06:06
|
|
|
World Community Grid has launched a new project, the Africa Rainfall Project. Its aim is to produce better, localized weather models for sub-Saharan Africa, which will increase crop yields (and be generally good for everyone). Unfortunately, it's off to a bit of a rough start. It turns out that climate modelling is not hugely parallelizable in the way that some other problems are. You can split the problem up into localized forecast areas, and run each of those in parallel, but you can't parallelize along the time axis; the results of the first chunk of the simulation for each forecast area becomes the initial conditions for the next time-chunk for that area. WCG staff anticipated this being problematic, and set the "WUs per machine" limit to 1 for this project, rather than the usual "unlimited". They also asked people to leave it set at 1. Predictably, people did not, and pretty much the entire initial allocation of WUs was gone by the time most of the US woke up. In three days, with 6 machines crunching, I've seen one ARP1 WU, and some people still haven't seen any at all. WGC staff said that the average completion time for an ARP1 WU, during the beta test period, was 20 hours. So this process is slow on top of everything else. Wingman WUs are slow; reworks are slow; it's all slow. It's frustrating. Some people are also complaining about the dataset size. Each WU is a 100MB download and a 60MB upload. I think those people are just being whiny though, because 1) I was routinely doing Debian installs and updates over a 56kbps modem back in the day, which was a lot more data 2) Deselect that project then dude. No one is forcing you to participate in it, and the other subprojects haven't gone anywhere. ...Except the ones that have. Smash Childhood Cancer is on a hiatus while the most recent batch of returned data is analyzed and the researchers decide on their next targets. OpenZika is very, very close to being completely finished. Until 27 October, there were about 300k ZIKA WUs being returned per day. The six days since then have seen 166k, 103k, 82k, 71k, 56k, and 40k.
|
|
#
?
Nov 3, 2019 05:49
|
|
|
On Asteroids@Home, Team Something Awful isn't that big and we're nowhere near the top ten, but for some odd reason I do find a bit of pride in our recent credit score: I don't know who is actually in charge of these goon teams, looks like it's mainly inactive goons...but perhaps they should be linked to the OP, so that goons actually know that they're there in case they want to join them.
|
|
#
?
Nov 3, 2019 10:18
|
|
|
Tuxide posted:On Asteroids@Home, Team Something Awful isn't that big and we're nowhere near the top ten, but for some odd reason I do find a bit of pride in our recent credit score: Added! If you know of any others, let me know and I'll be happy to update the OP.
|
|
#
?
Nov 3, 2019 19:47
|
|
|
Tuxide posted:On Asteroids@Home, Team Something Awful isn't that big and we're nowhere near the top ten, but for some odd reason I do find a bit of pride in our recent credit score: I'll put some time into mapping asteroids and pulling further ahead of 4chan.
|
|
#
?
Nov 4, 2019 01:38
|
|
|
I just made myself captain of the Something Awful team on World Community Grid by accident. That's a little awkward.
|
|
#
?
Nov 5, 2019 10:44
|
|
|
Hello, New Captain! I just joined Asteroids and the goon team as well, because yay science and gently caress 4chan. No workunits so far. I can't actually tell from the very sparse homepage: is this a CPU project, or a GPU project?
|
|
#
?
Nov 5, 2019 21:44
|
|
|
mdxi posted:I just joined Asteroids and the goon team as well, because yay science and gently caress 4chan. No workunits so far. I can't actually tell from the very sparse homepage: is this a CPU project, or a GPU project? Asteroids@Home workunits are pretty sporadic. One day, you'll see a batch of a million WUs ready to send, and then four days later they're all gone. Then you have to wait a few days for the next batch to appear. From what I can tell, Asteroids@Home claims to support the following:
The last bullet is worth pointing out because BOINC projects that support ARM Linux are rare, and in case anyone has a dozen Raspberry Pis they want to throw into this thing. You're not going to churn out WUs with them, but it does have an effect on the recent average credit score during times when everyone else is waiting for the next WU batch.
|
|
#
?
Nov 5, 2019 22:24
|
|
|
mdxi posted:If you know of any others, let me know and I'll be happy to update the OP. These are the goon teams I'm aware of. Note that most of these, but not all of them, are called "Something Awful". I find them by looking for team names with the word "awful" in them, though sometimes this returns non-teams registered by spambots.
|
|
#
?
