| Baseline 10k flows | Baseline 4 flows (1 flow/port) | POD12 CI before tuning 2 flows (1 flow/port) (12 Jun 2017) | POD12 CI current results | POD 12 Sandbox after updating OVS and DPDK | RH
|
|---|---|---|---|---|---|---|
64B - 1 PMD core
| P2P 6.1 Mpps 4.1 Gbps PVP 2.3 Mpps 1.5 Gbps | PVP 3.8 Mpps 2.6 Gbps | P2P 14.1 Mpps PVP 3.7 Mpps | |||
64B - 2 PMD cores | P2P 16.1 Mpps 10.8 Gbps PVP 4.7 Mpps 3.1 Gbps | PVP 6.8 Mpps 4.6 Gbps | P2P 23.5 Mpps PVP 7.5 Mpps | P2P 15.1 Mpps PVP 4.63 Mpps | ||
64B | P2P (4 ports - 4 PMD cores, Fortville) 36.0 Mpps (Fortville limit) 24.2 Gbps PVP 10.1 Mpps 6.8 Gbps (20B added to 64B) | PVP 12.9 Mpps 8.6 Gbps (20B added to 64B) | P2P (2 ports - 2 PMD cores, Niantic) 23.3 Mpps (Niantic limit) 11.9 Gbps PVP 7.9 Mpps 4.0 Gbps | PVP 4.2 Mpps (15Aug) | P2P 23.5 Mpps PVP_tput 12.7 Mpps (is this correct?) I ran it four times to confirm, Yes that is correct. If I used the ubuntu vloop image, I was able to get this number consistently. I can run it again and provide a log. Note this was only 30 second trials, if I ran 2 minute trials it dropped to 11 something. | PVP 6.6 Mpps 2 Cores/ 4 Hyperthreads |
Pkts | IPv4, and Ethernet headers Increment SRC and DEST IP | |||||
Flow configuration | 4 NIC ports with 1 flow per port | 2 NIC ports with 1 flow per port | ||||
Frame loss | 0% | 0% | 0% | |||
Resolution | 0.1 | 0.1 | 0.1 | |||
Test duration | 60 sec | 30 sec | 30 sec | 20 min | ||
NIC | Fortville X710-DA2 8.0 GT/s, x8 lanes | Niantic 82599ES PCIe v2.0 (5.0GT/s) | Spring Fountain X520 PCIe v2.0 (5.0GT/s) | |||
PCI-E | 2 x PCI-E 3.0 x8 slot | |||||
Ports | 4 x 10GE | |||||
OVS | 2.7.0 | 2.6.9 | ||||
DPDK | 16.11 | 17.02 | 16.07 | |||
Host OS | Ubuntu-16.04 4.4.0-36-generic | CentOS 7.3 | Centos 7.3 | |||
Guest VM | Ubuntu-16.04 4.4.0-21-generic | |||||
QEMU | QEMU-KVM version 2.8.0 | 2.5 | 2.6 | |||
Processor | E5-2695 v4 Broadwell-EP 18 cores 2.10 GHz (3.30 GHz) | E5-2699 v4 Broadwell-EP 22 cores 2.20GHz (3.60 GHz) | Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz 44 Cores/ 88 with HT | E5-2699 v3 Haswell 18 cores 2.30GHz (3.60 GHz) | ||
Hyperthreading | Disable | Enabled | Enabled | |||
Turbo mode | Disable | Enable | DIsabled | |||
Traffic Generator | Ixia | Ixia | Ixia | Xena |
Anuket Project
Overview
Content Tools