NVIDIA Jetson Nano 開発者キットを Raspberry Pi 3と性能比較してみたベンチマークレビュー

NVIDIA Jetson Nano 開発者キット B01 ASIN: B084DSDDLT マクニカ　［NVIDIA社国内正規代理店］が販売 ※ 2020年出荷の B01版 ※ B01版は Intel 8260NGWが使える

SUCCUL ACアダプター 5V 4A 大手メーカーOEM社製品 センタープラス スイッチング式 最大出力20W 出力プラグ外径5.5mm(内径2.1mm)PSE取得品 ASIN: B015RKFAA2 ※ 4A電源を使用する場合は右のジャンパーピンも必要

Bullet ジャンパーピン5色セット 取手付き JP01 ASIN: B005KVKKZY ※ 4A電源を使用する場合に必要

NVIDIA Jetson Nanoケース、冷却ファン付きNVIDIA Jetson Nano開発者キット用アクリルケース（3.0-5.8V）シェルエンクロージャークーラー. ASIN: B07VLXHZLF ※ 4A電源を使用するなら冷却対策も！

Raspberry Pi 3をシャットダウンした時に電源が切れたかを簡単に確認する方法

Raspberry Piでメモリを馬鹿食いするアプリ用に不要なサービスを停止してフリーメモリを増やす方法

【成功版】Raspberry Piで Darknet Neural Network Frameworkをビルドする方法

$ uname -a
Linux user-desktop 4.9.140-tegra #1 SMP PREEMPT Wed Mar 13 00:32:22 PDT 2019 aarch64 aarch64 aarch64 GNU/Linux

$ free -h
              total        used        free      shared  buff/cache   available
Mem:           3.9G        514M        3.0G         26M        371M        3.2G
Swap:            0B          0B          0B

$ cat /proc/cpuinfo
processor       : 0
model name      : ARMv8 Processor rev 1 (v8l)
BogoMIPS        : 38.40
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x41
CPU architecture: 8
CPU variant     : 0x1
CPU part        : 0xd07
CPU revision    : 1

processor       : 1
model name      : ARMv8 Processor rev 1 (v8l)
BogoMIPS        : 38.40
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x41
CPU architecture: 8
CPU variant     : 0x1
CPU part        : 0xd07
CPU revision    : 1

processor       : 2
model name      : ARMv8 Processor rev 1 (v8l)
BogoMIPS        : 38.40
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x41
CPU architecture: 8
CPU variant     : 0x1
CPU part        : 0xd07
CPU revision    : 1

processor       : 3
model name      : ARMv8 Processor rev 1 (v8l)
BogoMIPS        : 38.40
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x41
CPU architecture: 8
CPU variant     : 0x1
CPU part        : 0xd07
CPU revision    : 1

$ sudo apt-get update

$ sudo apt-get -y install cmake

$ cd
$ git clone https://github.com/pjreddie/darknet.git --depth 1

$ cd darknet
$ git show
commit 61c9d02ec461e30d55762ec7669d6a1d3c356fb2 (grafted, HEAD -> master, origin/master, origin/HEAD)
Author: Joseph Redmon <pjreddie@gmail.com>
Date:   Fri Sep 14 08:03:20 2018 -0700

$ sudo apt-get -y install nano
$ nano Makefile

# GPUと CUDNNを有効にする
$ sed -i -e "s/GPU=0/GPU=1/g" Makefile
$ sed -i -e "s/CUDNN=0/CUDNN=1/g" Makefile

# nvcc: not found
$ make
/bin/sh: 1: nvcc: not found
Makefile:92: recipe for target 'obj/convolutional_kernels.o' failed
make: *** [obj/convolutional_kernels.o] Error 127

$ /usr/local/cuda/bin/nvcc
nvcc fatal   : No input files specified; use option --help for more information

$ /usr/local/cuda/bin/nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sun_Sep_30_21:09:22_CDT_2018
Cuda compilation tools, release 10.0, V10.0.166

