PSTSWM SP3-200 Communication Protocol Performance

Performance Studies using

PSTSWM

IBM SP3-200 Protocol Performance

(transpose LT experiment II-B2 / O(P) sendrecv transpose algorithm)

Date/Person: June 6, 1999 / P. Worley
Platform: IBM SP3 at Oak Ridge National Laboratory (morgan.ccs.ornl.gov):
     62 2-way Winterhawk I SMP nodes (200 MHz POWER3 with 4MB L2 cache, equivalent to RS/6000 Model 260)
Environment: AIX 4.3.2;   PSSP 3.1
Code Version: 6.5
Compilation Options: mpxlf -O3 -qarch=auto -qtune=auto -qcache=auto
Link Options: -bmaxdata:0x70000000
Math Library: ESSL
Communication Library: MPI over the switch using user space for internode communication
MPI using IP for intranode communication
Parallel Algorithm: srtrans
(assuming transpose FFT)
Number of Timesteps: 12
Results:

8x8 Processors (Row Major Ordering) / Problem T42L16
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  3.4800e-01   0.10   0.09   0.23 
Three Fastest
Protocols
1st2nd3rd
  e4   e3   a4 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  3   9   31 

8x8 Processors (Row Major Ordering) / Problem T85L32
Runtime Statistics
min(mean-min)/min(median-min)/min(max-min)/min
  2.6511e+00   0.09   0.07   0.70 
Three Fastest
Protocols
1st2nd3rd
  a1   a4   a5 
       Number of Proctocols With
Runtimes Within X% of Min
1%5%25%
  1   10   30 

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:16:59 EDT.