PSTSWM Paragon Point-to-Point Communication Performance

Performance Studies using

PSTSWM


Intel Paragon SWAP Performance

(ordered swap of 2MB message using MPI)

Date/Person: January 30, 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: 262144 REAL*8 floating point values each direction
Message size: Largest - 262144 REAL*8 floating point values
Smallest - 256 REAL*8 floating point values
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. 117.76 74.59 6.7%
10 iter. 117.73 74.02 6.4%

ordered swap using nonblocking send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 117.00 92.75 4.4%
10 iter. 117.49 91.56 6.4%
1 iter. w/overlap 116.76 108.38 3.8%
10 iter. w/overlap 117.00 99.31 6.1%

ordered swap using nonblocking receive
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.63 93.39 4.9%
10 iter. 116.93 93.69 8.1%
1 iter. w/overlap 116.86 105.24 4.8%
10 iter. w/overlap 117.38 95.67 7.2%

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. 116.47 93.80 6.0%
10 iter. 117.10 93.48 7.7%
1 iter. w/overlap 116.41 138.02 3.9%
10 iter. w/overlap 116.89 120.88 5.8%

ordered swap using ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.10 156.83 18.3%
10 iter. 116.37 155.45 20.2%
1 iter. w/overlap 116.68 110.09 5.4%
10 iter. w/overlap 117.21 99.23 7.1%

ordered swap using nonblocking ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.14 158.96 18.9%
10 iter. 116.24 155.78 21.1%
1 iter. w/overlap 116.57 141.67 4.3%
10 iter. w/overlap 117.01 125.95 6.8%

synchronous
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.92 107.80 10.8%
10 iter. 117.49 104.37 11.0%


Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:29:15 EDT.
86508 accesses since 1/2/96.