# nvccにパスを通す
$ export PATH=${PATH}:/usr/local/cuda/bin
$ make
gcc -Iinclude/ -Isrc/ -DGPU -I/usr/local/cuda/include/ -DCUDNN  -Wall -Wno-unused-result -Wno-unknown-pragmas -Wfatal-errors -fPIC -Ofast -DGPU -DCUDNN obj/captcha.o obj/lsd.o obj/super.o obj/art.o obj/tag.o obj/cifar.o obj/go.o obj/rnn.o obj/segmenter.o obj/regressor.o obj/classifier.o obj/coco.o obj/yolo.o obj/detector.o obj/nightmare.o obj/instance-segmenter.o obj/darknet.o libdarknet.a -o darknet -lm -pthread  -L/usr/local/cuda/lib64 -lcuda -lcudart -lcublas -lcurand -lcudnn -lstdc++  libdarknet.a

$ ls -l darknet
-rwxrwxr-x 1 user user 1792208  4月 19 22:51 darknet

$ ./darknet
usage: ./darknet <function>

$ wget https://pjreddie.com/media/files/yolov2.weights
--2019-04-19 22:56:42--  https://pjreddie.com/media/files/yolov2.weights
Resolving pjreddie.com (pjreddie.com)... 128.208.4.108
Connecting to pjreddie.com (pjreddie.com)|128.208.4.108|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 203934260 (194M) [application/octet-stream]
Saving to: ‘yolov2.weights’

yolov2.weights      100%[===================>] 194.49M  5.57MB/s    in 34s

2019-04-19 22:57:18 (5.65 MB/s) - ‘yolov2.weights’ saved [203934260/203934260]

$ ./darknet detect cfg/yolov2.cfg yolov2.weights data/person.jpg
layer     filters    size              input                output
    0 conv     32  3 x 3 / 1   608 x 608 x   3   ->   608 x 608 x  32  0.639 BFLOPs
    1 max          2 x 2 / 2   608 x 608 x  32   ->   304 x 304 x  32
    2 conv     64  3 x 3 / 1   304 x 304 x  32   ->   304 x 304 x  64  3.407 BFLOPs
    3 max          2 x 2 / 2   304 x 304 x  64   ->   152 x 152 x  64
    4 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    5 conv     64  1 x 1 / 1   152 x 152 x 128   ->   152 x 152 x  64  0.379 BFLOPs
    6 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    7 max          2 x 2 / 2   152 x 152 x 128   ->    76 x  76 x 128
    8 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
    9 conv    128  1 x 1 / 1    76 x  76 x 256   ->    76 x  76 x 128  0.379 BFLOPs
   10 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
   11 max          2 x 2 / 2    76 x  76 x 256   ->    38 x  38 x 256
   12 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   13 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   14 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   15 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   16 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   17 max          2 x 2 / 2    38 x  38 x 512   ->    19 x  19 x 512
   18 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   19 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   20 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   21 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   22 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   23 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   24 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   25 route  16
   26 conv     64  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x  64  0.095 BFLOPs
   27 reorg              / 2    38 x  38 x  64   ->    19 x  19 x 256
   28 route  27 24
   29 conv   1024  3 x 3 / 1    19 x  19 x1280   ->    19 x  19 x1024  8.517 BFLOPs
   30 conv    425  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 425  0.314 BFLOPs
   31 detection
mask_scale: Using default '1.000000'
Loading weights from yolov2.weights...Done!
data/person.jpg: Predicted in 0.730307 seconds.
horse: 82%
dog: 86%
person: 86%

NVIDIA Jetson Nanoは 0.7秒 10Wモード
Raspberry Pi 3は 161秒
data/person.jpg: Predicted in 161.068979 seconds.
horse: 91%
dog: 85%
person: 85%

$ wget http://pjreddie.com/media/files/vgg-conv.weights
URL transformed to HTTPS due to an HSTS policy
--2019-04-19 22:59:37--  https://pjreddie.com/media/files/vgg-conv.weights
Resolving pjreddie.com (pjreddie.com)... 128.208.4.108
Connecting to pjreddie.com (pjreddie.com)|128.208.4.108|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 58858768 (56M) [application/octet-stream]
Saving to: ‘vgg-conv.weights’

vgg-conv.weights    100%[===================>]  56.13M  5.21MB/s    in 14s

2019-04-19 22:59:52 (3.91 MB/s) - ‘vgg-conv.weights’ saved [58858768/58858768]

