1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

NFS@home challenge

Discussion in 'Team OcUK Distributed Computing Projects' started by hanluc, Dec 10, 2012.

  1. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    First sorry for the late reply.
    Let me try to explain this in a simple manner :D:

    --------------------------------------------------------------------------
    2. Sieving for Relations

    The sieving step is not the theoretically most complex part of the algorithm of factorization, but it is the most time consuming part because it iterates over a large domain with some expensive calculations like division and modulo, although some of these can be avoided by using logarithms.
    In general optimization of the sieving step will give the biggest reduction in actual running time of the algorithm. It is easy to use a large amount of memory in this step, and one should be aware of this and try to reuse arrays and use the smallest possible data types. The factor bases can for record factorizations contain millions of elements, so one should try to obtain the best on-disk/in-memory tradeoff.

    The purpose of the sieving step is to find usable relations, i.e. elements (a, b) with the following properties
    • gcd(a, b) = 1
    • a + bm is smooth over the rational factor base
    • b^deg(f)*f(a/b) is smooth over the algebraic factor base

    Finding elements with these properties can be done by various sieving methods like the classical line sieving or the faster lattice sieving, the latter being used at NFS@Home.

    The lattice sieving was proposed by John Pollard in "Lattice sieving, Lecture Notes in Mathematics 1554 (1991), 43–49.". The factor bases are split into smaller sets and then the elements which are divisible by a large prime q are sieved. The sizes of the factor bases have to be determined empirically, and are dependent on the precision of the sieving code, if all smooth elements are found or if one skips some by using special-q methods.

    One advantage the lattice siever has is the following. The yield rate for the line siever decreases over time because the norms get bigger as the sieve region moves away from the origin. The lattice siever brings the sieve region "back to the origin" when special-q's are changed. This might be its biggest advantage (if there is one).
    --------------------------------------------------------------------------

    So we were finding less elements per region than we were expecting to be able to build the matrix for the post-processing phase.

    Carlos
     
  2. andy_taximan

    Wise Guy

    Joined: Apr 20, 2014

    Posts: 1,648

    Location: Herts/uk

    Ah so That explains it !! clear as mud :rolleyes:
     
  3. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    Bad news.
    Major corrupted file on G6p490 which got propagated to the backup before being discovered. Needs to redo up to 1500M (now at 255M).

    Help requested, application lasieve5f.

    Thank you in advance,

    Carlos
     
  4. hanluc

    Mobster

    Joined: Oct 22, 2010

    Posts: 2,731

    Location: Ratae Corieltauvorum

    Recent works units seem to have half the runtime of past ones, but give out the the same points, hence why our production has increased :)
    Nice little cluster of teams to stomp ahead ;)
     
  5. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    You're correct. The wu's are sieving the same region size but because they have less "content" they crunch faster.
     
  6. PDW

    Wise Guy

    Joined: Feb 21, 2014

    Posts: 1,440

    Hi Carlos, thanks for the explanation you gave earlier, I did read it and not sure I fully mastered the concept (I'd probably need a second lesson for that) but it did help.

    How many wu are left to do now ?
     
  7. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    Not sure how much is still left to do regarding G2m1285. Probably we will go up to 3600M (my guess, still awaiting for admin feedback).--->quick wu's: 209,000 wus. My last wu received as now is 3182M.

    Normal wu's from G6p490b from q=260M to 1500M (redo work 620,000 wu's).
     
  8. andy_taximan

    Wise Guy

    Joined: Apr 20, 2014

    Posts: 1,648

    Location: Herts/uk

    hey im back up to 5th spot in top hosts :)
     
  9. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    Got the reply.

    Greg queued through 3700M, means we have ~239k tasks still to process.
     
  10. PDW

    Wise Guy

    Joined: Feb 21, 2014

    Posts: 1,440

    Looks like the majority of the ~239k quicker wu's (G2m1285) have been done as the G6p490b wu's are coming through now.

    Once these G6p490b wu's are done is that your 'emergency' over ?
     
  11. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    Yes.
    Thank you for the team support.
     
  12. hanluc

    Mobster

    Joined: Oct 22, 2010

    Posts: 2,731

    Location: Ratae Corieltauvorum

    Abandoning this for now, too many computation errors on the latest batch.
     
  13. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    6,490b was pushed to 1600M.
    90,000 tasks to be done with leftovers.

    Thank you for your support.

    Carlos
     
  14. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    Hi,

    Can I request support from you all to run lasieve5f application where we queued more than 1M wus to sieve 6,383-, 7,353+, 7,353-, and 2,1009+ composites.

    Thank you in advance,

    Carlos
    NFS@Home Moderator
     
  15. colinmcr

    Wise Guy

    Joined: Mar 17, 2009

    Posts: 2,465

    Location: Scotland

    I'll try if I can fix my OS failing to boot properly.

    EDIT: Ok I'm running again selected only lasieve5f WU to DL.
     
    Last edited: Jun 28, 2016
  16. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

  17. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    More than 2M wus to process. Come and join the Pirates.
     
  18. hanluc

    Mobster

    Joined: Oct 22, 2010

    Posts: 2,731

    Location: Ratae Corieltauvorum

    We're in.
     
  19. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    We have a set of 750k 16e wus to immediately start crushing and more 2M to come afterwards. Every help is needed and I suppose you need some help on formula boinc.
     
  20. Carlos Pinho

    Associate

    Joined: Dec 12, 2012

    Posts: 73

    Location: Milton Keynes , UK

    7/10-14/10 Need Faster Ships, Pirates Ahoy part deux NFS@Home Challenge

    http://boincstats.com/en/stats/challenge/team/chat/815

    Name Need Faster Ships, Pirates Ahoy part deux NFS@Home Challenge
    Status Upcoming
    Project NFS@Home
    Issued by Anguillan Pirates
    Start time 2016-10-07 00:00 UTC
    End time 2016-10-14 00:00 UTC
    Late entrants allowed? Yes
    Number of teams participating » 7
    Number of users participating 0