Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
World Community Grid Forums
Category: Community Forum: Chat Room Thread: Supporting more than one project - Pros and Cons |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 14
|
Author |
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Currently I only crunch World Community Grid.
I would be interest in hearing opinions Pros and Cons (from single crunchers and multiple crunchers), for crunching and/or not crunching multiple projects eg seti@home, rosetta@home, climate prediction etc. There are a lot of projects to be crunched and I am trying to decide if I should stick with only the one, or add more projects to my crunching. I only have the one computer so either way my impact is limited, so I am leaning to sticking only to WCG, but I would still be interested in hearing opinions. Thank You dseto |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Don't take my post to serious and I truly don't want to say bad things about the other project but this is my opinion.
climate prediction: I think it's almost completely wasting your power because those predictions really aren't correct. Those 'green' people say the sea level will rise 10 centimeters a year and half a year later they say it's only 3 centimeters. The sea level changed a lot since the existence of our planet but never before in history it is measured so much. It's the some with the ozone hole in our atmosphere. They began to measure and they saw the hole was growing and now they so it's shrinking above antartica, so again they were wrong. 10 years ago there were plans to heat up the earth because they thought there was the beginning of a new ICE AGE! One of the american presidents had a talk about that in Rusland but now it's all forgotten. I'm not saying we're not destroying the earth, I think we are. But measuring the earth for 50 years doesn't give a good image of how to save our planet. You can better all use your bicycle when you go to work. I personally almost never use the car. My school is 15 kilometers away but I'm faster by bike then by bus or by car. And all the other places I go to are less then 4 kilometres away. And here you may not drink alcohol when you drive a car so I still not get it why every one is complaining about high oil prizes, just go by bike. And the most easiest way to save the earth is a birth restriction on the complete planet, but that's almost the hardest solution. Rossetta: as I heard they make new software for folding molecules. They made the software for hpf on WCG. By crunching they can make better software. Seti: It would really remarkable if we would find aliens. But the downside is that they are light years away from us. And like a Dutch comedian said: "there are only two possibilities if you find aliens: 1., they are smarter than us 2. they are dumber than us. If they are smarter, they will find us before we will find them. If they are dumber I don't want to meet them" WCG: It is a really good project but the only downside is I think the program source is not free. If it was you could really see if you are really crunching for humanity and you could compile the workunit yourselves what would speed up about 10 %. |
||
|
mike047
Senior Cruncher Joined: Aug 22, 2006 Post Count: 262 Status: Offline Project Badges: |
I am a small farmer[46 boxes] and participate in several projects. WCG is my main work.
----------------------------------------If I had only one box, I would crunch only one project. For science, WCG is my top choice, if you want points...there are others that "pay" better
mike
Crunch Hard, Crunch Often |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
WCG: It is a really good project but the only downside is I think the program source is not free. If it was you could really see if you are really crunching for humanity and you could compile the workunit yourselves what would speed up about 10 %. It can't be, impossible. The complexity and dependencies of the algorithms working in the right order with the right timings do not permit individual science application compilation at the present time. Several projects do use open source science like Autodock on FA@H and upcoming Dengue, but as it is, if 3 work units were processed by different compilations, they would not come back as equal.... no way to know which one is the right one. That's why WCG uses Homogeneous Redundancy distribution for the 4 different operating systems serviced, as also each OS is of impact on the computation. The work units are just data sets, separate from the science app. They don't need to be 'compiled'. Anyone can open them up to see what's inside. Opposed to molecular, atomic and energy level computation, I'm reasonably sure that a project like ABC@home could do with self compilations. The output is simply a set of 3 numbers (abc-triples). sorry
WCG Global & Research > Make Proposal Help: Start Here!
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jun 24, 2007 4:20:52 PM] |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Further to what Sekerob said: the techs have explained in the past that the biggest difficulty with getting consistent (identical) results is compiler optimisation. To a compiler expert, this makes perfect sense. Floating point operations are very sensitive - tiny changes in the input can change the output a lot, and changing the order of operations can alter the precision in unexpected ways. To the rest of us, this all sounds baffling and counter-intuitive.
