PSTSWM T3E-900 Algorithm Comparison

Performance Studies using

PSTSWM

SGI/Cray Research T3E-900 Protocol Performance

(distributed LT experiment II-C )

Date/Person: May 11, 1998 / P. Worley
Platform: T3E-900 at National Energy Research Scientific Computing Center (mcurie.nersc.gov):
   532 450-MHz DEC Alpha EV5 RISC processors
Environment: UNICOS/mk 2.0.2.28
mpt 1.2.0.2
Cray CF90 Version 3.0.2.1
Code Version: 6.3
Compilation Options: -dp -Oscalar3 -Wl,-Dstreams=on
Math Library: libsci
Communication Library: MPI
SHMEM
Number of Timesteps: 12
Results:

Distributed LT (2) (mpi) / Column Major Processor Ordering
Algorithm Comparison
  T42L2     T42L2     T42L16    T85L4     T85L4     T85L32 
  P=2x32     P=4x16     P=8x8     P=2x32     P=4x16     P=8x8  
  optimal algorithm   halfsum  halfsum  halfsum  halfsum  halfsum  halfsum 
  (allreduce-min)/min     0.412    0.123    0.061    0.531    0.213    0.087 
  (generic-min)/min     0.400    0.150    0.016    0.057    0.029    0.014 

Distributed LT (2) (shmem) / Column Major Processor Ordering
Algorithm Comparison
  T42L2     T42L2     T42L16    T85L4     T85L4     T85L32 
  P=2x32     P=4x16     P=8x8     P=2x32     P=4x16     P=8x8  
  optimal algorithm   halfsum  halfsum  halfsum  ringsum  halfsum  ringsum 
  (generic-min)/min     0.382    0.194    0.029    0.095    0.076    0.014 

Distributed LT (2) (combined) / Column Major Processor Ordering
Communication Library Comparisons
  T42L2     T42L2     T42L16    T85L4     T85L4     T85L32 
  P=2x32     P=4x16     P=8x8     P=2x32     P=4x16     P=8x8  
  optimal library   shmem  shmem  shmem  shmem  shmem  shmem 
  (allreduce-min)/min     1.113    0.744    0.324    0.773    0.416    0.177 
  (mpi-min)/min     0.496    0.553    0.249    0.157    0.167    0.084 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:24:42 EDT.