Results 1 to 12 of 12

Thread: Getting more people to do factoring

  1. #1
    Senior Member dmbrubac's Avatar
    Join Date
    Dec 2002
    Location
    Ontario Canada
    Posts
    112

    Getting more people to do factoring

    I know there has been quite a bit of talk about shifting some emphasis over to factoring or getting new people involved, but I must point out that the biggest obstacle has to be figuring out how to do it.

    Can I suggest someone create a sticky post with simple instructions and links to all the files you need? Wading through the 14 page discussion to figure out what to do is a huge disincentive.

    Maybe a call for help on the SoB news page is called for too. It looks like only a few more people doing factoring would keep us ahead of the PRP effort. Sieving is over 175T, so we must be well into diminishing returns there (at east as opposed to factoring)

    Thanks all!

  2. #2
    Sieve it, baby!
    Join Date
    Nov 2002
    Location
    Potsdam, Germany
    Posts
    959
    A big problem is that the current Windows client crashes when used on a P4.
    Otherwise, there'd be some more ppl doing factoring (e.g. me).

  3. #3
    Senior Member dmbrubac's Avatar
    Join Date
    Dec 2002
    Location
    Ontario Canada
    Posts
    112
    Hmmm. Didn't know that. It's a good thing I'm doing factoring on my Athlon, Sieving on my P4 and PRPing on my dual P3.

    But that proves my point even further. A sum post that outlines all of those critical facts is badly needed.

  4. #4
    Originally posted by dmbrubac
    Hmmm. Didn't know that. It's a good thing I'm doing factoring on my Athlon, Sieving on my P4 and PRPing on my dual P3.

    But that proves my point even further. A sum post that outlines all of those critical facts is badly needed.
    dmbrubac, You can improve our output SIGNIFICANTLY if you change from this allocation. Let your P4 do the PRPing - my guess it it alone will produce more than your dual P3. Let your Athlon do the sieving - they rock at it. And let your dual P3 do the factoring - as long as the machine has more 256MB of memory. Others will also tell you the same or similar allocations are more efficient.

  5. #5
    Senior Member dmbrubac's Avatar
    Join Date
    Dec 2002
    Location
    Ontario Canada
    Posts
    112
    Unfortunately my P4 does not get along too well with PRPing (and yes you are right, my dual P3 only does 250 cem/s combined). I have the most recent client - I think I have some slightly dodgy RAM in there.

    Maybe I'll pull the stick (it has 3 * 512 Mb sticks now)

    The dual p3 is similarly endowed, so perhaps I will reconfigure as you suggest.

  6. #6
    Senior Member Frodo42's Avatar
    Join Date
    Nov 2002
    Location
    Jutland, Denmark
    Posts
    299
    Getting more people to do factoring would be a good thing but the current problems with the Win-client decreases the potiental P-1'rs a lot.
    I agree that a sticky-post with a an explanation of what P-1 is and how to do it would be good.

    10 more Linux 2.4 GHz P4's would make give us the same speed as prp'ing with an value of app. 1.3.
    At present I have the feeling that I'm able to do something like 1/10 of what is necessary but with more people we could raise the value of factors to 1.5 or more.

    Which values for P-1 is optimal at present? I run it with the command line
    ./sbfactor x y 47 1.3 128 &
    at present (I have 256 MB-ram in this machine and to be able to use it when doing stage 2 and GCD I have to keep the memory limit to that, when the box is not being used for anything else for longer time I raise the memory factor to 194, as the core and fluxbox only use app. 40 MB-ram)

  7. #7
    Senior Member dmbrubac's Avatar
    Join Date
    Dec 2002
    Location
    Ontario Canada
    Posts
    112
    Considering I scored 17000+ on the last P-1 factor and 26 total on the last 10 seive factors I submitted, I think I will move PRP to my P4 when it's done it's current sieve and run P-1 on my dual P3 and my Athlon.

