Announcement

Collapse
No announcement yet.

Possible work-around for genome-crashes.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Possible work-around for genome-crashes.

    One solution popping up in the "official" genome-group is this:

    Then crashing, change the seed-number in seed.dat. It's a 32-bit-number, therefore max 2^32.
    Stefan's comment to this is:
    <blockquote>This will work, and not cause any major problems. However, I wouldn't make a habit of it, since it's harder for us to verify the data if the seed has been changed.</blockquote>

    Of course, a wu crashing in first sequence is probably corrupted, and should be discarded. Another thing, if re-looping same wu, don't choose the same seed as you've already used for this wu...

  • #2
    After you've changed the seed, will the entire unit be recalculated? If so, I'd take the safe way out and just get myself a new unit.

    Martin

    Comment


    • #3
      Originally posted by Ees
      After you've changed the seed, will the entire unit be recalculated? If so, I'd take the safe way out and just get myself a new unit.

      Martin
      It continues from the sequence where it crashed.

      > Hi,
      >
      > I had the 0.98 client cause a GPF on the 25th sequence of a 99aa this
      > morning. It takes nearly a day to process that size of gene on that
      > particular machine I found that if I changed the seed.dat manually,
      > the work unit would continue processing with no problems. Since the
      > 30th sequence is the only one added to the database, how would this
      > affect the results if I uploaded the completed WU? Is this a bad idea
      > to change the seed? Inquiring minds want to know...
      Quoted post from Yahoo

      Kim
      P5B Deluxe, C2D E6600, Scythe Ninja, G.Skill 2GBHX
      Raptor 150x3, Plextor PX-760SA, X-Fi Elite, 7900GT, 21" CM813ET Plus, CM Stacker

      Comment


      • #4
        Hopefully fixed the genome-bug.

        <blockquote>In this same vein, I've tracked down the infamous memory access violation bug, and pending the success of my overnight tests, should have a new version out tomorrow.</blockquote>

        from http://groups.yahoo.com/group/genomeathome/message/2817

        If released tomorrow, this version will not contain http-support.

        Comment

        Working...
        X