Requirements

Hardware Requirement
For CPU Based
Minimum requirement
Intel CPU with 4 Cores.
Linux 18.04 or latest.
Docker Runtime latest.
For GPU Based
Minimum requirement
Intel CPU with 4 Cores.
Linux 18.04 or latest.
Nvidia GPU.
Docker Runtime latest.
Docker Compose latest.
Nvidia-docker version 2.5.0 or latest.
NVIDIA CUDA Backend version 11.6.2 or latest. Follow this link to install.
Hardware Benchmark
Face Match
E5-2620 2.2GHz 16 Core → 8 hit/sec parallel, latency 1.5-2s
Xeon E5-2620 2.2GHz 16 Core x2 → 16 hit/sec parallel, latency 1.5-2s
Xeon E5-2620 2.2Ghz 16 Core + Nvidia T4 GPU 16GB → average 40 hit/sec (latencies 150-300ms)
Face Recognition
Xeon E5-2620 2.2Ghz 16 Core + Nvidia T4 GPU 16GB → 8 Hit/Sec for Face Recognition with 100.000 Face Enrollment
Requests      [total, rate, throughput]         20, 1.05, 1.05
Duration      [total, attack, wait]             19.123s, 19s, 123.199ms
Latencies     [min, mean, 50, 90, 95, 99, max]  121.933ms, 124.254ms, 124.136ms, 126.9ms, 126.979ms, 126.998ms, 126.998ms
​
​
Requests      [total, rate, throughput]         100, 5.05, 5.02
Duration      [total, attack, wait]             19.922s, 19.8s, 122.101ms
Latencies     [min, mean, 50, 90, 95, 99, max]  117.716ms, 140.956ms, 121.466ms, 154.547ms, 235.006ms, 516.922ms, 517.784ms
​
​
Requests      [total, rate, throughput]         200, 10.05, 8.29
Duration      [total, attack, wait]             24.12s, 19.9s, 4.22s
Latencies     [min, mean, 50, 90, 95, 99, max]  173.806ms, 2.326s, 2.409s, 3.857s, 4.04s, 4.196s, 4.22s
Chicken Estimation
Xeon E5-2620 2.1Ghz 32 Cores + Nvidia Tesla P4 GPU 7.6GB → 1 Hit/Sec
Requests      [total, rate, throughput]         10, 1.11, 1.01
​
Latencies     [mean, 50, 95, 99, max]           837.41ms, 831.32ms, 871.59ms, 871.59ms, 871.59ms
​
​
Requests      [total, rate, throughput]         100, 1.01, 1.001
​
Latencies     [mean, 50, 95, 99, max]           812.31ms, 827.73ms, 858.22ms, 892.19ms, 894.99ms
​

Last modified 7mo ago