Hi All, Quick update re recent major additions to CSIT-2009 report available at https://docs.fd.io/csit/rls2009/report/ :
1. Benchmark results for AMD 2n-zn2 testbed (EPYC Zen2 7532) with VPP and DPDK performance data: - Testbed description in [12] - Sample throughput graphs in [13] - Sample latency graphs in [14] - Sample benchmark tables in [15] - Sample VPP operational data (“show runtime”) in [16] 2. New NAT44 Endpoint Dependent stateful benchmarks - Methodology for Connections-Per-Second (CPS) and Packets-Per-Second (PPS) for UDP and TCP/IP [17] - Sample benchmark graphs UDP CPS [18], TCP CPS [19], UDP PPS [20], TCP PPS [21] Happy browsing! :) Cheers, Maciek [12] https://docs.fd.io/csit/rls2009/report/introduction/physical_testbeds.html#node-amd-epyc-zen2-2n-zn2 [13] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/ip4-2n-zn2-xxv710.html [14] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_latency/ip4-2n-zn2-xxv710.html [15] https://docs.fd.io/csit/rls2009/report/detailed_test_results/vpp_performance_results_2n_zn2/ip4_xxv710.html [16] https://docs.fd.io/csit/rls2009/report/test_operational_data/vpp_performance_operational_data_2n_zn2/ip4_xxv710.html [17] https://docs.fd.io/csit/rls2009/report/introduction/methodology_nat44.html#nat44-endpoint-dependent [18] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/nat44-ed-udp-cps.html [19] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/nat44-ed-tcp-cps.html [20] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/nat44-ed-udp-pps.html [21] https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/nat44-ed-tcp-pps.html > On 14 Oct 2020, at 21:58, Maciek Konstantynowicz (mkonstan) > <mkons...@cisco.com> wrote: > > Hi All, > > FD.io CSIT-2009 report is available on FD.io docs site: > > https://docs.fd.io/csit/rls2009/report/ > > Great thanks to all contributors in CSIT and VPP communities! > > Below summary and pointers to specific sections in the report. > Welcome all comments, best by email to csit-...@lists.fd.io. > > Cheers, > -Maciek > > > CSIT-2009 Release Summary > ------------------------- > > NEW TESTS > > - A new category of tests using TRex ASTF stateful APIs and traffic > profiles with up to 16M UDP and TCP/IP sessions. Initial stateful > tests include VPP NAT44 Endpoint Dependent (NAT44ed) > connections-per-second and packets-per-second throughput (with > controlled packet size). (Note: report test runs are still to be > executed in their fullness, expect them to appear in maintenance > report versions next week and week after next. Maintenance reports are > published on a weekly basis if there are changes.) > > - Refactored existing NAT44 Deterministic (NAT44det) throughput tests > and added higher session scale, up to 16M UDP sessions. Continue to > use TRex STL stateless APIs and traffic profiles. > > - Added NAT44ed uni-directional UDP throughput tests using TRex STL > stateless APIs and traffic profiles, as a way to verify stateful tests > performance. > > - IPsec async mode VPP performance tests, with HW crypto only for now, > meaning Xeon Haswell testbeds only. > > - Full suite of tests now running on Mellanox ConnectX5-2p100GE NICs in > 2n-clx (Intel Xeon Cascadelake) testbeds using VPP native rdma driver. > For the first time one case see linear multi-core speedup into 72 Mpps > region (L2 on 2 cores, IPv4 on 4 cores), in some cases NIC is the > limit again (like it was to date in CSIT labs with FVL 2p25GE NICs). > > BENCHMARKING > > - AMD 2n-zn2 testbed onboarded with EPYC 7532 32-Core Processor. Full > set of CSIT-2009 results to be included in one of the upcoming > maintenance reports, following completion of calibrating dry runs that > are currently ongoing. > > - Optimization and calibration of TRex STL and ASTF multi-core > configurations, a small impact on test results as captured in current > vs. previous release performance comparisons. > > > Pointers to CSIT-2009 Report sections > ------------------------------------- > > 1. FD.io CSIT test methodology [1] > 2. VPP release notes [2] > 3. VPP 64B/IMIX throughput graphs [3] > 4. VPP throughput speedup multi-core [4] > 5. VPP latency under load [5] > 6. VPP comparisons v20.09 vs. v20.05 [6] > 7. VPP performance all pkt sizes & NICs [7] > 8. DPDK 20.08 apps release notes [8] > 9. DPDK 64B throughput graphs [9] > 10. DPDK latency under load [10] > 11. DPDK comparisons 20.08 vs. 20.02 [11] > > Functional device tests (VPP_Device) are also included in the report. > > [1] https://docs.fd.io/csit/rls2009/report/introduction/methodology.html > [2] > https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/csit_release_notes.html > [3] > https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_throughput_graphs/index.html > [4] > https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/throughput_speedup_multi_core/index.html > [5] > https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/packet_latency/index.html > [6] > https://docs.fd.io/csit/rls2009/report/vpp_performance_tests/comparisons/current_vs_previous_release.html > [7] > https://docs.fd.io/csit/rls2009/report/detailed_test_results/vpp_performance_results/index.html > [8] > https://docs.fd.io/csit/rls2009/report/dpdk_performance_tests/csit_release_notes.html > [9] > https://docs.fd.io/csit/rls2009/report/dpdk_performance_tests/packet_throughput_graphs/index.html > [10] > https://docs.fd.io/csit/rls2009/report/dpdk_performance_tests/packet_latency/index.html > [11] > https://docs.fd.io/csit/rls2009/report/dpdk_performance_tests/comparisons/current_vs_previous_release.html >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#18419): https://lists.fd.io/g/vpp-dev/message/18419 Mute This Topic: https://lists.fd.io/mt/79184087/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
