PSTSWM T3E-900 Point-to-Point Communication Performance

Performance Studies using

PSTSWM

SGI/Cray Research T3E-900 SWAP Performance

(unordered swap of 8KB message using SHMEM)

Date/Person: January 30, 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
PrgEnv 3
Cray CF90 Version 3.0.1.3
Communication Library: SHMEM
SWAP size: 1024 REAL*8 floating point values each direction
Message size: Largest - 1024 REAL*8 floating point values
Smallest - 1 REAL*8 floating point value
Processors: 0 and 1
Latency Definition:(T1024-T512)/512
Results:

unordered swap using nonblocking send
Data Statistics
bidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 258.44 6.92 15.7%
10 iter. 345.70 6.67 28.1%
1 iter. w/overlap 245.76 6.67 11.9%
10 iter. w/overlap 350.97 6.21 26.6%

unordered swap using nonblocking receive
Data Statistics
bidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 264.60 6.64 10.7%
10 iter. 314.59 6.42 25.3%
1 iter. w/overlap 268.45 6.77 11.1%
10 iter. w/overlap 332.61 5.80 24.5%

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:28:17 EDT.
86869 accesses since 1/2/96.