PSTSWM T3E-900 Communication Protocol Performance

Performance Studies using

PSTSWM

SGI/Cray Research T3E-900 Protocol Performance

(transpose LT experiment I-B / O(log P) transpose algorithm)

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
Parallel Algorithm: logtrans
(assuming distributed FFT)
Number of Timesteps: 12
Results:

4x16 Processors (Row Major Ordering) / Problem T21L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.3797e-01   0.08   0.08   0.16 
Three Fastest
Protocols
1st2nd3rd
  a6   a2   a1 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  1   8   25 

2x32 Processors (Row Major Ordering) / Problem T21L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.8096e-01   0.09   0.10   0.16 
Three Fastest
Protocols
1st2nd3rd
  c0   c6   c3 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  2   4   14 

8x8 Processors (Row Major Ordering) / Problem T42L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  4.4050e-01   0.04   0.05   0.08 
Three Fastest
Protocols
1st2nd3rd
  a0   a1   c1 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   13   25 

4x16 Processors (Row Major Ordering) / Problem T42L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  4.0930e-01   0.05   0.07   0.09 
Three Fastest
Protocols
1st2nd3rd
  a0   a6   c2 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  5   12   25 

2x32 Processors (Row Major Ordering) / Problem T42L32
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  6.9075e-01   0.06   0.10   0.11 
Three Fastest
Protocols
1st2nd3rd
  a6   a0   a1 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  4   12   25 

8x8 Processors (Row Major Ordering) / Problem T85L32
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  3.7525e+00   0.02   0.03   0.03 
Three Fastest
Protocols
1st2nd3rd
  a0   a6   c1 
      Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  12   25   25 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:24:58 EDT.