PSTSWM Paragon Point-to-Point Communication Performance

Performance Studies using

PSTSWM


Intel Paragon SWAP Performance

(ordered swap of 128KB message using NX)

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: NX
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. 113.52 61.72 21.4%
10 iter. 115.00 60.81 22.6%

ordered swap using nonblocking send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 112.07 69.49 19.4%
10 iter. 113.51 68.74 20.8%
1 iter. w/overlap 111.39 79.62 21.9%
10 iter. w/overlap 112.94 76.12 20.5%

ordered swap using nonblocking receive
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 112.06 68.53 18.8%
10 iter. 113.72 68.36 21.2%
1 iter. w/overlap 112.61 77.51 22.4%
10 iter. w/overlap 114.46 73.68 21.2%

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. 111.82 68.68 21.7%
10 iter. 113.63 68.18 21.2%
1 iter. w/overlap 111.12 88.83 19.0%
10 iter. w/overlap 113.14 83.11 19.6%

ordered swap using ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 111.31 98.98 15.1%
10 iter. 109.66 97.81 17.0%
1 iter. w/overlap 111.88 77.21 21.8%
10 iter. w/overlap 114.51 73.95 20.7%

ordered swap using nonblocking ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 109.79 99.65 15.2%
10 iter. 109.60 97.56 17.2%
1 iter. w/overlap 111.85 89.28 19.4%
10 iter. w/overlap 113.12 83.72 18.9%

synchronous
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 110.89 85.06 18.8%
10 iter. 111.82 84.07 24.7%


Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:29:18 EDT.
86964 accesses since 1/2/96.