Grr. (BOINC irritations)

[altivo]

Over the weekend I seem to have crossed the 200,000 credit point mark. However, this is without any particular thanks to World Community Grid, whose work units are becoming unprocessable on all four machines I have running them. They sit in the queue, supposedly "running" but without accumulating any CPU time or showing any change in the hours to completion. This can continue for days on end, locking out all other projects trying to share time on that machine. Or they run up CPU time endlessly, but again without showing any progress toward completion. In a few cases, they ran for two to three days, then errored out with bad results. It seems evident to me that inadequate testing and planning has taken over in the management of the Fight AIDS @ Home project.

The attitude expressed by the project managers so far is just "It's OK, they just run longer." Maybe that's true on Windows, but on Linux they don't run at all. I'm particularly aggravated by the cavalier attitude about taking up all the time and locking out other projects. In another day or so I'll be ready to kick WCG units off the remaining four machines and lock them out. They are already self-selected out of three of my machines because they didn't bother to support the Alpha and because, for no rational reason, their code insists that one machine doesn't have enough RAM, even though it has the same amount as another that is running their stuff (or has been until now.)

Comments

[soanos.livejournal.com]

Sounds like they are underestimating the contributors, and only making the software supported on one platform. I hope a Linux client emerges soon :)
And that the project managers will see the light.

Addendum.

[soanos.livejournal.com]

Forgot to add, every computing second counts for the HIV patients around the world, desperately waiting for a cure.

I am sure the project managers would be urging to fix the bugs and ensure performance in all conditions, had they been tested positive. To me, it smells like the typical "I am not affected, so no hurry"-attitude.

Re: Addendum.

[altivo.livejournal.com]

Or something, yes. I've already been concerned that any useful information derived from these projects will simply end up in the hands of drug companies who will use it to produce patented and trademarked products for which they will charge exhorbitant and impossible prices.

[soanos.livejournal.com]

That would really undermine their respectability.

I just wonder... Perhaps I should start, too.

[altivo.livejournal.com]

No harm in trying it. You can always drop out if you find it causes trouble. There are plenty of projects to choose and not all have commercial aspects. Most of the astronomical research, for instance, is pure science.

[heavens-steed.livejournal.com]

I don't get the whole BOINC thing. I don't have a clue how it works or what exactly it does.

[soanos.livejournal.com]

Not sure about BOINC itself, but distributed processing is a good thing in my opinion.

If I have understood right, it is a distributed processing project, where users can donate their CPU time for number crunching, like trying to find a cure for cancer and AIDS and other illnesses while they do less CPU-intensive work like typing or listening to music, or browsing the web. Basically, client downloads a "work unit", which is a chunk of data to be processed on client's PC, complexity of which is determined by the hardware specs of the user. When the chunk is processed, whuch usually takes from a few hours to a few days, the result is sent back to the server. The server keeps track of the processed data and combines it, and gives the client a new bit of code to chew on.

This method allows researchers to use volunteer CPU power in addition to their expensive supercomputers for their important work.

[altivo.livejournal.com]

In simple terms...

You know about supercomputers that work on very large modeling processes, like astronomical models, weather analysis, or gene mapping, right? They do massive amounts of numeric calculation at high speed, and still require days or weeks to arrive at a solution. Such machines are very expensive to set up and operate, but certain kinds of scientific work require them.

BOINC is a system that allows those calculations to be done on ordinary desktop microcomputers by breaking the task down into smaller parts. Volunteers who run BOINC on their machines can choose projects they wish to support, and "donate" their spare CPU time. The BOINC software runs in the background, like a screen saver or a virus protection program, and only uses the computer when it isn't busy (which is actually most of the time.) Work units are downloaded over the internet from the chosen project, processed, and the results are sent back to the project managers to be reassembled into useful data.

In theory, this all happens automatically without any meddling required on the part of the home or office user, and in many cases that is true. However, if you volunteer time for new projects in the testing stages, or projects that are not always well managed, then you can have various snags in the processing, which is what I've been experiencing. I expect that on the two newer astronomy projects I've been supporting, but I don't think it should happen on old, established projects with well tested code, hence my complaint.

There are about a hundred widely varying projects that can be selected. Some are pure math, like prime number research or searching for proofs of theorems, others are applied sciences like astronomical research or game theory, and quite a few are humanitarian work, such as medical research on protein models, new drug compounds, or vaccines.

I have BOINC running on seven computers (Rex has more) and contribute time to half a dozen different projects.

[shadow-stallion.livejournal.com]

I have to wonder...do you really know that any of this is doing any good? I have seen you and Rex speak of this stuff and in some instances I have to wonder if it isn't just a huge waste of your own electricity.

[soanos.livejournal.com]

Well, depends how much electricity your PC is consuming. My laptop has got a 65 watt power supply, which means it cannot draw any more than that at any time. :P

[shadow-stallion.livejournal.com]

True that. I was not speaking of just one machine though and my comment was directed moreso towards Rex. I think at one time I remember reading that he had over a dozen machines at his place doing this sort of work. Personally I see that as a waste of resources due to the heat generated by said PC's, the cost to cool said room(s) they are in as well as the power to run them day in and day out.

I mean I get that this is supposedly for a good cause but the cynical and zaded unicorn in me says...really? Is this really doing any good?

[altivo.livejournal.com]

Note that Rex reduces the number of machines in summer. Last winter he was actually using them to heat a cold apartment. ;p

Do we ever know whether any contribution we make to a charity does any good? Sometimes yes, but mostly no. There are other side benefits in many cases, though.

