Re: Support for NVIDIA Tesla

From: Paul H. Hargrove (PHHargrove_at_lbl_dot_gov)
Date: Tue Feb 17 2009 - 14:12:59 PST

  • Next message: Tahar Amari: "Re: shared data and operator"
    Jeff,
    
      The Berkeley UPC runtime is quite portable to POSIX-type systems.  
    However, the Tesla taken alone is lacking nearly all the support that a 
    language runtime would require.
      At present I am not aware of any plans to port a UPC runtime to the 
    Tesla (either in a heterogeneous system such as host node + Tesla or on 
    the Tesla alone).   Others on this list might know more.
    However:
      + I am not aware of any specific reasons why one could not write UPC 
    code for the host node in which the individual UPC threads utilize the 
    Tesla for offloading computation.
      + If somebody has ported MPI to the Tesla, it might be possible to run 
    the Berkeley UPC runtime using its support for running over MPI.
    
    -Paul
    
    Jeff Glickman wrote:
    >
    > Are there any plans to support the NVIDIA Tesla CPU/system architecture?
    >
    > (http://www.nvidia.com/object/tesla_computing_solutions.html)
    >
    >  
    >
    > Jeff Glickman
    >
    >  
    >
    >  
    >
    
    
    -- 
    Paul H. Hargrove                          PHHargrove_at_lbl_dot_gov
    Future Technologies Group                 Tel: +1-510-495-2352
    HPC Research Department                   Fax: +1-510-486-6900
    Lawrence Berkeley National Laboratory     
    

  • Next message: Tahar Amari: "Re: shared data and operator"