PSTSWM Paragon Point-to-Point Communication Performance

Performance Studies using

PSTSWM


Intel Paragon SWAP Performance

(ordered swap of 2MB message using NX)

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: NX
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. 117.68 50.37 9.8%
10 iter. 117.37 50.41 10.6%

ordered swap using nonblocking send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 117.58 58.00 7.7%
10 iter. 117.37 58.65 9.0%
1 iter. w/overlap 117.68 66.03 7.5%
10 iter. w/overlap 117.06 64.08 9.1%

ordered swap using nonblocking receive
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 117.10 57.09 8.4%
10 iter. 117.31 58.16 9.9%
1 iter. w/overlap 117.72 65.05 8.7%
10 iter. w/overlap 117.18 63.04 10.0%

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. 117.35 58.51 10.3%
10 iter. 116.92 57.87 10.0%
1 iter. w/overlap 117.72 75.52 7.1%
10 iter. w/overlap 117.09 72.41 9.6%

ordered swap using ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.78 88.69 4.9%
10 iter. 116.55 87.91 4.9%
1 iter. w/overlap 117.97 65.03 8.1%
10 iter. w/overlap 117.22 63.21 10.1%

ordered swap using nonblocking ready send
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 116.72 88.52 4.5%
10 iter. 116.60 88.58 5.0%
1 iter. w/overlap 117.26 76.86 7.9%
10 iter. w/overlap 116.93 72.59 9.0%

synchronous
Data Statistics
unidirectional bandwidth estimated latency model error
(peak MByte/sec) (usec/msg) (max. rel. error)
1 iter. 117.55 81.05 11.1%
10 iter. 117.19 84.71 21.4%


Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:29:15 EDT.
86941 accesses since 1/2/96.