Results 1 to 9 of 9

Thread: APS@home (new boinc project)

  1. #1
    NeoGen's Avatar
    NeoGen is offline AMD Users Alchemist Moderator
    Site Admin
    Join Date
    Oct 2003
    Location
    North Little Rock, AR (USA)
    Posts
    8,451

    APS@home (new boinc project)

    APS@home is a research project that uses Internet-connected computers to do research, currently in atmospheric dispersion as it relates to the accuracy of climate prediction. You can participate by downloading and running a free program on your computer.

    Please contact James Dorsey with any enquiries. See project personnel pages. APS@home is based at the Centre for Atmospheric Science at the at University of Manchester.

    http://www.data-dump.org/


    Account Creation is closed

  2. #2
    NeoGen's Avatar
    NeoGen is offline AMD Users Alchemist Moderator
    Site Admin
    Join Date
    Oct 2003
    Location
    North Little Rock, AR (USA)
    Posts
    8,451

  3. #3
    Join Date
    Jan 2007
    Location
    Vermont, USA
    Posts
    1,379
    I emailed the project sponser (Dr. J.R. Dorsey, <james@surface-exchange.net>)....This is another Climate Prediction Project.


    Hi Bruce,

    I'm not really selecting at the moment. It's basically a free for all.
    I'm sending limited numbers of work units out (5 per user per day) to
    help me test changes to the server software. If you'd like to help out,
    you're very welcome. Always nice to have people with different
    platforms...

    The work generator is down at the moment, but I hope it should be back
    by the end of the day. Looking forward to working with you...

    Cheers,
    James


    So, I'm in and crunching my '5 wu's a day'. I sent a return email requesting access for a few more AMD members (he is looking for Mac and Linux testing platforms also). I'll let you know. James sounds pretty layed back about access.


    EDIT: James just replied. He wants to keep the user list below 100. So I can give out the access code to 2 people max. I'd like to get at least 1 Linux user for him if possible. Let me know.

    Also, I DID create the Team (for later xfer to Vaughan). The Team .gif did not display correctly. It may be a APS site problem...
    Last edited by Bender10; 05-17-2007 at 06:08 PM. Reason: Access code.
    Logic is the art of being wrong with confidence.


  4. #4
    Join Date
    Jan 2007
    Location
    Vermont, USA
    Posts
    1,379
    There is some headway being made in the points-for-work system. Hopefully without a quorum. We'll see. The test wu's aren't too long. About 90-100 minutes or so on a Pentium D 3.2 Ghz. I'm running a wu on my X2 4200 now. AMD is supposed to run them faster.... The plan is to keep the wu's short. There is already a Version .03 in the works to fix more problems (V.02 is current now).

    It's going to be a little while before the project goes public. Plus there is the 5 wu a day limit (per computer).
    Logic is the art of being wrong with confidence.


  5. #5
    Join Date
    Jan 2007
    Location
    Vermont, USA
    Posts
    1,379
    Right now my (stock) X2 4200 is running wu's in about 70 minutes, verses 90 minutes for the Intel D 3.2....
    Logic is the art of being wrong with confidence.


  6. #6
    NeoGen's Avatar
    NeoGen is offline AMD Users Alchemist Moderator
    Site Admin
    Join Date
    Oct 2003
    Location
    North Little Rock, AR (USA)
    Posts
    8,451

  7. #7
    Join Date
    Jan 2007
    Location
    Vermont, USA
    Posts
    1,379
    Credit reporting news (excerpts from the board).

    Since there's an API call to determine how many calculations were made, this strikes me as the fairest way to grant credit. It's then not tied to benchmark values, which has apparently been a source of controversy in the past.

    Can anyone tell me why they think we *shouldn't* go with a system which grants credit on a "per floating point operation" basis? It seems the fairest way to me, and should be impossible to manipulate.

    Thanks again for taking the time to engage in the discussion. Hopefully if we can reach a robust consensus about this, then I won't spend the next two years explaining it to every newcomer, 'cause you lot can take turns explaining it too. ;-)
    Logic is the art of being wrong with confidence.


  8. #8
    Join Date
    Jan 2007
    Location
    Vermont, USA
    Posts
    1,379
    I've been smurfing around the user stats on APS. It seems there is quite a bit of positive feedback FOR the "points per floating point operation" credit system.

    Which is strange....Checking 3 of the top 5 users (those I can view), they are all running "Optimized Clients" during the TEST phase of this project. I suppose because they can.

    Here is an X2 3800...
    Measured floating point speed 6140.23 million ops/sec
    Measured integer speed 14380.25 million ops/sec

    and my stock X2 4200 on a standard BOINC client....I know the 3800's are nice, but..
    Measured floating point speed 2109.9 million ops/sec
    Measured integer speed 3922.07 million ops/sec



    Anyway...
    If nobody here wants to join the test, should I advertise in the general population (as it were)? Inquiring minds would like to know..

    Here is the invite code (apshomez0275). Don't everyone jump on this, I'll look like a weasel. A couple 3-4 new users and all the cpu's you want to use. Remember, 5 wu's (per computer) a day. BOINC.BE and UK BOINC Team already have 12 or so Team members here crunching.

    And Dr. James Dorsey really wants our input...and seems to listen.
    Logic is the art of being wrong with confidence.


  9. #9
    Join Date
    Jul 2003
    Location
    Sydney, Australia
    Posts
    5,642
    No Linux here atm.

    Try sending a PM to AMDave or Daddygeek.


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •