ncnn-benchmark

The benchmark of ncnn that is a high-performance neural network inference framework optimized for the mobile platform https://github.com/Tencent/ncnn

Hardware Platform

Device	System	CPU-Family	CPU-Num	Freq
RK3288	Android 5.1	Cortex-A17	4	1.8GHz
Qualcomm820	Android 6.0	Kryo	2+2	2.15GHz/1.6GHz
Mi5	Android 7.1.2	Kryo	2+2	1.8GHz/1.3GHz

Runtime Environment

1.Using the cpu working in performance model.

echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor

2.Loop a hundred times to take the minimum value of Inference time-consuming.

Result

Devices	Models	Input Size	Single-Thread(ms)	Multi-Threads(ms)
RK3288	SqueezeNet v1.1	227x227x3	194	80
	MobileNet v1.0	224x224x3	324	115
	ResNet18	224x224x3	807	291
	GoogleNet v1.0	224x224x3	740	261
	VGG16	224x224x3	3449	1506
	MobileNet-SSD	300x300x3	652	245
Qualcomm820	SqueezeNet v1.1	227x227x3	91	47
	MobileNet v1.0	224x224x3	150	70
	ResNet18	224x224x3	356	171
	GoogleNet v1.0	224x224x3	371	161
	VGG16	224x224x3	1956	846
	MobileNet-SSD	300x300x3	330	160
Mi5	SqueezeNet v1.1	227x227x3	98	51
	MobileNet v1.0	224x224x3	189	79

User Guide

1. Build the benchmark demo

build demo for Linux-x86

./build.sh linux

build demo for Android

./build.sh android

2. How to run the executable files.

If you build demo for linux success,and want to run the benchmark demo.

$ cp ./models/classification/squeezenet.param  ./build-linux/install/bin/
$ cp ./models/classification/squeezenet.bin  ./build-linux/install/bin/
$ cd ./build-linux/install/bin/
$ ./ncnn_classify squeezenet.param squeezenet.bin 227 227 1 1

Demo running params:

./ncnn_classify <ncnn-param-file> <ncnn-model-bin-file> <input-width> <input-height> <loops-num> <threads-num>

Example:

bug1989@ubuntu:~/ncnn-benchmark/build-linux/install/bin$ ./ncnn_classify squeezenet.param squeezenet.bin 227 227 10 2
--- NCNN Classification Benchmark Demo --- 22:41:09 Dec 26 2017
Loops   : 10
Threads : 2
Time cost: Max 263.338 ms, Min 247.209 ms, Avg 252.652 ms.

The End

Thanks to ncnn's author nihui and all the contributors for sharing this framework.

webnoise/ncnn-benchmark