But WCG will make the source available for projects using open source code. Well, they have to - that's how open source works! But so far, nobody has shown an interest in providing useful optimisations that WCG can incorporate into the main codebase. We're moving into new territory. Personally, I expect that it will bring a whole raft of problems with it - but the BOINC system is resilient, and when members start compiling stuff and trying to optimise it, and failing - BOINC will cut them off. This will provide a strong incentive for any optimisers to make sure they maintain bitwise compatibility in their results. Oh, and dseto: your original question. Think carefully about what you want to achieve by crunching. If you want to go alien hunting, that's your choice. Equally, you may have other priorities. Technically speaking, you can more or less do what you want - one project or a dozen. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello dseto,
Here is a list of public distributed computer projects: http://distributedcomputing.info/projects.html Here is a list of BOINC projects: http://boinc.berkeley.edu/projects.php If you want to run several BOINC projects, think about your profile settings for Leave in Memory (best turned off for BOINC 5.8 and higher) and the time to spend per project (60 minutes by default, but I think that is a very small slice of time). Also think about the weight allocated to each project (equal weights is easiest to remember but a very small weight allows you to select an emergency project that normally will only be crunched if your other projects go offline for longer than your cache buffer size can cover). Happy crunching Lawrence |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
And one last consideration: UC Berkeley make no representations or guarantees about any BOINC project. Remember that the project will be running their code on your computer, and it could potentially be absolutely anything. All the projects listed by Lawrence are reputable, and won't damage your computer on purpose - but they don't all have the resources of WCG.
All WCG science applications have been audited by IBM security analysts, delivered over a secure connection, and currently represent the safest way of grid computing. So far there are no malicious BOINC projects, but to me it seems an inevitability. BOINC has already been packaged in a trojan and misused that way. Basically, I'm saying: research the projects before you attach to them. Crunch safe. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello dseto, Here is a list of public distributed computer projects: http://distributedcomputing.info/projects.html Here is a list of BOINC projects: http://boinc.berkeley.edu/projects.php If you want to run several BOINC projects, think about your profile settings for Leave in Memory (best turned off for BOINC 5.8 and higher) and the time to spend per project (60 minutes by default, but I think that is a very small slice of time). Also think about the weight allocated to each project (equal weights is easiest to remember but a very small weight allows you to select an emergency project that normally will only be crunched if your other projects go offline for longer than your cache buffer size can cover). Happy crunching Lawrence Why would it be best to Leave in Memory turned off? |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Only because it uses up your memory. The scheduler in 5.8 and above tries to unload tasks after they have checkpointed, so you lose a minimum of work.
Before 5.8, removing tasks from memory had a heavy cost in lost work, and if the task didn't manage to checkpoint before it got removed, it would make no progress at all. |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
In 5.8 there may have been 1 or 2 conditions left to keep LIM on. Did not pay attention during benchmarks while testing 5.10, but if they now keep the job in memory automatically (?), LIM is more or less outdated. For sure it did not do that in 5.4 and it was the only reason i left it on in 5.8.... anyone?
----------------------------------------Benefit is, that if you do scheduled project switching, the job is in memory, whereas if LIM is off, each time you do a project switch, it has to fetch it from disk.... lost CPU cycles, with the said price of hogging RAM. It's main function was keeping a job loaded, but from 5.8, the project switching is done at checkpoint saves. NOW HERE'S SOMETHING TO KNOW: If you are writing a long post or been thinking and you want to see if someone posted in meantime making your post a possible repetition, hit the post preview button and scroll down. You will see any additional posts, like now seeing that Didactylos already cut the grass before my feet :D
WCG Global & Research > Make Proposal Help: Start Here!
Please help to make the Forums an enjoyable experience for All! |
||
|
|