ppd increase w/ p1159

2»

Comments

  • edited April 2006
    So I found out that core 78 is specially made for AMD SSE. I tried deleting all the other cores except for this one, but F@H still downloads the 7a core. How can I force F@H to use a specific core? This should help crank through the WU faster right?
  • QCHQCH Ancient Guru Chicago Area - USA Icrontian
    edited April 2006
    When F@H grabs a protein, it has to use a specific CORE. You cannot force a particular core without deleting the protein and hope for another that DOES use the Core78... But that really flies in the face of "research Project". I would just stick with the "No Deadline" work units and hope for the best.

    Unless you have several systems and want to jockey the proteins around manually to maximize them to the CPU type... that's been done before and still brings us closer to cures... Deleting proteins just to get the best fit is really not a ethical thing...
  • edited April 2006
    So what does machine ID mean? I read about when 2 computers have the same user id# that they both work on the same unit, and whichever one finished first gets credit. The slow once just gets discarded. I set up the msi installer, so if I were to install it with machine id=1 on 5 computers, would they all be working on the same thing and only 1 gets credit?
  • Sledgehammer70Sledgehammer70 California Icrontian
    edited April 2006
    Machine Id is referring to the core of the CPU. If you have a Intel Pentium 4 chip with HT you would create 2 folders with the F@H program load the .exe for folder 1 set it up and select Machine ID #1 than go into your 2nd folder and do the same thing but select machine ID #2...

    In my case my rig is a dual Xeon with HT so I have 4 folders with machine id's 1 thru 4
  • edited April 2006
    Wow, so how come it doesn't set affinity to the correct core for me? I have machine id 1 and 2. Yet the cores both have affinity set to both cores unless I manually set them in task manager.
  • Sledgehammer70Sledgehammer70 California Icrontian
    edited April 2006
    Here is an image or 2 that might help you out with the setup :) set your F@H with these settings and your CPU should be at 100% across the board :)

    Just use your username instead of mine :)
  • edited April 2006
    It's not a problem with them not using 100% cpu. It has to do with the threads. The program launches with affinity on both cores. If you go to task manager and look at affinity, the core is set to core0 and core1. I manally make each F@H core to their own thread. Does setting to seperate cores help at all?

    This is my config:
    [settings]
    machineid=2
    username=skankinred
    team=93
    bigpackets=yes

    [http]
    active=no
    usereg=no

    [core]
    cpuusage=100
    checkpoint=10

    [power]
    battery=yes

    [clienttype]
    type=2
  • csimoncsimon Acadiana Icrontian
    edited April 2006
    if you have -advmethods configured you should disable that to increase your chances of getting tinkers IMO.
  • edited April 2006
    What do you think of the seti@home project? The program's much more advanced, you can set it to as many processors(threads) as you have handy and the statistics website seems much more involved.
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited April 2006
    SETI may or may not come up with anything. We all know that Cancer, Parkinson's, etc, are real and need as much research applied to a solution as we can muster.

    You can run FAH on any number of processors, and can run two instances on a single-CPU rig with HT. There are plenty of flashy stats available at sites like this. :fold:
  • QCHQCH Ancient Guru Chicago Area - USA Icrontian
    edited April 2006
    Seti is OK... I look at our F@H project as something that is good for all of humanity. Unless the little green men have a cure for cancer, I'll feel better letting F@H run on all my systems.

    Seti was around before F@H and they have gone to great efforts to make the Seti application look and feel better.
Sign In or Register to comment.