Difference between revisions of "Emulex OneConnect OCe14102-NM"

From Define Wiki
Jump to navigation Jump to search
Line 79: Line 79:
 
</syntaxhighlight>
 
</syntaxhighlight>
  
'''osu_bw'''
+
'''osu_bibw'''
  
 
Command Line Run:
 
Command Line Run:
 
<syntaxhighlight>
 
<syntaxhighlight>
#
+
[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_bibw
 
</syntaxhighlight>
 
</syntaxhighlight>
  
 
*Output
 
*Output
 
<syntaxhighlight>
 
<syntaxhighlight>
#
+
 
 +
# OSU MPI Bi-Directional Bandwidth Test v4.4.1
 +
# Size    Bi-Bandwidth (MB/s)
 +
1                      0.64
 +
2                      1.29
 +
4                      2.59
 +
8                      5.17
 +
16                    10.38
 +
32                    20.24
 +
64                    39.61
 +
128                    79.93
 +
256                  153.93
 +
512                  296.42
 +
1024                  349.68
 +
2048                  373.45
 +
4096                  413.55
 +
8192                  423.65
 +
16384                413.16
 +
32768                423.40
 +
65536                234.21
 +
131072                235.67
 +
262144                247.31
 +
524288                236.90
 +
1048576              242.36
 +
2097152              237.07
 +
4194304              237.14
 
</syntaxhighlight>
 
</syntaxhighlight>
  
'''osu_bibw'''
+
'''osu_bw'''
  
  
Line 130: Line 155:
 
</syntaxhighlight>
 
</syntaxhighlight>
  
'''osu_mr_mbr'''
+
'''osu_latency'''
  
 
Command Line Run:
 
Command Line Run:
Line 167: Line 192:
 
4194304            18007.57
 
4194304            18007.57
  
 +
</syntaxhighlight>
 +
 +
'''osu_mbr_mr'''
 +
 +
Command Line Run:
 +
<syntaxhighlight>
 +
[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_mbr_mr
 +
</syntaxhighlight>
 +
 +
*Output
 +
<syntaxhighlight>
 +
 +
# OSU MPI Multiple Bandwidth / Message Rate Test v4.4.1
 +
# [ pairs: 1 ] [ window size: 64 ]
 +
# Size                  MB/s        Messages/s
 +
1                      0.33        325917.53
 +
2                      0.65        326412.92
 +
4                      1.31        326714.85
 +
8                      2.60        325151.66
 +
16                      5.23        326563.82
 +
32                      9.79        305954.68
 +
64                    19.61        306468.15
 +
128                    39.19        306191.99
 +
256                    79.89        312087.08
 +
512                  147.86        288783.11
 +
1024                  175.93        171803.09
 +
2048                  198.99          97163.82
 +
4096                  206.89          50509.06
 +
8192                  218.85          26714.51
 +
16384                231.62          14136.78
 +
32768                233.28          7119.02
 +
65536                231.89          3538.37
 +
131072                234.39          1788.27
 +
262144                235.76            899.37
 +
524288                236.60            451.27
 +
1048576              236.96            225.98
 +
2097152              237.13            113.07
 +
4194304              237.23            56.56
 
</syntaxhighlight>
 
</syntaxhighlight>

Revision as of 07:23, 4 August 2015

System Configuration

  • Drivers: elx-be2net-dd-rhel6-10.6.144.21-1.tar
  • OS: CentOS 6.6 - Default Kernel

All steps in "10G NICs General Optimization" Followed

Systems

hft1 hft2

  • Cards: OCE14102-NM-NFR

Tests

  • SockPerf
  • iPerf
  • OSU MPI Benchmarks - bw, bibw, latency, mr_mbw

Benchmarks

SockPerf

Command Line Run:

[root@hft1 ~]# taskset -c 8 sockperf sr
[root@hft2 ~]# taskset -c 8 sockperf pp -i 10.10.10.2 -t 5 -m 12
  • Output
[root@hft2 elx-rhel6-be2net-dd-10.6.144.21-1]# taskset -c 8 sockperf pp -i 10.10.10.2 -t 5 -m 12
sockperf: == version #2.5.244 ==
sockperf[CLIENT] send on:sockperf: using recvfrom() to block on socket(s)

[ 0] IP = 10.10.10.1      PORT = 11111 # UDP
sockperf: Warmup stage (sending a few dummy messages)...
sockperf: Starting test...
sockperf: Test end (interrupted by timer)
sockperf: Test ended
sockperf: [Total Run] RunTime=5.100 sec; SentMessages=333984; ReceivedMessages=333983
sockperf: ========= Printing statistics for Server No: 0
sockperf: [Valid Duration] RunTime=5.000 sec; SentMessages=327436; ReceivedMessages=327436
sockperf: ====> avg-lat=  7.616 (std-dev=0.060)
sockperf: # dropped messages = 0; # duplicated messages = 0; # out-of-order messages = 0
sockperf: Summary: Latency is 7.616 usec
sockperf: Total 327436 observations; each percentile contains 3274.36 observations
sockperf: ---> <MAX> observation =    9.900
sockperf: ---> percentile  99.99 =    8.233
sockperf: ---> percentile  99.90 =    7.957
sockperf: ---> percentile  99.50 =    7.847
sockperf: ---> percentile  99.00 =    7.815
sockperf: ---> percentile  95.00 =    7.729
sockperf: ---> percentile  90.00 =    7.687
sockperf: ---> percentile  75.00 =    7.640
sockperf: ---> percentile  50.00 =    7.606
sockperf: ---> percentile  25.00 =    7.578
sockperf: ---> <MIN> observation =    7.337

iPerf

Command Line Run:

[root@hft1 ~]# taskset -c 8,9 iperf -s
[root@hft2 ~]# taskset -c 8,9 iperf -c 10.10.10.1
  • Output
------------------------------------------------------------
Client connecting to 10.10.10.1, TCP port 5001
TCP window size: 9.54 MByte (default)
------------------------------------------------------------
[  3] local 10.10.10.2 port 30403 connected with 10.10.10.1 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  10.6 GBytes  9.10 Gbits/sec

osu_bibw

Command Line Run:

[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_bibw
  • Output
# OSU MPI Bi-Directional Bandwidth Test v4.4.1
# Size    Bi-Bandwidth (MB/s)
1                       0.64
2                       1.29
4                       2.59
8                       5.17
16                     10.38
32                     20.24
64                     39.61
128                    79.93
256                   153.93
512                   296.42
1024                  349.68
2048                  373.45
4096                  413.55
8192                  423.65
16384                 413.16
32768                 423.40
65536                 234.21
131072                235.67
262144                247.31
524288                236.90
1048576               242.36
2097152               237.07
4194304               237.14

osu_bw


  • Pinning Cores to the run made little to no difference for OSU

Command Line Run:

[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_bw
  • Output
# OSU MPI Bandwidth Test v4.4.1
# Size      Bandwidth (MB/s)
1                       0.30
2                       0.61
4                       1.23
8                       2.48
16                      4.97
32                      9.77
64                     18.96
128                    38.91
256                    77.56
512                   146.62
1024                  175.88
2048                  198.64
4096                  206.83
8192                  219.13
16384                 231.53
32768                 233.15
65536                 231.67
131072                234.22
262144                235.75
524288                236.57
1048576               236.94
2097152               237.13
4194304               237.23

osu_latency

Command Line Run:

[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_latency
  • Output
# OSU MPI Latency Test v4.4.1
# Size          Latency (us)
0                      12.26
1                      12.27
2                      12.27
4                      12.27
8                      12.28
16                     12.28
32                     12.29
64                     12.50
128                    13.16
256                    14.43
512                    16.94
1024                   23.27
2048                   59.06
4096                  112.98
8192                  115.23
16384                 137.94
32768                 213.93
65536                 402.38
131072                910.54
262144               1490.14
524288               2450.02
1048576              4761.29
2097152              9175.97
4194304             18007.57

osu_mbr_mr

Command Line Run:

[root@hft1 pt2pt]# mpirun -H 10.10.10.1,10.10.10.2 -np 2 --allow-run-as-root ./osu_mbr_mr
  • Output
# OSU MPI Multiple Bandwidth / Message Rate Test v4.4.1
# [ pairs: 1 ] [ window size: 64 ]
# Size                  MB/s        Messages/s
1                       0.33         325917.53
2                       0.65         326412.92
4                       1.31         326714.85
8                       2.60         325151.66
16                      5.23         326563.82
32                      9.79         305954.68
64                     19.61         306468.15
128                    39.19         306191.99
256                    79.89         312087.08
512                   147.86         288783.11
1024                  175.93         171803.09
2048                  198.99          97163.82
4096                  206.89          50509.06
8192                  218.85          26714.51
16384                 231.62          14136.78
32768                 233.28           7119.02
65536                 231.89           3538.37
131072                234.39           1788.27
262144                235.76            899.37
524288                236.60            451.27
1048576               236.96            225.98
2097152               237.13            113.07
4194304               237.23             56.56