$ ./darknet nightmare cfg/vgg-conv.cfg vgg-conv.weights data/scream.jpg 10
policy: Using default 'constant'
max_batches: Using default '0'
layer     filters    size              input                output
    0 conv     64  3 x 3 / 1   224 x 224 x   3   ->   224 x 224 x  64  0.173 BFLOPs
    1 conv     64  3 x 3 / 1   224 x 224 x  64   ->   224 x 224 x  64  3.699 BFLOPs
    2 max          2 x 2 / 2   224 x 224 x  64   ->   112 x 112 x  64
    3 conv    128  3 x 3 / 1   112 x 112 x  64   ->   112 x 112 x 128  1.850 BFLOPs
    4 conv    128  3 x 3 / 1   112 x 112 x 128   ->   112 x 112 x 128  3.699 BFLOPs
    5 max          2 x 2 / 2   112 x 112 x 128   ->    56 x  56 x 128
    6 conv    256  3 x 3 / 1    56 x  56 x 128   ->    56 x  56 x 256  1.850 BFLOPs
    7 conv    256  3 x 3 / 1    56 x  56 x 256   ->    56 x  56 x 256  3.699 BFLOPs
    8 conv    256  3 x 3 / 1    56 x  56 x 256   ->    56 x  56 x 256  3.699 BFLOPs
    9 max          2 x 2 / 2    56 x  56 x 256   ->    28 x  28 x 256
   10 conv    512  3 x 3 / 1    28 x  28 x 256   ->    28 x  28 x 512  1.850 BFLOPs
   11 conv    512  3 x 3 / 1    28 x  28 x 512   ->    28 x  28 x 512  3.699 BFLOPs
   12 conv    512  3 x 3 / 1    28 x  28 x 512   ->    28 x  28 x 512  3.699 BFLOPs
   13 max          2 x 2 / 2    28 x  28 x 512   ->    14 x  14 x 512
   14 conv    512  3 x 3 / 1    14 x  14 x 512   ->    14 x  14 x 512  0.925 BFLOPs
   15 conv    512  3 x 3 / 1    14 x  14 x 512   ->    14 x  14 x 512  0.925 BFLOPs
   16 conv    512  3 x 3 / 1    14 x  14 x 512   ->    14 x  14 x 512  0.925 BFLOPs
   17 max          2 x 2 / 2    14 x  14 x 512   ->     7 x   7 x 512
Loading weights from vgg-conv.weights...Done!
Iteration: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, done
0 scream_vgg-conv_10_000000

# GPUと CUDNNを無効にする
$ sed -i -e "s/GPU=1/GPU=0/g" Makefile
$ sed -i -e "s/CUDNN=1/CUDNN=0/g" Makefile

gcc -Iinclude/ -Isrc/ -Wall -Wno-unused-result -Wno-unknown-pragmas -Wfatal-errors -fPIC -Ofast obj/captcha.o obj/lsd.o obj/super.o obj/art.o obj/tag.o obj/cifar.o obj/go.o obj/rnn.o obj/segmenter.o obj/regressor.o obj/classifier.o obj/coco.o obj/yolo.o obj/detector.o obj/nightmare.o obj/instance-segmenter.o obj/darknet.o libdarknet.a -o darknet -lm -pthread  libdarknet.a

real    0m46.887s
user    0m57.984s
sys     0m3.388s

$ ls -l darknet
-rwxrwxr-x 1 user user 573688  4月 20 14:38 darknet

$ ./darknet detect cfg/yolov2.cfg yolov2.weights data/person.jpg
layer     filters    size              input                output
    0 conv     32  3 x 3 / 1   608 x 608 x   3   ->   608 x 608 x  32  0.639 BFLOPs
    1 max          2 x 2 / 2   608 x 608 x  32   ->   304 x 304 x  32
    2 conv     64  3 x 3 / 1   304 x 304 x  32   ->   304 x 304 x  64  3.407 BFLOPs
    3 max          2 x 2 / 2   304 x 304 x  64   ->   152 x 152 x  64
    4 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    5 conv     64  1 x 1 / 1   152 x 152 x 128   ->   152 x 152 x  64  0.379 BFLOPs
    6 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    7 max          2 x 2 / 2   152 x 152 x 128   ->    76 x  76 x 128
    8 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
    9 conv    128  1 x 1 / 1    76 x  76 x 256   ->    76 x  76 x 128  0.379 BFLOPs
   10 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
   11 max          2 x 2 / 2    76 x  76 x 256   ->    38 x  38 x 256
   12 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   13 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   14 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   15 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   16 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   17 max          2 x 2 / 2    38 x  38 x 512   ->    19 x  19 x 512
   18 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   19 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   20 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   21 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   22 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   23 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   24 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   25 route  16
   26 conv     64  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x  64  0.095 BFLOPs
   27 reorg              / 2    38 x  38 x  64   ->    19 x  19 x 256
   28 route  27 24
   29 conv   1024  3 x 3 / 1    19 x  19 x1280   ->    19 x  19 x1024  8.517 BFLOPs
   30 conv    425  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 425  0.314 BFLOPs
   31 detection
mask_scale: Using default '1.000000'
Loading weights from yolov2.weights...Done!
data/person.jpg: Predicted in 36.355337 seconds.
horse: 82%
dog: 86%
person: 86%

