Sample 2 - IMB-MPI1 PingPing Allreduce

The following example shows the results of the PingPing

<..> 
 -np 6 IMB-MPI1
  pingping allreduce -map 2x3 -msglen Lengths -multi 0
Lengths file:
0
100
1000
10000
100000
1000000
#---------------------------------------------------
# Intel (R) MPI Benchmark Suite V3.2.2, MPI1 part
#---------------------------------------------------
# Date                  : Thu Sep 4 13:26:03 2008
# Machine               : x86_64
# System                : Linux
# Release               : 2.6.9-42.ELsmp
# Version               : #1 SMP Wed Jul 12 23:32:02 EDT 2006
# MPI Version           : 2.0
# MPI Thread Environment: MPI_THREAD_SINGLE
# New default behavior from Version 3.2 on: # the number of iterations per message size is cut down # dynamically when a certain run time (per message size sample) # is expected to be exceeded. Time limit is defined by variable # SECS_PER_SAMPLE (=> IMB_settings.h) # or through the flag => -time
  # Calling sequence was: # IMB-MPI1 pingping allreduce -map 3x2 -msglen Lengths #          -multi 0
# Message lengths were user-defined # # MPI_Datatype                        :   MPI_BYTE # MPI_Datatype for reductions    :   MPI_FLOAT # MPI_Op                                :   MPI_SUM  # # # List of Benchmarks to run: # (Multi-)PingPing # (Multi-)Allreduce #-------------------------------------------------------------- # Benchmarking Multi-PingPing # ( 3 groups of 2 processes each running simultaneously ) # Group  0:     0    3 # # Group  1:     1    4 # # Group  2:     2    5 # #-------------------------------------------------------------- # bytes #rep.s t_min[μsec] t_max[μsec] t_avg[μsec] Mbytes/sec       0   1000       ..          ..          ..          ..     100   1000          1000   1000         10000   1000        100000    419      1000000     41     #-------------------------------------------------------------- # Benchmarking Multi-Allreduce # ( 3 groups of 2 processes each running simultaneously ) # Group  0:     0    3 # # Group  1:     1    4 # # Group  2:     2    5 # #-------------------------------------------------------------- #bytes #repetitions  t_min[μsec]  t_max[μsec]  t_avg[μsec]       0         1000         ..          ..           ..     100         1000           1000         1000          10000         1000         100000          419       1000000           41         #-------------------------------------------------------------- # Benchmarking Allreduce # #processes = 4; rank order (rowwise): #     0    3 # #     1    4 # # ( 2 additional processes waiting in MPI_Barrier) #-------------------------------------------------------------- # bytes #repetitions  t_min[μsec]  t_max[μsec]  t_avg[μsec]       0         1000         ..          ..           ..     100         1000           1000         1000          10000         1000        100000          419      1000000           41   #-------------------------------------------------------------- # Benchmarking Allreduce # # processes = 6; rank order (rowwise): #     0    3 # #     1    4 # #     2    5 # #-------------------------------------------------------------- # bytes #repetitions  t_min[μsec]  t_max[μsec]  t_avg[μsec]       0         1000         ..          ..           ..     100         1000           1000         1000          10000         1000        100000          419      1000000           41     # All processes entering MPI_Finalize 

Submit feedback on this help topic