Announcement

Collapse
No announcement yet.

Re-crunching Genome Units. Faster or slower?

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

  • Re-crunching Genome Units. Faster or slower?

    Yet again I'm looking for some confirmation

    I've run a 58aa and a 59aa together, using two ghclients at the same time.
    Although the 58aa should be shorter than the 59aa, it's running slower than the 59aa. But that's not my question.

    I ask you recrunchers if you've seen a speed-up in crunch time, after the first crunch of 30 sequences has finished? I mean, the first crunch is ready - does the crunching of the second set of 30 sequences go any faster on your machine, is it slower or does it stay the same?

    Plus, does any of the recrunchers stop after a second or third time being crunched? I had one of my 59aa's stop on its own after it had crunched the unit a second time.

    Jord.
    Jordâ„¢

  • #2
    Don't know about the times yet as I have on machine that I'm testing that on now -2 58 aa's. As for the re-crunch stopping, I found my client was either crashing (had to re-install the client on 3 machines because the client would crash with any wu's it had) or the wu had the dreaded error message. I had one machine last week that I hadn't checked -I have to manually connect to send wu's as they're on my home phone line- and it had 10! genes ready to transmit! So I guess if the client doesn't crash it will recrunch forever? Just my observations not scientific data but I hope it helps as I wonder about the same things.
    Pete

    Comment


    • #3
      I'm pretty sure I've noticed it being faster! Though I've also noticed the crash problems after a while - the install just corrupts

      P.
      Meet Jasmine.
      flickr.com/photos/pace3000

      Comment


      • #4
        Well, the second time it should be a little bit faster than the first time, because it doesn't need to run filter.

        If I remember correctly it's nearly constant time-difference on the result-files if I've let the machine run unattended. While using, it's of course marked variation.

        As for 59aa faster than 58aa, this is since different molecules has different complexity and therefore different run-times.
        It can also be because the split between the clients aren't 50-50. Atleast on NT, the default is that the foreground-application gets more cpu than background-applications, if you're not running server.

        Comment

        Working...
        X