# NVIDIA Jetson Nanoは 10Wモード
# GPUと CUDNNを無効にした状態では 36秒
# GPUと CUDNNを有効にした状態では 0.7秒

pi@raspberrypi:~ $ git clone https://github.com/pjreddie/darknet.git --depth 1
pi@raspberrypi:~ $ cd darknet
pi@raspberrypi:~/darknet $ git show
commit 61c9d02ec461e30d55762ec7669d6a1d3c356fb2

pi@raspberrypi:~/darknet $ ls -l darknet
-rwxr-xr-x 1 pi pi 476192 Apr 20 06:51 darknet

pi@raspberrypi:~/darknet $ wget https://pjreddie.com/media/files/yolov2.weights
yolov2.weights      100%[===================>] 194.49M  1.52MB/s    in 85s
2019-04-20 06:53:29 (2.28 MB/s) - ‘yolov2.weights’ saved [203934260/203934260]

pi@raspberrypi:~/darknet $ ./darknet detect cfg/yolov2.cfg yolov2.weights data/person.jpg
layer     filters    size              input                output
    0 conv     32  3 x 3 / 1   608 x 608 x   3   ->   608 x 608 x  32  0.639 BFLOPs
    1 max          2 x 2 / 2   608 x 608 x  32   ->   304 x 304 x  32
    2 conv     64  3 x 3 / 1   304 x 304 x  32   ->   304 x 304 x  64  3.407 BFLOPs
    3 max          2 x 2 / 2   304 x 304 x  64   ->   152 x 152 x  64
    4 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    5 conv     64  1 x 1 / 1   152 x 152 x 128   ->   152 x 152 x  64  0.379 BFLOPs
    6 conv    128  3 x 3 / 1   152 x 152 x  64   ->   152 x 152 x 128  3.407 BFLOPs
    7 max          2 x 2 / 2   152 x 152 x 128   ->    76 x  76 x 128
    8 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
    9 conv    128  1 x 1 / 1    76 x  76 x 256   ->    76 x  76 x 128  0.379 BFLOPs
   10 conv    256  3 x 3 / 1    76 x  76 x 128   ->    76 x  76 x 256  3.407 BFLOPs
   11 max          2 x 2 / 2    76 x  76 x 256   ->    38 x  38 x 256
   12 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   13 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   14 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   15 conv    256  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x 256  0.379 BFLOPs
   16 conv    512  3 x 3 / 1    38 x  38 x 256   ->    38 x  38 x 512  3.407 BFLOPs
   17 max          2 x 2 / 2    38 x  38 x 512   ->    19 x  19 x 512
   18 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   19 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   20 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   21 conv    512  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 512  0.379 BFLOPs
   22 conv   1024  3 x 3 / 1    19 x  19 x 512   ->    19 x  19 x1024  3.407 BFLOPs
   23 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   24 conv   1024  3 x 3 / 1    19 x  19 x1024   ->    19 x  19 x1024  6.814 BFLOPs
   25 route  16
   26 conv     64  1 x 1 / 1    38 x  38 x 512   ->    38 x  38 x  64  0.095 BFLOPs
   27 reorg              / 2    38 x  38 x  64   ->    19 x  19 x 256
   28 route  27 24
   29 conv   1024  3 x 3 / 1    19 x  19 x1280   ->    19 x  19 x1024  8.517 BFLOPs
   30 conv    425  1 x 1 / 1    19 x  19 x1024   ->    19 x  19 x 425  0.314 BFLOPs
   31 detection
mask_scale: Using default '1.000000'
Loading weights from yolov2.weights...Done!
data/person.jpg: Predicted in 292.377208 seconds.
horse: 82%
dog: 86%
person: 86%

Raspberry Pi 3の CPUパワーで 暗号通貨を採掘する（非実用）。CPU Miner cpuminer-multiで採掘する

$ cd
$ git clone https://github.com/tpruvot/cpuminer-multi

$ cd cpuminer-multi/
$ ./build.sh
configure: error: OpenSSL crypto library required
make: *** No targets specified and no makefile found.  Stop.

