
   Description of the test1
   ========================

   The HMPI call HMPI_Recon is tested.
    
   The benchmark code used for the updation of processor
   speeds is the serial matrix-matrix multiplication.

   A simple performance model definition shown below is used:

   algorithm simpleGrid(int p, int q)
   {
      coord I=p, J=q;
      parent [0];
   };

   Returns the performances of the processors after updation.

   Files
   -----

   simpleGrid.c     
   simpleGrid.h ---->     Performance model definition

   Test_recon_mxm.c --->  Call to HMPI_Recon containing 
                          the benchmark code to serial matrix-matrix
                          multiplication

   HOW TO RUN
   ----------
   shell$ hmpibcast Test_recon_mxm.c simpleGrid.*

   shell$ hmpiload -o Test_recon_mxm Test_recon_mxm.c

   shell$ hmpirun Test_recon_mxm
Executing the benchmark code in call to HMPI_Recon
Total number of processes available for computation is = 4
Creating a grid (2, 2) of processes
Updated processor performances are :
348.00 348.00 348.00 348.00

   Description of the test2
   ========================

   The HMPI call HMPI_Recon is tested.
    
   The benchmark code used for the updation of processor
   speeds is the serial Cholesky Factorization.

   A simple performance model definition shown below is used:

   algorithm simpleGrid(int p, int q)
   {
      coord I=p, J=q;
      parent [0];
   };

   Returns the performances of the processors after updation.

   Files
   -----

   simpleGrid.c     
   simpleGrid.h ---->     Performance model definition

   Test_recon_cholesky.c --->  Call to HMPI_Recon containing 
                          the benchmark code to serial 
                          Cholesky Factorization

   HOW TO RUN
   ----------

   shell$ hmpibcast Test_recon_cholesky.c simpleGrid.*

   shell$ hmpiload -o Test_recon_cholesky Test_recon_cholesky.c

   shell$ hmpirun Test_recon_cholesky
Executing benchmark code of cholesky factorization
Total number of processes available for computation is = 4
Creating a grid (2, 2) of processes
Updated processor performances are :
213.00 213.00 213.00 213.00

   Description of the test3
   ========================

   The HMPI call HMPI_Recon is tested.
    
   The benchmark code used for the updation of processor
   speeds is the serial matrix-matrix multiplication.

   A simple performance model definition shown below is used:

   algorithm simpleGrid(int p, int q)
   {
      coord I=p, J=q;
      parent [0];
   };

   Returns the performances of the processes available for 
   global computation.

   Also returns the performances of the processes of the group.

   Files
   -----

   simpleGrid.c     
   simpleGrid.h ---->     Performance model definition

   Test_recon_mxm_processes_perf.c --->  Call to HMPI_Recon containing 
                                         the benchmark code to serial 
                                         matrix-matrix multiplication

   HOW TO RUN
   ----------

   shell$ hmpibcast Test_recon_mxm_processes_perf.c simpleGrid.*

   shell$ hmpiload -o Test_recon_mxm_processes_perf Test_recon_mxm_processes_perf.c

   shell$ hmpirun Test_recon_mxm_processes_perf
Executing the benchmark code in call to HMPI_Recon
Total number of processes available for computation is = 16
Creating a grid (4, 4) of processes
Updated performances of the processes available for global computation are :
145.87 115.95 259.27 111.45 122.11 168.58 122.48 168.02 130.64 133.62 185.32 185.69 104.59 153.51 127.33 92.41
Updated performances of the processes of the group are :
145.87 115.95 259.27 130.64 133.62 185.32 122.11 168.58 122.48 104.59 153.51 127.33 111.45 185.69 168.02 92.41

