Soldato
- Joined
- 19 Jan 2005
- Posts
- 2,729
- Location
- Oxford/Bristol
Beyond B0rkley....The Wider World of BOINC
Here it is, a bit later than planned, the very first all things BOINC newsletter.
A combination of work, charity quizzes, work, unplanned trips to Henley, work, surprise England wins in the 6 Nations (and the ensuing celebration) and yet more work left me a bit short on time. Sorry.
Here's a run down of some of the news from the last month or so.
New project: Trojan@Home
I know everyone is in the hunt for more and more credits but one individual took things much too far.... enough about Shady though, as you may have heard someone boosted their score by installing BOINC unsolicited on other computers using a trojan. The excerpt below is taken from this CPDN forum post.
Although we all joke about stealth borgs and throw the odd knowing wink about, hopefully everybody frequenting these forums is aware that the actions of the user in question are utterly reprehensible and, I’m fairly certain, are unlawful in many countries around the World including all EU member states.
Einstein@Home
The Einstein servers have been up and down like a meerkat on a trampoline since the new year but recently the problem was diagnosed and, fingers crossed, fixed.
The issue was with the unique way Einstein issues work in batches. Several WU's can be crunched from a single download which reduces bandwidth for the project and the participant. The S5R1 search completed before data from S6 was ready to be issued and so a more rigorous search of the S5 dataset was commenced called S5RI. This search focussed on a narrower frequency spectrum and resulted in more requests for the server to search through the work unit tables, thus slowing everything down. A fix has been implemented and the number of database requests is down from 950 a second to just 80.
In addition there was found to be hardware clash between the server boards and the RAID controller in the HDD's. This was only diagnosed after a complete server was boxed up and packaged off to the HDD manufacturer in Taiwan!
More technical explanations from Bruce Allen in his forum post.
You can see the progress of the S5RI search in the E@H server pages here.
Leiden@Home
Leiden have had validation issues for a long time. Rounding errors in each WU differ from platform to platform, and even from OS to OS for otherwise identical systems and result in invalid work units. Until now the best solution has been to only send corresponding work to like systems eg Pentium4's with Windows XP but the success rate still seems to be arround 2/3 for validation. A disparity between the returned results in the initial quorum of 2 causes a third WU to be sent out and so on until two returned results agree. The results that disagree are marked invalid and receive 0 credit. Obviously this is extremely wasteful of resources.
Good news though, a newer version of the app, Classical 5.41, contains a new math library called crlib which may alleviate the problem. I'm crunching WU's with the new app now so hopefully there were will be some positive feedback from the project in the near future.
Other projects with the same issue (notably Docking@home) will also be eagerly awaiting the outcome of the trials of the new app.
Orbit@Home
It's taken an age but the Orbit@home project has secured a source of funding and may very soon be sending out work.This project seeks to identify potentially hazardous near earth orbit (NEO) asteroids and track them providing an early warning in the event that a significant event be predicted.
I've copied the Torino Scale for classifying hazardous space rocks below just to scare the bejesus out of you.
NO HAZARD
0. The likelihood of a collision is zero, or is so low as to be effectively zero. Also applies to small objects such as meteors and bodies that burn up in the atmosphere as well as infrequent meteorite falls that rarely cause damage.
NORMAL
1. A routine discovery in which a pass near the Earth is predicted that poses no unusual level of danger. Current calculations show the chance of collision is extremely unlikely with no cause for public attention or public concern. New telescopic observations very likely will lead to re-assignment to Level 0.
MERITING ATTENTION BY ASTRONOMERS
2. A discovery, which may become routine with expanded searches, of an object making a somewhat close but not highly unusual pass near the Earth. While meriting attention by astronomers, there is no cause for public attention or public concern as an actual collision is very unlikely. New telescopic observations very likely will lead to re-assignment to Level 0.
3. A close encounter, meriting attention by astronomers. Current calculations give a 1% or greater chance of collision capable of localized destruction. Most likely, new telescopic observations will lead to re-assignment to Level 0. Attention by public and by public officials is merited if the encounter is less than a decade away.
4. A close encounter, meriting attention by astronomers. Current calculations give a 1% or greater chance of collision capable of regional devastation. Most likely, new telescopic observations will lead to re-assignment to Level 0. Attention by public and by public officials is merited if the encounter is less than a decade away.
