Folding@home v4.0

danball1976danball1976 Wichita Falls, TX
edited February 2004 in Folding@Home
How long ago was F@H version 4 released? I just happened to check up on the ElectronMicroscope website to check for updates (found EMIII 2.4.4) and noticed the comment about F@H 4. So today I downloaded it. Maybe this is more efficient than F@H 3.

BTW, I know I haven't been coming around all that much, too busy with other things.

Comments

  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited February 2004
    I believe it was around Nov or Dec.

    You'll probably want to use the " -forcesse" tag. Camman did a little write-up on it.

    Good to see you around, Dan! :cheers:
  • danball1976danball1976 Wichita Falls, TX
    edited February 2004
    I'm using -Forceasm right now. So it is better to do -forcesse?
  • a2jfreaka2jfreak Houston, TX Member
    edited February 2004
    Believe it or not I was just thinking "where is danball?" last night. Good to see you're stick around and kicking, Dan.
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited February 2004
    danball1976 wrote:
    I'm using -Forceasm right now. So it is better to do -forcesse?
    Might vary depending on the stability of your individual system, but the overall consensus seems to be "yes". On my SSE-capable rigs it has really pumped up production. :thumbsup:
  • MrBillMrBill Missouri Member
    edited February 2004
    danball1976 wrote:
    I'm using -Forceasm right now. So it is better to do -forcesse?
    forcesse is to version 4 what forceasm was to version 3. So, yes...you want to change it. ;)
  • mmonninmmonnin Centreville, VA
    edited February 2004
    Ditto to what MrBill said.:)
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited February 2004
    Also try using -advmethods. I have one box that folds faster with this sequence:

    -forcesse -forceasm -advmethods

    The change is seeable in 10-20% of a Gromacs fold, and -advmethods is more likley to get you Gromacs, which on a fast machine, you DO want.

    Technically, the possible barriers to -forcesse and -forceasm are simply what happens to speed of production and how it affects stability. Use them and the client will warn you that if the folding client is unstable you need to discontinue them, essentially. If both keys of -forcesse and -forceasm do not improve things or make things unstable, use -forcesse only. The client will say if -forcesse is ok further down in log, but it is not perfect in that regard.

    I would experiment if you have an AMD box, with using both -forcesse and -forceasm. I know on my Barton it improved things markedly(I got on average a 45 second to 1 min drop in time expended per percentage of WU completed with Gromacs WUs by using both on AMD box with Barton processor), and that on the Intel P4 box that -forceasm slowed things down a tib when used with -forcesse but -forcesse alone of the two was the most likely to give best performance results.

    Your milage might vary depending on if and how much you OC your box, though, so experiment with these two flags. But -advmethods works quite well now (overall) to reduce the instances of Tinkers per instances of Gromacs assigned.

    John D.
  • mmonninmmonnin Centreville, VA
    edited February 2004
    -forcesse overides -forceasm just like -advmethods overides your WU preference (fah/gah/no-pref).
  • danball1976danball1976 Wichita Falls, TX
    edited February 2004
    I've enabled -forceSSE. Been getting tinkers alot with the old Ver3 client, I hope this new ver4 doesn't get as many

    Also, -advsmethod is enabled
  • a2jfreaka2jfreak Houston, TX Member
    edited February 2004
    was -advsmethod a typo dan, or is that actually what you're using?
    It should be -advmethods
  • danball1976danball1976 Wichita Falls, TX
    edited February 2004
    Oh, yeah, its a typo... I meant -advmethods
  • a2jfreaka2jfreak Houston, TX Member
    edited February 2004
    :thumbsup:
    Thought so, but just wanted to be sure.
  • edcentricedcentric near Milwaukee, Wisconsin Icrontian
    edited February 2004
    you don't need the caps in -forcesse either.
    you did at first but it has been fixed
  • csimoncsimon Acadiana Icrontian
    edited February 2004
    dan where the hell you been?
    use: -advmethods -forcesse
    keep an eye on it for a few days and if you begin to get lock ups/freezing let us know.
  • mmonninmmonnin Centreville, VA
    edited February 2004
    Heh, the client wouldnt start if he typed -advsmethods.
  • csimoncsimon Acadiana Icrontian
    edited February 2004
    you tryed it? lol
  • mmonninmmonnin Centreville, VA
    edited February 2004
    No I have mistyped before, missed a space once in awhile.
  • danball1976danball1976 Wichita Falls, TX
    edited February 2004
    Its running right now with -advmethods and -forcesse. Its the final v4 client
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited February 2004
    Then the only change I would make is to try adding -forceasm for about 1\4 of same WU and then compare times per 10 percent beofre and after adding and see if better or worse. If better, leave, if not take out only the -forceasm. Some boxes fold faster with it also used, others NOT. Right now that box looks warm, using both could raise temps some, so see what temps rise to also with all three used (two you use plus the one I suggested you experiment with). I am getting results not explained by command docs here with 4.0 on a Barton box. Using all three is improving productive output more than with just the two you are using now. So much so, that I have been using all three since a week after I got the final of 4.0, but only on the Barton box have I gotten these results.

    John D.
Sign In or Register to comment.