15:05:09 #startmeeting FD.io CSIT Meeting 15:05:09 Meeting started Wed Nov 1 15:05:09 2023 UTC and is due to finish in 60 minutes. The chair is mackonstan. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:05:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:05:09 The meeting name has been set to 'fd_io_csit_meeting' 15:06:54 #info Maciek Konstantynowicz 15:06:59 #info Kai Ji 15:07:21 #info Low attendance due to public holiday in many countries (All Saints). 15:07:31 #info Will keep this meeting focused and short. 15:08:12 #topic Agenda bashing 15:08:29 #info Due to low attendance will only cover updates from Kai and Maciek 15:08:56 #topic Projects update - TSC 15:11:20 #info Maciek in last TSC meeting reviewed proposed FD.io CSIT lab extension proposal that included following SUT server HW: i) 4 of EMR TG TRex servers based on Intel Xeon Emerald Rapids build, ii) 1 of 2 node Intel SierraForest E-Core, iii) 2 of 2-Node performance testbeds based on Marvell Octeon 10 DPU (Arm Neoverse N2, HW VPP), iv) 2 of 2-Node-GraceGrace performance testbeds based on Nvidia Grace-Grace processors, v) 2 of 2-Nod 15:11:20 e-Zn4 performance testbeds based on AMD Zen4 processors, vi) 1 Rack with power 15:11:48 #info v) 2 of 2-Nod e-Zn4 performance testbeds based on AMD Zen4 processors, vi) 1 Rack with power 15:12:32 #topic Intel HW accelerator 15:13:19 #info SPR QAT2.0 - Kai - need Peter to confirm that provided SPR patch is working. It has been tested in Intel labs in Shannon, and working fine. 15:15:30 #info SNR QAT2.0 - Kai - replacement systems got shipped. Need to check with Vexxhost. 15:16:29 #info SPR DSA - DSA - SuperMicro has finally published the BIOS v1.4, see link below. We will upgrade all SPR systems after CSIT-2310 rls testing cycle is completed. 15:17:08 #link [Motherboard: Super X13DEG-QT-P](https://www.supermicro.com/en/support/resources/downloadcenter/firmware/MBD-X13DEG-QT/BIOS) 15:24:12 #info E810 GTPU offload - Kai - reviewed the results from CSIT, compared to internal, and CSIT 1c results are better. But CSIT 2c and 4c results do not show expected multi-core speedup for neither gtpuhw nor gtpusw, see link below. In Intel labs they don't either, while running the same CSIT tests. This is likely due to small number of flows used in baseline tests. May need to add scale tests and/or to look into test definitions a 15:24:12 nd test code to make sure RSS kicks in. 15:24:52 #link https://csit.fd.io/report/#eNrtmMtugzAQRb-GbqqpsIHQTRdN-Y-Ix_BQCXE9JlW6yLfXiSINUR-bFJESbwD5ejRjXx1dCTIbjSvC9smLll689GTcFPbhBc_39qVbkoHwYasUyODBfulc7n2ofBGlizIuxT4TEQRdBqQ0CN-vUCqBj8LP36BQxSs0KlyZvuuwJViEGYgc0NR2tTKqr98PepYSQlfoQ1_5cur7ZQhWi96wakc7U7aoWfx5Zq5R9Y4LfjkJV6QaUy4ZHJC3GKTBjN8dmPeWOl0jNR_IBfaiWM-tRSyJ_LyL2amBerrFODnumNJP5fz8Az_VtH7SrfFJ8-aTbo1Pmh2f0uXnRX7KK-NTuvwc0c-J-XT5OS8-XX7-ez5Dl58X-RleGZ-hy88R_ZyYT5ef8-LT5edof 15:24:52 EbJXbfR6-N_3Cj5BHd-A2o 15:25:28 #topic Release CSIT-2310 15:27:48 #info Maciek reporting on behalf of Tibor - rls testing is in progress, no new issues identified so far. Live testing data is being published automatically into the csit.fd.io as it arrives, e.g. 2n-spr (6 runs) see live link below. 15:27:59 #link https://csit.fd.io/report/#eNrtVMsOgjAQ_Bq8mDW0BPHiQeU_TIFVSQDXbSXBr7cY4kKMiRc9eegrM7udziS17sy4t1itg3gbJNtAJ2XhpyDazP3CldWRCqElAh0t_I6xQmMRdAOWGFQYHlGTwpUK8wuY9gAlLSFZZaByQHfqT35kfU3N3LfWu6H1yz2CFlcnqL99grTIAk5kCY1O3YjzTqzwDaORAq9YIId2JGb6JmEd2NRoyxsK1bsgeO6NFkjl0_6uoxE6GJWkD8aXU2mKfyofpTIY9ZtU6J_KZ6nQM5U4nTVnrh8_WZzeAUipS68 15:32:19 #endmeeting