PSTSWM Paragon Point-to-Point Communication Performance

Performance Studies using

PSTSWM

Intel Paragon SWAP Performance

(ordered swap of 128KB message using MPI)

Date/Person: February 2, 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
Communication Library: MPI
SWAP size: 16384 REAL*8 floating point values each direction
Message size: Largest - 16384 REAL*8 floating point values
Smallest - 16 REAL*8 floating point value
Processors: 0 and 1
Latency Definition:(T1024-T512)/512
Results:

ordered simple swap
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 106.70 97.22 27.4%
10 iter. 111.65 85.57 17.6%

ordered swap using nonblocking send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 104.66 105.32 13.7%
10 iter. 108.60 103.67 17.8%
1 iter. w/overlap 104.53 124.62 16.6%
10 iter. w/overlap 107.66 113.16 16.6%

ordered swap using nonblocking receive
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 106.37 105.42 15.7%
10 iter. 108.74 103.74 18.1%
1 iter. w/overlap 108.48 119.02 15.2%
10 iter. w/overlap 109.85 107.14 17.8%

ordered swap using nonblocking send and receive
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 107.33 105.42 13.0%
10 iter. 109.31 104.81 17.5%
1 iter. w/overlap 102.75 152.81 15.1%
10 iter. w/overlap 106.94 134.88 14.9%

ordered swap using ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 102.13 152.24 16.0%
10 iter. 103.77 151.49 19.4%
1 iter. w/overlap 107.52 121.99 13.1%
10 iter. w/overlap 110.12 110.53 17.5%

ordered swap using nonblocking ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 103.35 154.66 19.9%
10 iter. 105.00 151.14 20.4%
1 iter. w/overlap 103.98 159.85 12.7%
10 iter. w/overlap 105.86 138.86 15.7%

synchronous
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 104.88 117.67 22.4%
10 iter. 108.41 115.87 21.9%

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:29:17 EDT.
85760 accesses since 1/2/96.