BOINC porting

[MtM]

I did read your first post. You don't seem to understand my points.

Yes, boinc has cuda support now. Yes it offers smp support now. But do you think their recently added support is on par with the cores/clients which are being used and developed by f@h / PG considerably longer and with more support? F@h userbase is pushing developments, and the dev team and the donors are in constant dialog to improve faster and further. I don't think that boinc will add much to f@h right now, it would only hinder further development IMHO.

Got me on the last point, I did do boinc for awhile but concentrated on hcc. I know the rice project ect and while I don't think any boinc project is bad, I'm here to find cures for deseases, not to research new variaties of rice to feed the third world.

[cenit]

ok, ok... it seems that my words do not pass on you.

Why do you think PandeGroup isn't sharing their ability with D.C.? Don't you think that if they were more "open" they would had helped boinc development (back in the days) so that now every project could benefit? More or less (apart from small experiments) every other distributed computing project runs on BOINC. Boinc it's not World Community Grid (which you wrongly think is Rice project), Boinc is not Einstein@Home, Boinc is not Quake Catcher (qcn, a STANFORD UNIVERSITY PROJECT), Boinc is not Rosetta@Home (a project dedicated to protein structure, not folding), Boinc is not SIMAP (a database of protein similarity really used by many many scientists), BOINC is not gpugrid (a folding@home-like project from a spanish university, if i'm not wrong), BOINC IS A PLATFORM TO ENABLE AND EASE DISTRIBUTED COMPUTING.

Final point: areyou sure you are here to find cure for deseases? I appreciate the folding@home project, but I think that we're really far away from it...

[MtM]

Berkely open infrastructure tells me enough about what boinc is thank you. Please quit assuming you know what I think, or what I know or do not know

The main reason PG isn't releasing the source for the clients has been given already: to ensure integrity of the submitted results. The core's are built up on open source software, and I want to bet you that the gromacs team ( http://www.gromacs.org ) does benefit from what PG does internally. So no, they don't directly aid boinc projects, but any boinc project using a gromacs based core is already benefiting from f@h's advances.

So, you know the reason the source code isn't open, you know how any other project is already benefiting from f@h, what was your last question?

If I'm here to cure deseases? Yes I am. I'm here because I think the process of proteine folding holds the key to allot of things which can go wrong in the human body. I don't expect f@h to lead to a cure directly though, maybe that's your point. f@h isn't doing specific research to find a drug for desease x, their doing the work which needs to be done before such a drug can be developed and that's a much bigger task.

Your question makes me wonder what you think we're doing here? Developing a new kind of nuclear weapon for the CIA?

edit: spelling

[bruce]

@cenit:

FAH does have requirements that are not adequately addressed by BOINC (such WU duplication/verification and a priority scheme intended to address the serial nature of FAH assignments as opposed to the parallel nature of most BOINC projects.) I do not believe that the changes needed by FAH should be addressed by a future version of BOINC.

Many of us have spent years contributing to BOINC and we're very familiar with what BOINC can do and what many of the projects that use BOINC are doing. Apparently you are an excellent evangelist for BOINC, but you're not going to be successful in converting FAH enthusiasts to the concept that BOINC is better than FAH. Similarly we will not able to convince you that it is neither better nor worse, only different, and with different objectives, and, in fact, incompatible.

Let's agree to disagree and drop it.

[cenit]

Don't worry bruce, I'm not trying to convert anyone. In fact, maybe I wouldn't be able, looking at this thread... Anyway I appreciate your answer, MtM and I seem unable to speak to each other
I appreciate the fact that you have considered BOINC, even if I cannot understand why you think that boinc shouldn't change to ease an eventual folding transition to that platform. It had received so many changes... but ok, let's stay on the way we are on now, maybe it's the best configuration.
And, at the end, I want to make sure that you can consider me open-minded. I'm not absolutely against something or favour something else, it's only for the fact that for me it would be really a better solution to have folding on boinc and I think I'm not alone.

As you have suggested, let's drop it.

[butc8]

Ive crunched a few projects on BOINC, but always seem to come back to Folding for two reasons.

First of all it does not have a quorum....not all BOINC projects have but most do.(IIRC wcg has a quorum of 15!!!)
I hate the idea of replicating work someone else has already done while here at FAH its not necessary.

Secondly I have not yet seen a single project publish half the papers FAH has.(I think climate prediction has the most, but im not 100% sure)

[Rattledagger]

Ive crunched a few projects on BOINC, but always seem to come back to Folding for two reasons.

First of all it does not have a quorum....not all BOINC projects have but most do.(IIRC wcg has a quorum of 15!!!)
I hate the idea of replicating work someone else has already done while here at FAH its not necessary.

To correct a very easy to make mis-understanding with WCG-quotas, while it looks like some of the sub-projects has very high replication, in reality each result for these sub-projects is unique. Each individual result has atleast one random parameter, and due to this randomization all results for wu is unique, even it looks like 19 copies of the same wu is sent-out. AFAIK FAH does something similar, but instead of listing it under the same wu, it's listed as different "Runs" (or is it Clones?) for the same "project".

Look on the WCG-FAQs for more info about the unusual quotas, a good starting-point would be "BOINC: Results Status page - What does xyz status mean? "


Generally for BOINC-projects, they'll use no replication if possible, but some projects uses replication 2 since there's no way for them to just look on the result and example say "oh, that C-H-bond is too short/long or in impossible angle to another C, it must be a bogus result". AFAIK the only project that uses 3 for quorum is LHC@home, but it's possible some alpha/beta-projects also uses high quorum.

Secondly I have not yet seen a single project publish half the papers FAH has.(I think climate prediction has the most, but im not 100% sure)

AFAIK Baker laboratory (Rosetta@home) has published the most papers.

[7im]

...
AFAIK Baker laboratory (Rosetta@home) has published the most papers.

Rosetta@home didn't launch until October 6, 2005 (5 years after fah). How many papers are actually from Rosetta, and how many are from Baker Labs, which started much ealier (1995?) ?!!!

If you want, we can start counting up the papers Vijay Pande published before starting F@h, you know, just to be fair and compare apples to apples.

Or just compare project related papers, and note that R@h is about 20 papers behind F@h.

[cenit]

I wanted do drop this thread, but now I cannot NOT re-enter the discussion.
Please stop comparing pubblications from numbers, most of all with rosetta

Did you read every paper published by Pande Group? How many of those really needed F@H, to be written?
On the other hand, Rosetta is a FREE software for everybody except commercial use. Don't you think that, even without taking into consideration these papers, many other from external people did use it?

Please stop this discussion about papers published, it's not fair to list everything, we should see the real value of the publications.

[Rattledagger]

My mistake, I forgot to include "of the BOINC-projects", but regardless of included or not, any comparison based on #papers wouldn't really mean anything, since it's basically on the same level as kids in kindergarden with "my daddy is..."

[cenit]

any comparison based on #papers wouldn't really mean anything, since it's basically on the same level as kids in kindergarden with "my daddy is..."

agreed

[bruce]

I don't think this discussion is productive so I'm going to close it.

Everyone should be allowed to choose what projects they choose to support using whatever criteria they choose to use. We don't have to agree.