THREATENING
5. A close encounter posing a serious, but still uncertain threat of regional devastation. Critical attention by astronomers is needed to determine conclusively whether a collision will occur. If the encounter is less than a decade away, governmental contingency planning may be warranted.
6. A close encounter by a large object posing a serious but still uncertain threat of a global catastrophe. Critical attention by astronomers is needed to determine conclusively whether a collision will occur. If the encounter is less than three decades away, governmental contingency planning may be warranted.
7. A very close encounter by a large object, which if occurring this century, poses an unprecedented but still uncertain threat of a global catastrophe. For such a threat in this century, international contingency planning is warranted, especially to determine urgently and conclusively whether a collision will occur.
CERTAIN COLLISIONS
8. A collision is certain, capable of causing localized destruction for an impact over land or possibly a tsunami if close offshore. Such events occur on average between once per 50 years and once per several 1000 years.
9. A collision is certain, capable of causing unprecedented regional devastation for a land impact or the threat of a major tsunami for an ocean impact. Such events occur on average between once per 10,000 years and once per 100,000 years.
10. A collision is certain, capable of causing global climatic catastrophe that may threaten the future of civilization as we know it, whether impacting land or ocean. Such events occur on average once per 100,000 years, or less often.
Given the projects history of early announcements we should maybe take the claims with a pinch of salt but account creation is enabled so if you want to attach do so here.
Won't you be proud to see your name on the news when you find that number 9 that we're presently long overdue for.
Predictor@Home
While we're on the subject of projects coming back from the dead Predictor, one of the earliest BOINC projects, has returned from it's long hiatus and is progressing with test batches of WU's on a new app called dTASSER. Anyone who wants to help with the next rounds of tests on dTASSER should attach.
BURP
Another project that announced itself to the world rather early is the Big Ugly Rendering Project - BURP. Although officially still in alpha it has recently started issuing rendering jobs and a move into an advanced beta and full production looks to be on the cards. If you want to attach make sure you meet the RAM requirements, some projects require 1GB, and be prepared for a few gremlins as bugfixes and revisions seem to be a daily occurrence on their news pages.
Some examples of jobs already rendered can be found here.
LHC@Home
Although not quite back from it’s long enforced absence LHC is moving towards being a fully active project again.
The project has relocated from CERN to St Mary’s college, London and new servers were installed early this year for hosting the project. Alex and Neasan, the new project admins, list their to do list as:
* Initial Database Migration
* Get test system working with database
* Ensure e-mail is working properly(trickier than it sounds)
* Test migrate again
* Real migration
* Sort out job submission over AFS by SixTrack User
I recently received some WU’s that I assume were test batches for the new scheduler so maybe a relaunch isn’t far away now. Stats still aren’t being updated though.
SIMAP
The most recent protein databases were released back in February and WU's were issued to search for orthologs (similarities) last week.
The SIMAP WU's are very nearly complete the HMMER's are approximately 50% done.
Finally a quick glance at Team OCUK's current status as of 12/03/07:
In the future the amount of actual news during the week(or more likely fortnight) may be actually quite small so I’m considering compiling longer articles on a different topic each issue. Ideas I’ve had so far are-
-Anatomy of a BOINC project – What do all those servers do?
-Validation – Everyone hates it, Why do we need it?
-Does SETI deserve my cycles? (A good one for all the board to get involved in hopefully)
-Focus articles on some of the projects. Science, objectives, feedback and so on.
I think that’s plenty to be getting on with. I look forward to reading all of your comments, feedback (both positive and negative), discussion and, most importantly, corrections below.
Honwards and Hupwards,
Stew
Most recent version of BOINC Client is 5.8.15. for Windows, Mac and Linux.
--------------------------------------------------------------------------
ABC@Home, BOINC Alpha Test, BURP, Chess960@Home, Climate Prediction(CPDN), DepSpid, Einstein@Home, Leiden Classical, LHC@Home, Malaria Control, Nano-Hive@Home, Orbit@Home, Predictor@Home, Primegrid@Home, Proteins@Home, QMC@Home, RALPH@Home, Rectilinear Crossing Number, RenderFarm@Home, Rieselsieve, Rosetta@Home, SETI@Home, SETI@Home and Astropulse Beta, Simap@Home, SpinHenge@Home, STZAKI Desktop Grid, Tanpaku, TMRL Rainbow Table Generator, World Community Grid, XtremLab,
Last edited:
)