PSTSWM SP2-66 Algorithm Comparison

Performance Studies using

PSTSWM

IBM SP2-66 Algorithm Comparison

(distributed FFT experiment A2)

Date/Person: March 22, 1996 / P. Worley
Platform: IBM SP2 at NASA Ames Research Center (babbage.nas.nasa.gov):
   160 RS6000/590 nodes ("wide", 66.7 MHz POWER2 RISC chip set)
Environment: AIX 4.1.3.3;   POE 2.1.0.8
Code Version: 6.0
Compilation Options: mpxlf -O3 -qarch=pwr2
Math Library: none
Communication Library: MPI
MPL
Number of Timesteps: 12
Results:

Distributed FFT (mpi)
Algorithm Comparison
  T21L2     T42L2     T85L1     T85L4  
  P=16     P=8     P=16     P=8  
  optimal algorithm   dfft  dfft  dfft  dfft 
  (generic-min)/min     0.558    0.017    0.029    0.090 

Distributed FFT (mpl)
Algorithm Comparison
  T21L2     T42L2     T85L1     T85L4  
  P=16     P=8     P=16     P=8  
  optimal algorithm   generic  dfft  dfft  dfft 
  (generic-min)/min     0.000    0.004    0.025    0.078 

Distributed FFT (combined)
Communication Library Comparisons
  T21L2     T42L2     T85L1     T85L4  
  P=16     P=8     P=16     P=8  
  optimal library   mpi  mpl  mpl  mpl 
  (mpi-min)/min     0.000    0.049    0.053    0.003 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:13:04 EDT.