PSTSWM Paragon Communication Protocol Performance

Performance Studies using

PSTSWM

Intel Paragon Protocol Performance

(transpose FFT experiment A1 / O(P) sendrecv transpose algorithm)

Date/Person: May 13, 1998 / P. Worley
Platform: Intel Paragon XP/S 150 MP at Oak Ridge National Laboratory:
     1024 MP nodes (3 50-MHz iPSC/860 processors per node)
Environment: Paragon OSF/1 Release 1.0.4 Server 1.4 R1_4_5
f77/Paragon Paragon Version R5.0.3
Code Version: 6.3
Compilation Options: if77 -O4 -Mnodepchk -Knoieee -Msafealloc
Math Library: KAI
Communication Library: MPI
Parallel Algorithm: srtrans
Partition: 1x8, 1x16, or 1x32
Results:

16x1 Processors / Problem T10L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  5.1916e-01   0.37   0.21   1.14 
Three Fastest
Protocols
1st2nd3rd
  a1   a3   a2 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   6   13 

32x1 Processors / Problem T10L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  7.7380e-01   0.43   0.38   1.14 
Three Fastest
Protocols
1st2nd3rd
  a1   a2   a3 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  4   7   11 

8x1 Processors / Problem T21L8
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.1602e+00   0.12   0.13   0.27 
Three Fastest
Protocols
1st2nd3rd
  a3   a2   b3 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  6   10   22 

16x1 Processors / Problem T21L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.3592e+00   0.22   0.26   0.54 
Three Fastest
Protocols
1st2nd3rd
  a3   b0   a1 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  5   10   12 

32x1 Processors / Problem T21L32
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.8301e+00   0.71   0.71   1.69 
Three Fastest
Protocols
1st2nd3rd
  b0   b2   b3 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  4   8   10 

8x1 Processors / Problem T42L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.0641e+01   0.04   0.03   0.24 
Three Fastest
Protocols
1st2nd3rd
  b1   b0   b2 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   17   25 

DISCUSSION

Partitions of the Paragon processor grid were used that match the processor subsets that the parallel algorithms would run on in a two dimensional data decomposition: 1x8, 1x16, and 1x32.
Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:30:07 EDT.