Nov 6, 2019 05:20
|
|
|
Tuxide posted:These are the goon teams I'm aware of. Note that most of these, but not all of them, are called "Something Awful". I find them by looking for team names with the word "awful" in them, though sometimes this returns non-teams registered by spambots. Updated. Thank you. In unrelated news, I've started posting monthly subproject summaries from the WCG forums to r/boinc. Here's the most recent one: https://www.reddit.com/r/BOINC/comments/dwc7fg/wcg_project_updates_november_2019/
|
|
#
?
Nov 18, 2019 21:02
|
|
|
peepsalot posted:Also, thehustler I would still be interested in a cloud computing guide if you have the energy to pick that up again. I'm kinda confused about all the different options on AWS etc, and would like to get the most bang for my 0 bucks. Hello. Sorry, I didn't see this for so long. I could maybe write something this weekend.
|
|
#
?
Nov 27, 2019 12:05
|
|
|
Tuxide posted:The last bullet is worth pointing out because BOINC projects that support ARM Linux are rare, and in case anyone has a dozen Raspberry Pis they want to throw into this thing. You're not going to churn out WUs with them, but it does have an effect on the recent average credit score during times when everyone else is waiting for the next WU batch. My RPI3B did fairly okay and provided a nice boost to my other 10 hosts in the cloud, although you really need to figure out which projects work well/fast on it and only attach those projects.
|
|
#
?
Nov 27, 2019 12:06
|
|
|
I signed the Something Awful team at World Community Grid up for a competition to see which team can earn the most points between today and December 2nd, so turn your machines on if they're idle goons!
|
|
#
?
Nov 30, 2019 23:57
|
|
|
Cool article on Rosetta@Home: http://www.dailyuw.com/wellness/article_6e89b56c-77a1-11e9-a9bb-fbd11958b0cf.html The following isn't BOINC stuff, but I figure there's a good chance that anyone here is also interested in general scientific computing news.
|
|
#
?
Dec 11, 2019 20:54
|
|
|
OhFunny posted:I signed the Something Awful team at World Community Grid up for a competition to see which team can earn the most points between today and December 2nd, so turn your machines on if they're idle goons! I contributed some WUs to this during the time period, did we win anything? In other news, in Asteroids@Home Team Something Awful has broken the Top 100 in recent average credit for the first time since I've been tracking it. This score goes up and down quite dynamically depending on availability of WUs and how many goons we have dedicated to the project, but this is the first time I've seen us break 100--we're #99 now. Our total credit score has been going up too, we're now #361 in rank. This is an awfully big deal to me because I don't want to see space rocks with dumb names like (9001) Derpyhooves going around our sun. Asteroids@Home actually submitted 762 new asteroid models to DAMIT back in September based on WUs between then and last year (I think), so they actually do produce results.
|
|
#
?
Dec 14, 2019 09:16
|
|
|
Tuxide posted:I contributed some WUs to this during the time period, did we win anything? We came in 8th place with 1 million points earned. Which I thought was a rather respectable showing.
|
|
#
?
Dec 14, 2019 09:44
|
|
|
WCG's OpenZika subproject has finished compute work: https://www.worldcommunitygrid.org/about_us/viewNewsArticle.do?articleId=614
|
|
#
?
Dec 15, 2019 07:09
|
|
|
Tuxide posted:In other news, in Asteroids@Home Team Something Awful has broken the Top 100 in recent average credit for the first time since I've been tracking it. This score goes up and down quite dynamically depending on availability of WUs and how many goons we have dedicated to the project, but this is the first time I've seen us break 100--we're #99 now. Our total credit score has been going up too, we're now #361 in rank. This is an awfully big deal to me because I don't want to see space rocks with dumb names like (9001) Derpyhooves going around our sun. I just started up this week and joined the team so I'm glad to see it's making a difference! I have a question though, I've been trying running the BOINC client in several different OS to get the best bang for my buck here. It seems like running the client in a VM running Ubuntu is better than just in a BSD jail which is kind of backwards to what I was originally thinking. What is the best OS and does the BOINC client version also matter?
|
|
#
?
Dec 15, 2019 19:03
|
|
|
|
| # ? Apr 18, 2024 21:43 |
|
|
Cracked 1,000 results returned on WCG. 
|
|
#
?
Dec 16, 2019 21:55
|
|