BOINC has a scoring system that lets participants accumulate "credits" for the units of computing work that they have done. There are teams that compete and rankings on work done, similar to bowling or soccer leagues or whatever. Some people derive considerable personal satisfaction from this sort of competition, so they are getting a benefit that way.

Each project also has an educational side that lets you take advantage of an opportunity to learn a lot more about the research and the results if you choose to do so. And there are social facilities such as mailing lists, forums/BBSes, and chat rooms as well, which interests a fair number of folks.

Supporting a project in the testing phase is a technological challenge in some cases, and that can interest certain people too.

On the whole, I'd say participating in BOINC costs less than writing a check to a charity, and offers more back to the participant, if he/she is so inclined, than just a glossy newsletter and another request for more money four or five times a year. ;p

[altivo.livejournal.com]

In my case, the computers are running anyway, so I'm not wasting any more electricity whether I do it or not. That's the theory behind the whole thing. CPU cycles are usually wasted. Very few microcomputers run at even half their capacity. Why not take advantage of that? Even when I am using my machine, I'm not actually absorbing more than 40% of what it can do, as a rule.

As for whether it does any good, well, that depends on the project. I'm in favor of theoretical research in astronomy, climate change, medicine and genetics. It isn't costing me anything to contribute computing power, really.

[soanos.livejournal.com]

I favor medicine, but then again, I am biased. I want this cure for AIDS out of the way so they can start concentrating on my condition. But I know it will be long time away for me. :P

[soanos.livejournal.com]

Do you know if anyone of them is searching for a cure for Crohn's Disease?
If not, I will start participating in (Fight)AIDS@home. :)

(Edit: Typo)

[altivo.livejournal.com]

You can support more than one project at the same time. You can even set the percentage of your free CPU time that will be given to each project.

I'm not aware of a specific project on Crohn's. However, there are several that work on generic protein folding and their findings can be applied to a wide variety of diseases and disorders, either to find treatments or prevention. Rosetta@Home and SIMAP are two such projects. The specific FightAIDS project is part of WorldCommunityGrid which sponsors half a dozen humanitarian goals, including vaccine research for Dengue Fever, and a project on rice genetics aimed at reducing famine.

[soanos.livejournal.com]

Darn. I need more computers! :D
Anyway, working to Conquer Cancer and cure AIDS.
I feel like I am a geek superhero fighting to save
the world now. :)

[marzolan.livejournal.com]

I have a few units freeze before on my Linux box, but a restart of the BOINC daemon fixed that easily enough every time... They'd restart from the last checkpoint and continue on without further trouble. That only happens on my old machine though, heaven't noticed such things with a friends newer linux computer or with Mac or Windows.

[altivo.livejournal.com]

It does depend on the particular project. SETI@Home never causes trouble. FightAIDS@Home was equally reliable until a couple of weeks ago. SIMAP also runs smoothly, though they've had less work recently and are often idle for weeks at a time.

Cosmology and MilkyWay are two newer astronomical projects I've been running and they've had all sorts of problems with failing units that loop forever or do other undesirable things. It does seem to be true that most projects test more thoroughly on Windows and tend to give short shrift to the other platforms. Statistically, though, fully half of the machines running BOINC are running some kind of Linux environment.

[saythename.livejournal.com]

This is why I don't burn up cycles constantly now with these things.
As admirable as they are, I think they could be managed better.

[altivo.livejournal.com]

Some are better than others. SETI has been running smoothly for years. Rosetta hasn't given me any trouble yet. SIMAP runs well when they have work, but the volume of data downloaded strains my dialup connection at home.

Cosmology and MilkyWay are beta test projects, so you should expect a few glitches and I do. Generally the code runs well enough, but the flow of work is uneven.

WCG, though, has been irritating me for a while. They operate several projects, and even though they ask you which ones you are willing to work on, they have a tendency to dump work for the others on you if you aren't watching them. Their staff seem to have a very casual attitude at times, one that reminds me of the cockroach commander in the old "Fat Freddy's Cat" comics back in the 60s: "Never mind the casualties, there's plenty more where they came from." So they just seem to figure that if they offend volunteers and lose a few, so what? There will be more along soon enough.

[soanos.livejournal.com]

The attitude of staff doesn't sound like community spirit to me. They cannot imagine that if the old volunteers stay, they get even more processing power. Anyway. working on my units, hoping to finish my first unit today.

[altivo.livejournal.com]

Joined a team yet? We have a furry team for WCG. It's called Us Furries (doesn't have to mean "US" but rather "We") and we'd be glad to have you. We have our own community on LJ, too: us_furries

[soanos.livejournal.com]

Okay, I will join you shortly.

Oh, one question. Will I accumulate points for myself, too? :)

[altivo.livejournal.com]

Yes, you keep all your own points, but the same number are credited to the team as well. You can change teams any time you choose. The old team will get to count the points you made while a member, and the new team counts any new points you make after switching. :D

[soanos.livejournal.com]

Okay. I am in the team. I was just asking because I like to keep track of my own personal progress, too :)

[altivo.livejournal.com]

You might want to sign up at a statistical site as well. I use BoincStats. It will accumulate your credits and give a daily update with various graphical views of progress.

You can also use BoincStats to control your machines and issue commands to them, but that's more useful if you have many machines running.

[soanos.livejournal.com]

Well, I probably have no use for the site yet, but when I get more processing muscle on/under my desk I will probably do that.

[soanos.livejournal.com]

Made an account there, although it will mostly just sit there for now. :P