$ sudo apt-get -y install automake autoconf pkg-config libcurl4-openssl-dev libjansson-dev libssl-dev libgmp-dev make g++

$ ./build.sh

/home/user/cpuminer-multi/algo/cryptonight.c:327: undefined reference to `fast_aesb_pseudo_round_mut'
/tmp/ccjffwmY.ltrans10.ltrans.o:/home/user/cpuminer-multi/algo/cryptonight.c:329: more undefined references to `fast_aesb_pseudo_round_mut' follow
collect2: error: ld returned 1 exit status
Makefile:941: recipe for target 'cpuminer' failed
make[2]: *** [cpuminer] Error 1
make[2]: Leaving directory '/home/user/cpuminer-multi'
Makefile:2766: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/user/cpuminer-multi'
Makefile:585: recipe for target 'all' failed
make: *** [all] Error 2
strip: 'cpuminer': No such file

$ ./build-linux-arm.sh

make[2]: Leaving directory '/home/user/cpuminer-multi'
make[1]: Leaving directory '/home/user/cpuminer-multi'
  => done.
$ ls -l cpuminer
-rwxrwxr-x 1 user user 1412352  4月 19 23:31 cpuminer
Stripping...
  => done.

$ ./cpuminer -help
** cpuminer-multi 1.3.6 by tpruvot@github **

$ ./cpuminer --benchmark
** cpuminer-multi 1.3.6 by tpruvot@github **
BTC donation address: 1FhDPLPpw18X4srecguG3MxJYe4a1JsZnd (tpruvot)

[2019-04-19 23:42:53] 4 miner threads started, using 'scrypt' algorithm.
[2019-04-19 23:42:54] CPU #0: 1.66 kH/s
[2019-04-19 23:42:54] CPU #1: 1.66 kH/s
[2019-04-19 23:42:54] CPU #2: 1.66 kH/s
[2019-04-19 23:42:54] CPU #3: 1.66 kH/s
[2019-04-19 23:42:54] Total: 6.64 kH/s
[2019-04-19 23:42:58] Total: 6.65 kH/s
[2019-04-19 23:43:03] CPU #0: 1.66 kH/s
[2019-04-19 23:43:03] CPU #1: 1.66 kH/s
[2019-04-19 23:43:03] CPU #2: 1.66 kH/s
[2019-04-19 23:43:03] CPU #3: 1.66 kH/s
[2019-04-19 23:43:03] Total: 6.65 kH/s
[2019-04-19 23:43:08] Total: 6.65 kH/s
[2019-04-19 23:43:13] CPU #2: 1.66 kH/s
[2019-04-19 23:43:13] CPU #1: 1.66 kH/s
[2019-04-19 23:43:13] CPU #3: 1.66 kH/s
[2019-04-19 23:43:13] Total: 6.65 kH/s
NVIDIA Jetson Nanoは 10Wモード

ラズパイ３は
[2017-10-14 08:28:00] thread 2: 4096 hashes, 0.95 khash/s
[2017-10-14 08:28:00] thread 0: 4096 hashes, 0.95 khash/s
[2017-10-14 08:28:00] thread 1: 4096 hashes, 0.93 khash/s
[2017-10-14 08:28:05] thread 0: 4733 hashes, 0.96 khash/s
[2017-10-14 08:28:05] thread 2: 4753 hashes, 0.95 khash/s
[2017-10-14 08:28:05] Total: 2.84 khash/s

$ ./cpuminer --benchmark -a sha256d
** cpuminer-multi 1.3.6 by tpruvot@github **
BTC donation address: 1FhDPLPpw18X4srecguG3MxJYe4a1JsZnd (tpruvot)

