PSTSWM T3E-900 Algorithm Comparison

Performance Studies using

PSTSWM

SGI/Cray Research T3E-900 Algorithm Comparison

(transpose FFT experiment B)

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:

Transpose FFT (mpi) / Row Major Processor Ordering
Algorithm Comparison
  T21L16    T21L16    T42L16    T42L32    T42L16    T85L32 
  P=32x2     P=16x4     P=8x8     P=32x2     P=16x4     P=8x8  
  optimal algorithm   alltoallv  alltoallv  alltoallv  alltoallv  alltoallv  alltoallv 
  (alltoallv-min)/min     0.000    0.000    0.000    0.000    0.000    0.000 
  (generic-min)/min     0.335    0.198    0.048    0.088    0.123    0.015 

Transpose FFT (shmem) / Row Major Processor Ordering
Algorithm Comparison
  T21L16    T21L16    T42L16    T42L32    T42L16    T85L32 
  P=32x2     P=16x4     P=8x8     P=32x2     P=16x4     P=8x8  
  optimal algorithm   swtrans  srtrans  srtrans  srtrans  srtrans  srtrans 
  (generic-min)/min     0.182    0.126    0.046    0.052    0.077    0.021 

Transpose FFT (combined) / Row Major Processor Ordering
Communication Library Comparisons
  T21L16    T21L16    T42L16    T42L32    T42L16    T85L32 
  P=32x2     P=16x4     P=8x8     P=32x2     P=16x4     P=8x8  
  optimal library   shmem  shmem  shmem  shmem  shmem  shmem 
  (alltoallv-min)/min     0.475    0.536    0.262    0.148    0.197    0.077 
  (mpi-min)/min     0.475    0.536    0.262    0.148    0.197    0.077 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:28:44 EDT.