Parallel Programming with Oracle Developer Tools

An Oracle White Paper May 2010 Paralel Programing with Oracle ® Developer Tols Oracle White Paper—Parallel Programming with Oracle Developer Tools Introduction.............................................1! Target Audience.......................................2! Multicore Procesor Technology.............................3! Basic Concepts in Paralel Programing......................6! What is a Thread?......................................6! Why Paralelization?....................................7! What is Paralelization?..................................8! Paralel Architectures.....................................9! Cache Coherence......................................9! The Symetric Multiprocesor (SMP) Architecture............9! The Non-Uniform Memory Aces (NUMA) Architecture.......10! The Hybrid Architecture................................1! The Cache Coherent Non-Uniform Memory Aces (c-NUMA) Architecture................................12! Paralel Programing Models..............................13! An Overview of Comon Paralel Programing Models.......13! Automatic Paralelization................................15! The

OpenMP Paralel Programing Model.................16! The Mesage Pasing Interface (MPI) Paralel Programing Model...................................18! The Hybrid Paralel Programing Model...................21! An Example Program Paralelized Using Various Models........2! The Example Program.................................2! Automatic Paralelization Using The Oracle Solaris Studio C Compiler..........................................25! Paralelization Strategy.................................27! Paralelizing The Example Using OpenMP..................28! Paralelizing The Example Using MPI......................39! Paralelizing The Example Using The Hybrid Model...........50! Aditional Considerations Important In Paralel Aplications......5! Oracle White Paper—Parallel Programming with Oracle Developer Tools Paralel Computing, Floating-Point Numbers And Numerical Results.............................................5! Elapsed Time And CPU Time............................58! Paralel Overheads....................................59! Paralel Sped Up And Paralel Eficiency..................60! Amdahl's Law........................................61! Performance Results....................................64! Conclusion............................................6! References............................................68! Apendix A - Aditional Source Codes.......................69! The check_numerical result() function.....................69! The get_num_threads() Function.........................70! The get_wal_time() Function............................71! The setup_data() Function..............................72! Apendix B - Ful C Source Code Of The MPI Implementation....73! Apendix C - Ful Fortran Source Code Of The Hybrid Implementation.........................................7! Oracle White Paper—Parallel Programming with Oracle Developer Tools 1 Introduction This technical white paper targets developers interested in learning about paralel programing. After introducing major concepts neded to get started writing paralel aplications, it delves into two comonly used paralel programing models: OpenMP and the Mesage Pasing Interface (MPI). OpenMP provides for paralelization of aplications runing within a single multicore-based system. MPI provides for...

Website: developers.sun.com | Filesize: 2387kb
No of Page(s): 87
Download Parallel Programming with Oracle Developer Tools.pdf