[2019-04-19 23:45:16] 4 miner threads started, using 'sha256d' algorithm.
[2019-04-19 23:45:19] CPU #0: 847.45 kH/s
[2019-04-19 23:45:19] CPU #1: 846.30 kH/s
[2019-04-19 23:45:19] CPU #3: 847.65 kH/s
[2019-04-19 23:45:19] CPU #2: 847.41 kH/s
[2019-04-19 23:45:21] Total: 3389 kH/s
[2019-04-19 23:45:26] CPU #0: 846.96 kH/s
[2019-04-19 23:45:26] CPU #1: 847.05 kH/s
[2019-04-19 23:45:26] CPU #3: 847.71 kH/s
[2019-04-19 23:45:26] Total: 3389 kH/s
[2019-04-19 23:45:26] CPU #2: 847.47 kH/s
[2019-04-19 23:45:31] Total: 3389 kH/s
[2019-04-19 23:45:36] CPU #0: 846.90 kH/s
[2019-04-19 23:45:36] CPU #1: 847.10 kH/s
[2019-04-19 23:45:36] CPU #3: 847.72 kH/s
[2019-04-19 23:45:36] Total: 3389 kH/s
NVIDIA Jetson Nanoは 10Wモード

ラズパイ３は
[2017-10-14 08:30:18] thread 0: 3385205 hashes, 672.18 khash/s
[2017-10-14 08:30:23] thread 1: 3399853 hashes, 679.74 khash/s
[2017-10-14 08:30:23] thread 2: 3402732 hashes, 680.62 khash/s
[2017-10-14 08:30:23] Total: 2033 khash/s

pi@raspberrypi:~ $ cd

pi@raspberrypi:~ $ sudo apt-get -y install automake autoconf pkg-config libcurl4-openssl-dev libjansson-dev libssl-dev libgmp-dev make g++

pi@raspberrypi:~ $ git clone https://github.com/tpruvot/cpuminer-multi
pi@raspberrypi:~ $ cd cpuminer-multi/

pi@raspberrypi:~/cpuminer-multi $ git show
commit 39fff9e5b91690ad21a89e1ffd2a5d7cdc444a1a
Author: Tanguy Pruvot <tanguy.pruvot@gmail.com>
Date:   Sat Mar 16 17:16:41 2019 +0100

pi@raspberrypi:~/cpuminer-multi $ ./build-linux-arm.sh

make[2]: Leaving directory '/home/pi/cpuminer-multi'
make[1]: Leaving directory '/home/pi/cpuminer-multi'
  => done.
$ ls -l cpuminer
-rwxr-xr-x 1 pi pi 1892376 Apr 20 07:44 cpuminer
Stripping...
  => done.

pi@raspberrypi:~/cpuminer-multi $ ./cpuminer -help
** cpuminer-multi 1.3.6 by tpruvot@github **
BTC donation address: 1FhDPLPpw18X4srecguG3MxJYe4a1JsZnd (tpruvot)

