PSTSWM SP3-375 Communication Protocol Performance

Performance Studies using

PSTSWM

IBM SP3-375 Winterhawk II Protocol Performance

(distributed LT experiment C / ring-pipeline sum algorithm)

Date/Person: December 12, 1999 / P. Worley
Platform: IBM SP3 at Oak Ridge National Laboratory (bobcat.ccs.ornl.gov):
     62 4-way Winterhawk II SMP nodes (375 MHz POWER3 with 8MB L2 cache)
Environment: AIX 4.3.3;   PSSP 3.1.1
Code Version: 6.6.4
Compilation Options: mpxlf -O3 -qarch=auto -qtune=auto -qcache=auto
Link Options: -bmaxdata:0x70000000
Environment Variables: export MP_SHARED_MEMORY="yes"
Math Library: ESSL
Communication Library: MPI over the switch using user space for internode communication
MPI using shared memory for intranode communication
Parallel Algorithm: ringpipe
Number of Timesteps: 12
Results:

4x16 Processors (Column Major Ordering) / Problem T42L2
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  1.1892e-01   0.26   0.28   0.57 
Three Fastest
Protocols
1st2nd3rd
  i4   i5   g4 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  1   4   18 

2x32 Processors (Column Major Ordering) / Problem T42L2
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  9.7336e-02   0.58   0.63   1.37 
Three Fastest
Protocols
1st2nd3rd
  i4   i5   f4 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  1   3   8 

8x8 Processors (Column Major Ordering) / Problem T42L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  4.3866e-01   0.06   0.07   0.13 
Three Fastest
Protocols
1st2nd3rd
  i4   f4   f5 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  2   13   50 

4x16 Processors (Column Major Ordering) / Problem T85L4
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  3.6539e-01   0.14   0.16   0.27 
Three Fastest
Protocols
1st2nd3rd
  i5   f5   i4 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   8   49 

2x32 Processors (Column Major Ordering) / Problem T85L4
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  3.1069e-01   0.32   0.37   0.65 
Three Fastest
Protocols
1st2nd3rd
  f4   i5   i4 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   4   14 

8x8 Processors (Column Major Ordering) / Problem T85L32
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  2.8228e+00   0.03   0.02   0.09 
Three Fastest
Protocols
1st2nd3rd
  a2   c4   f3 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  6   40   50 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:19:32 EDT.