pps (a tool developed by P. Saab and previously used and improved by K. Kennaway) blasting UDP packets in parallel threads is testing UDP send performance only. We run it to a local disc(4) (discard interface) for these tests, no even sending packets on the wire. We ran 3 reboot iterations, with each iteration having 1 warm up run and 10 recorded pps runs of 120 seconds.
On IPv6 UDP checksums are mandatory, so we did not turn them off for comparisons.
x hydra1/test_pps_udp4_local_discard.stat
+ hydra1/test_pps_udp4_local_discard_nocksum.stat
* hydra1/test_pps_udp6_local_discard.stat
+--------------------------------------------------------------------------+
| * * |
| *x x + * * |
| *x x + * ** |
| *x x ++ ++ * ** |
| **x x ++ ++ * * ** |
| x **x+ x ++ ++ * * * ****|
|xx *x **x+ x ++ ++ * * * ******|
| |__A__| |
| |____AM___| |
| |___A_M__||
+--------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 30 435357.61 446991.56 443735.63 443343.29 3028.2886
+ 30 439229.27 453317.25 448703.37 447586.43 4209.5686
Difference at 95.0% confidence
4243.14 +/- 1895.43
0.957079% +/- 0.42753%
(Student's t, pooled s = 3666.81)
* 30 487131.28 499788.75 497556.52 495750.29 3904.706
Difference at 95.0% confidence
52407 +/- 1806.14
11.8209% +/- 0.407391%
(Student's t, pooled s = 3494.09)