Results 1 to 6 of 6

Thread: GPUGrid: Too Many Work Units

  1. #1
    Gold Member
    FourOh's Avatar
    Join Date
    January 16th, 2013
    Location
    Memphis, TN
    Posts
    1,037

    GPUGrid: Too Many Work Units

    Does anyone know a way of limiting the number of GPUGrid WUs the project sends to a host? By default it sends 2 per GPU, but doesn't seem to distinguish between Nvidia and non-Nvidia cards. So, on a rig with one Nvidia and one AMD, I have 4 WUs at all times, making it virtually impossible to achieve the 24 hour bonus, even though the card can process a long run WU in 10-12 hours.



  2. #2
    Diamond Member
    zombie67's Avatar
    Join Date
    October 24th, 2010
    Location
    Reno, NV
    Posts
    7,290

    Re: GPUGrid: Too Many Work Units

    Set your resource share to zero. It will send only one.
    "Don't confront me with my failures, I had not forgotten them" - Jackson Browne

    Avatar source


  3. #3
    Gold Member
    FourOh's Avatar
    Join Date
    January 16th, 2013
    Location
    Memphis, TN
    Posts
    1,037

    Re: GPUGrid: Too Many Work Units

    Quote Originally Posted by zombie67 View Post
    Set your resource share to zero. It will send only one.
    Perfect! Thank you!



  4. #4
    Gold Member
    Slicker's Avatar
    Join Date
    October 25th, 2010
    Location
    South of Cheeseland
    Posts
    1,253

    Re: GPUGrid: Too Many Work Units

    Quote Originally Posted by FourOh View Post
    Does anyone know a way of limiting the number of GPUGrid WUs the project sends to a host? By default it sends 2 per GPU, but doesn't seem to distinguish between Nvidia and non-Nvidia cards. So, on a rig with one Nvidia and one AMD, I have 4 WUs at all times, making it virtually impossible to achieve the 24 hour bonus, even though the card can process a long run WU in 10-12 hours.
    That was supposed to have been fixed by a recent BOINC clients since it also means that if the nVidia is overworked, it won't pull AMD work from other projects either. If you are running the most recent BOINC version, you might want to report it via the boinc_dev or boinc_alpha email list as still not working.
    Spring 2008 Race: (1st Place)

  5. #5
    Gold Member
    FourOh's Avatar
    Join Date
    January 16th, 2013
    Location
    Memphis, TN
    Posts
    1,037

    Re: GPUGrid: Too Many Work Units

    Quote Originally Posted by Slicker View Post
    That was supposed to have been fixed by a recent BOINC clients since it also means that if the nVidia is overworked, it won't pull AMD work from other projects either. If you are running the most recent BOINC version, you might want to report it via the boinc_dev or boinc_alpha email list as still not working.
    I don't have exclude_gpu for the AMD GPU in my cc_config... maybe that would help? Although I will still have the problem of getting two WUs at a time - if they are both very long 12+ hour units I still won't get the bonus on the 2nd and subsequent ones. Setting the project's resource share to zero is working; I just take the risk that when I complete a unit the server is unavailable to send another.



  6. #6
    Gold Member
    EmSti's Avatar
    Join Date
    March 8th, 2012
    Location
    New York, USA
    Posts
    1,617

    Re: GPUGrid: Too Many Work Units

    Resource share set to zero is how I run it, ever since Zombie taught me that trick. You are right on possibility of server being down, that is the risk, but I prefer the bonus on the work done.

Posting Permissions

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