    Sieving at 175T just doesn't seem necessary when there are big gaps in P-1 that PRP will pass soon.

  8. #8
    The windows-P4 bug is fixed in v1.2 which i just released. Go the P-1 factorer thread to get it.

    And I agree with you guys, more resources should be on P-1 factoring. Eventually when v3 of the client comes out, P-1 will be done inline with PRPing so that all numbers get tested optimally. But until that time comes, it will help the project to put effort into factoring the ranges just ahead of the assignment level of the network.

    Keep up the good work guys!

    Cheers,
    Louie

  9. #9
    Senior Member Frodo42's Avatar
    Join Date
    Nov 2002
    Location
    Jutland, Denmark
    Posts
    299
    Eventually when v3 of the client comes out, P-1 will be done inline with PRPing so that all numbers get tested optimally.
    I think that is worth a second thought. The problem is that especially stage2 and GCD of P-1 -factoring is pretty memory consuming, and that would compromise one of the big advanteges of the prp-client, that it runs in the background without bothering the user.
    Of couse you could add an option not to use a lot of memory, but then P-1 factoring won't be done very effectively.
    Anyways just a thought, perhaps I'm wrong, that wouldn't be the first time.

  10. #10
    Originally posted by Frodo42
    I think that is worth a second thought. The problem is that especially stage2 and GCD of P-1 -factoring is pretty memory consuming, and that would compromise one of the big advanteges of the prp-client, that it runs in the background without bothering the user.
    Of couse you could add an option not to use a lot of memory, but then P-1 factoring won't be done very effectively.
    Anyways just a thought, perhaps I'm wrong, that wouldn't be the first time.
    True. True.

    In any case, there can at least be an option that would allow people to turn the P-1 factoring on. Also, if there were an option to have your system run factoring with no user intervention, I think more people would do it.

    I understand why people aren't rushing to do P-1 right now. Downloading obscure result.txt files, posting forum messages about reservation ranges, running a cryptic command line program, and then posting output from a text file to a hidden page on our website a few days later is a little beyond our average user.

    As long as 5% of users are down to run it, that's likely enough. SBFactor v1.2 probably won't convince that many people to get involved. v3 of the client should.

    Cheers,
    Louie

  11. #11
    Senior Member dmbrubac's Avatar
    Join Date
    Dec 2002
    Location
    Ontario Canada
    Posts
    112
    garo - you are right about the athlon rocking at the sieve! My P4 2.4 was averaging about 299 k/s, but my Athlon 2200+ (not even Barton) is doing 524! Core clock is 'only' 1800 on that puppy.

    Now that the P4's will factor, I moved the sieve onto one of the P3 CPUs and put the nearest to be passed factor range on the P4.

    PRP can wait for a while - after 26T I don't mind something different.

    Thanks Louie for all the great clients.

  12. #12
    Moderator Joe O's Avatar
    Join Date
    Jul 2002
    Location
    West Milford, NJ
    Posts
    643
    Originally posted by Frodo42
    Getting more people to do factoring would be a good thing but the current problems with the Win-client decreases the potiental P-1'rs a lot.
    I agree that a sticky-post with a an explanation of what P-1 is and how to do it would be good.

    10 more Linux 2.4 GHz P4's would make give us the same speed as prp'ing with an value of app. 1.3.
    At present I have the feeling that I'm able to do something like 1/10 of what is necessary but with more people we could raise the value of factors to 1.5 or more.

    Which values for P-1 is optimal at present? I run it with the command line
    ./sbfactor x y 47 1.3 128 &
    at present (I have 256 MB-ram in this machine and to be able to use it when doing stage 2 and GCD I have to keep the memory limit to that, when the box is not being used for anything else for longer time I raise the memory factor to 194, as the core and fluxbox only use app. 40 MB-ram)
    Does anyone have a FAQ, or the start of one? I would be willing to contribute, but if someone has already done the work, I would hate to redo it.

    We could target people who have machines that are not connected to the internet.
    Joe O

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •