2011-12-15 p4 @203041 UDP

Test runs

netperf-2.5 UDP_STREAM test was run over 3 reboot iterations, with each iteration having 1 warm up run and 10 recorded netperf runs.
netperf needed a patch to be able to run this test on IPv6.

On IPv6 UDP checksums are mandatory, so we did not turn them off for comparisons.

IPv4/UDP with and without offloading turned on, with and without checksumming turned on, and IPv6/UDP with and without offloading turned on

As we expect the IPv4 tests with checksumming turned off perform best. However IPv6 UDP is behind IPv4 UDP with checksumming turned on and zooming in gives us a clearer view.

IPv4/UDP w/ and w/o offloading, w/ and w/o checksumming,
IPv6/UDP numbers w/ and w/o offloading on

IPv4/UDP w/ and w/o offloading, w/ and w/o checksumming,
IPv6/UDP numbers w/ and w/o offloading on (zoomed in)

x hydra1/test_netperf_udp4_default_nocksum.stat
+ hydra1/test_netperf_udp4_nooffload_nocksum.stat
* hydra1/test_netperf_udp4_default.stat
% hydra1/test_netperf_udp4_nooffload.stat
# hydra1/test_netperf_udp6_default.stat
@ hydra1/test_netperf_udp6_nooffload.stat
+--------------------------------------------------------------------------+
|     O                                                                    |
|     O                                                                    |
|     O                                                   %*               |
|     O                                                   %*            +  |
|     O@                                                  %*            +  |
|     O@                                                  %*            +  |
|     OO                                                  %*            +  |
|     OO                                                  %**           +  |
|     OO                                                  %**          ++  |
|     OO                                                  %O*      x  x+*x |
|     OO                                                  %O*      x  **** |
|   ##OO                                               %OO%O*      xx **** |
|   #OOO                                             %%%OOOO*      *x **** |
|O@OOOOO                                             O%%OOOO*    *x*x+*****|
|                                                                   |_AM_| |
|                                                                    |_AM| |
|                                                        |_A|              |
|                                                      |_AM|               |
|   |_A|                                                                   |
|   |_A|                                                                   |
+--------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x  30        5318.7       5554.02          5470     5455.0687     69.018149
+  30       5301.76       5562.65       5497.15     5479.1397     54.711816
No difference proven at 95.0% confidence
*  30       4976.59       5174.88       5143.77     5126.1407     43.975246
Difference at 95.0% confidence
	-328.928 +/- 29.9126
	-6.02977% +/- 0.548345%
	(Student's t, pooled s = 57.8676)
%  30       4957.97        5147.2       5103.32      5083.228     50.926167
Difference at 95.0% confidence
	-371.841 +/- 31.3511
	-6.81643% +/- 0.574716%
	(Student's t, pooled s = 60.6506)
#  30       3502.51       3678.12       3648.75     3635.0957     38.189084
Difference at 95.0% confidence
	-1819.97 +/- 28.8314
	-33.363% +/- 0.528524%
	(Student's t, pooled s = 55.7759)
@  30       3509.09       3670.67       3650.99      3637.008     42.203557
Difference at 95.0% confidence
	-1818.06 +/- 29.5697
	-33.3279% +/- 0.542059%
	(Student's t, pooled s = 57.2042)