3.4.1. L2 Ethernet Switching

This section provides a summary of Testpmd Phy-to-Phy L2 Ethernet looping performance illustrating packet latency measured at 50% of discovered NDR throughput rate. Latency is reported for Testpmd running in multiple configurations of Testpmd worker thread(s), a.k.a. Testpmd data plane thread (s), and their physical CPU core(s) placement.

Title of each graph is a regex (regular expression) matching all plotted latency test cases, X-axis labels are indeces of csit-dpdk-perf-1701 jobs that created result output files used as data sources for the graph, Y-axis labels are measured packet Latency [uSec] values, and the graph legend identifes the plotted test suites.

Note

Data sources for reported test results: i) FD.io test executor jobs csit-dpdk-perf-1701-all and csit-dpdk-perf-1701-long , ii) archived FD.io jobs test result output files csit-dpdk-perf-1701-all#job-number.

Testpmd packet latency - running in configuration of one worker thread (1t) on one physical core (1c) - is presented in the figure below.

Figure 1. Testpmd 1thread 1core - packet latency for Phy-to-Phy L2 Ethernet Looping.

CSIT test cases used to generate results presented above can be found in CSIT git repository by filtering with specified regex as follows:

$ csit/tests/perf
$ grep -E "64B-1t1c-(eth|dot1q|dot1ad)-(l2xcbase|l2bdbasemaclrn)-ndrdisc" *

10ge2p1x520-eth-l2xcbase-ndrdisc.robot:| tc01-64B-1t1c-eth-l2xcbase-ndrdisc

Testpmd packet latency - running in configuration of two worker threads (2t) on two physical cores (2c) - is presented in the figure below.

Figure 2. Testpmd 2thread 2core - packet latency for Phy-to-Phy L2 Ethernet Looping.

CSIT test cases used to generate results presented above can be found in CSIT git repository by filtering with specified regex as follows:

$ csit/tests/perf
$ grep -E "64B-2t2c-(eth|dot1q|dot1ad)-(l2xcbase|l2bdbasemaclrn)-ndrdisc" *

10ge2p1x520-eth-l2xcbase-ndrdisc.robot:| tc07-64B-2t2c-eth-l2xcbase-ndrdisc

Testpmd packet latency - running in configuration of four worker threads (4t) on four physical cores (4c) - is presented in the figure below.

Figure 3. Testpmd 4thread 4core - packet latency for Phy-to-Phy L2 Ethernet Looping.

CSIT test cases used to generate results presented above can be found in CSIT git repository by filtering with specified regex as follows:

$ csit/tests/perf
$ grep -E "64B-4t4c-(eth|dot1q|dot1ad)-(l2xcbase|l2bdbasemaclrn)-ndrdisc" *

10ge2p1x520-eth-l2xcbase-ndrdisc.robot:| tc13-64B-4t4c-eth-l2xcbase-ndrdisc