Since we have a few strike team (or new name TBD) members heading off over the weekend, I thought I'd get a strategy discussion started now. That way, they can get an early start if we've got enough figured out by then. In general I'd like to get started this weekend and finish before March Madness folding. It might be nice to make a top 50 DC Vault standing our goal.
The general concept is to make a concerted attack on our DC Vault standings by spreading our members across the projects where we have a low score. In a nutshell, the DC Vault score for a project is your team's percentile rank times 100. Thus, our folding@home team, at the 99.999th percentile, gets a score of 10,000. If we have a team around 75th place out of 100, that's worth 2500 points. What I noticed in several projects is that the lower half or even more of teams have pretty low scores; from there on up it gets a lot harder. As an example, I was able to get our SIMAP team up to 7500 points running 2 systems for 2 weeks.
So, here's a list of projects where we are in the lowest 75 percent. The ones in bold look like good targets; strikethrough ones may have "issues"; "???" means I don't know enough. I would appreciate feedback from experts on the various projects, and also volunteers to lead the assault on each project (meaning help people out with setup instructions etc). I'll try to keep this thread updated with everyone's feedback. When we get the list firmed up, I hope people will volunteer for various projects so we can get good coverage. I know most people, including me, picked Biological/Medical for their preference, but I am happy to explore outside of that area for our "expeditions". Physical Science
- currently having issues with the servers, and DC Vault isn't listing the right team yet
Leiden Classic - 4338 points - pumps out cpu tasks worth ~30 credits that take 1-2 hours Muon1 Particle Accelerator Design
- 3561 points - non-BOINC, manual setup but not difficult
QMC@Home - 4820 points - ??? Biological and Medical Science Malariacontrol.net
- 7129 points - 1-2 hr CPU tasks - good chance of moving into top 20% in a week with more than one person running Poem@Home
- 3969 points - we are already cranking this one up RNA World @ Home
- 7103 points - CPU tasks that take at least 2 hours and sometimes over 5 days
The Lattice Project - 4237 points - BOINC project, first unit I got will take 72 hours, uses 800MB memory for 1 Mathematical Collatz Conjecture
- 5233 points - ATI GPU preferred Enigma@Home
- 4401 points - works on decoding the enigma machines and stuff.... Fairly quick wus, RHMash's 920 goes about 50 min to an hour....
GIMPS - 0 points - non-BOINC, manual setup, can take 2 weeks to complete 1 prime
NFS@Home - 5510 points - CPU tasks, 45 min to 1hr15min to complete OGR-72
- 1287 points - non-BOINC, but fairly flexible and powerful client
Prime Sierpinski Problem - 2414 points - non-BOINC prime number finder - Pumps out CPU (first pass) task worth ~7500 credits takes 3 to 4 days to complete while its CPU (double check) task worth ~1500 credits takes ~a day to complete. This is a non-BOINC project and it's quite confusing to configure
RC5-72 - 6663 points - use only AMD GPUs - Pumps out GPU task worth ~1500 credits per server communication and takes ~2 hrs 15mins to complete. This is a non-BOINC project and quite easy to install. There is a BOINC Moo! Wrapper for this project that is not optimized whereby its GPU Usage % is much lower.
- skip this one, it's about to be removed from DC Vault Miscellaneous Majestic-12
- 2142 points - this one uses a LOT of internet bandwidth, not much CPU. Yes, indeed it is at least 1Mbit bandwidth running 24/7 for a day to crawl half a million URLs. This is a non-BOINC project and easy to install manually
SZTAKI Desktop Grid - 6195 points - ???
<message edited by Punchy on Thursday, February 16, 2012 7:16 PM>