AMD flags?

deicistdeicist Manchester, UK
edited February 2006 in Folding@Home
Just setting up my new opty for folding and was wondering what's best for AMD systems. from what I've heard setting the 'deadlineless' option to 'yes' is good, and basically leave everything else at default. Is that true? What does the deadlineless option achieve? I mean, I know it requests work units without a dedline, but why do AMDs do better with that option (if indeed they do). I've also set the 'request large WUs' on both cores (I have 2GB of RAM) is that a good way to go?

Comments

  • csimoncsimon Acadiana Icrontian
    edited February 2006
    It just allows you to get larger wu's ...and they fold fast. Very fast on your opty! No flags.
  • csimoncsimon Acadiana Icrontian
    edited February 2006
    csimon wrote:
    It just allows you to get larger faster folding tinkers ...and they fold fast. Very fast on your opty! No flags.
  • lemonlimelemonlime Canada Member
    edited February 2006
    deicist wrote:
    Just setting up my new opty for folding and was wondering what's best for AMD systems. from what I've heard setting the 'deadlineless' option to 'yes' is good, and basically leave everything else at default. Is that true? What does the deadlineless option achieve? I mean, I know it requests work units without a dedline, but why do AMDs do better with that option (if indeed they do). I've also set the 'request large WUs' on both cores (I have 2GB of RAM) is that a good way to go?

    When you set your client to 'deadlineless', you'll get exclusively 241/239 point Tinker WUs, which get eaten up very quickly by AMD processors (especially 754/939 procs). Without that option, you'll get some small gromacs etc, which are worth very little point wise, and take longer to complete. I've been using that method for a while now and does very well.
  • Sledgehammer70Sledgehammer70 California Icrontian
    edited February 2006
    Yes Prof gve me some good insight getting my AMD's folding a bit faster... I just wish i coul suse all my towers at home to fold... I had to elimiate folding form 3 as they were running 100% CPU without Folding running! :(
  • JakeJake Alec Baldwin's Chest Hair
    edited February 2006
    Should Athlon XPs also use these same flags? Back in the day (WAY back in the day) I remember the switch from Tinkers to Gromacs being hailed by all because of the associated point increase; is this no longer the case for Athlon XPs, as it apparently is for Opterons/A64s?

    Also, should the flag for allowing large WUs be set to "no"?
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited February 2006
    Jake wrote:
    Should Athlon XPs also use these same flags? Back in the day (WAY back in the day) I remember the switch from Tinkers to Gromacs being hailed by all because of the associated point increase; is this no longer the case for Athlon XPs, as it apparently is for Opterons/A64s?

    Also, should the flag for allowing large WUs be set to "no"?
    It's a crapshoot, with the odds decidedly in favor of going "deadliness" for all AMD rigs. You may do slightly better on some of the bigger WU's (600-Pointers), but that is more than offset by the price you pay if you lay into a bunch of those low-point Gromacs. All AMD rigs that I've used the "deadliness" WU's on showed a ppw increase - sometimes as much as double. :eek2:

    This seems to be the case for Celeron rigs, too, though my test sample has been much smaller with them.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    that is more than offset by the price you pay if you lay into a bunch of those low-point Gromacs
    Those small Gromacs don't produce much with P4s either.
Sign In or Register to comment.