Usage: cpuminer-multi [OPTIONS]
Options:
  -a, --algo=ALGO       specify the algorithm to use
                          allium       Garlicoin double lyra2
                          axiom        Shabal-256 MemoHash
                          bitcore      Timetravel with 10 algos
                          blake        Blake-256 14-rounds (SFR)
                          blakecoin    Blake-256 single sha256 merkle
                          blake2b      Blake2-B (512)
                          blake2s      Blake2-S (256)
                          bmw          BMW 256
                          c11/flax     C11
                          cryptolight  Cryptonight-light
                          cryptonight  Monero
                          decred       Blake-256 14-rounds 180 bytes
                          dmd-gr       Diamond-Groestl
                          drop         Dropcoin
                          fresh        Fresh
                          geek         GeekCash
                          groestl      GroestlCoin
                          heavy        Heavy
                          jha          JHA
                          keccak       Keccak (Old and deprecated)
                          keccakc      Keccak (CreativeCoin)
                          luffa        Luffa
                          lyra2re      Lyra2RE
                          lyra2rev2    Lyra2REv2
                          lyra2v3      Lyra2REv3 (Vertcoin)
                          myr-gr       Myriad-Groestl
                          neoscrypt    NeoScrypt(128, 2, 1)
                          nist5        Nist5
                          pluck        Pluck:128 (Supcoin)
                          pentablake   Pentablake
                          phi          LUX initial algo
                          phi2         LUX newer algo
                          quark        Quark
                          qubit        Qubit
                          rainforest   RainForest (256)
                          scrypt       scrypt(1024, 1, 1) (default)
                          scrypt:N     scrypt(N, 1, 1)
                          scrypt-jane:N (with N factor from 4 to 30)
                          shavite3     Shavite3
                          sha256d      SHA-256d
                          sia          Blake2-B
                          sib          X11 + gost (SibCoin)
                          skein        Skein+Sha (Skeincoin)
                          skein2       Double Skein (Woodcoin)
                          sonoa        A series of 97 hashes from x17
                          s3           S3
                          timetravel   Timetravel (Machinecoin)
                          vanilla      Blake-256 8-rounds
                          x11evo       Permuted x11
                          x11          X11
                          x12          X12
                          x13          X13
                          x14          X14
                          x15          X15
                          x16r         X16R (Raven)
                          x16s         X16S (Pigeon)
                          x17          X17
                          x20r         X20R
                          xevan        Xevan (BitSend)
                          yescrypt     Yescrypt
                          zr5          ZR5
  -o, --url=URL         URL of mining server
  -O, --userpass=U:P    username:password pair for mining server
  -u, --user=USERNAME   username for mining server
  -p, --pass=PASSWORD   password for mining server
      --cert=FILE       certificate for mining server using SSL
  -x, --proxy=[PROTOCOL://]HOST[:PORT]  connect through a proxy
  -t, --threads=N       number of miner threads (default: number of processors)
  -r, --retries=N       number of times to retry if a network call fails
                          (default: retry indefinitely)
  -R, --retry-pause=N   time to pause between retries, in seconds (default: 30)
      --time-limit=N    maximum time [s] to mine before exiting the program.
  -T, --timeout=N       timeout for long poll and stratum (default: 300 seconds)
  -s, --scantime=N      upper bound on time spent scanning current work when
                          long polling is unavailable, in seconds (default: 5)
      --randomize       Randomize scan range start to reduce duplicates
  -f, --diff-factor     Divide req. difficulty by this factor (std is 1.0)
  -m, --diff-multiplier Multiply difficulty by this factor (std is 1.0)
  -n, --nfactor         neoscrypt N-Factor
      --coinbase-addr=ADDR  payout address for solo mining
      --coinbase-sig=TEXT  data to insert in the coinbase when possible
      --max-log-rate    limit per-core hashrate logs (default: 5s)
      --no-longpoll     disable long polling support
      --no-getwork      disable getwork support
      --no-gbt          disable getblocktemplate support
      --no-stratum      disable X-Stratum support
      --no-extranonce   disable Stratum extranonce support
      --no-redirect     ignore requests to change the URL of the mining server
  -q, --quiet           disable per-thread hashmeter output
      --no-color        disable colored output
  -D, --debug           enable debug output
  -P, --protocol-dump   verbose dump of protocol-level activities
      --hide-diff       Hide submitted block and net difficulty
  -S, --syslog          use system log for output messages
  -B, --background      run the miner in the background
      --benchmark       run in offline benchmark mode
      --cputest         debug hashes from cpu algorithms
      --cpu-affinity    set process affinity to cpu core(s), mask 0x3 for cores 0 and 1
      --cpu-priority    set process priority (default: 0 idle, 2 normal to 5 highest)
  -b, --api-bind        IP/Port for the miner API (default: 127.0.0.1:4048)
      --api-remote      Allow remote control
      --max-temp=N      Only mine if cpu temp is less than specified value (linux)
      --max-rate=N[KMG] Only mine if net hashrate is less than specified value
      --max-diff=N      Only mine if net difficulty is less than specified value
  -c, --config=FILE     load a JSON-format configuration file
  -V, --version         display version information and exit
  -h, --help            display this help text and exit

pi@raspberrypi:~/cpuminer-multi $ ./cpuminer --benchmark
** cpuminer-multi 1.3.6 by tpruvot@github **
BTC donation address: 1FhDPLPpw18X4srecguG3MxJYe4a1JsZnd (tpruvot)

[2019-04-20 07:46:26] 4 miner threads started, using 'scrypt' algorithm.
[2019-04-20 07:46:28] CPU #1: 0.31 kH/s
[2019-04-20 07:46:28] CPU #2: 0.31 kH/s
[2019-04-20 07:46:28] CPU #3: 0.31 kH/s
[2019-04-20 07:46:28] CPU #0: 0.30 kH/s
[2019-04-20 07:46:31] Total: 1.23 kH/s
[2019-04-20 07:46:36] CPU #3: 0.31 kH/s
[2019-04-20 07:46:36] Total: 1.22 kH/s
[2019-04-20 07:46:36] CPU #2: 0.30 kH/s
[2019-04-20 07:46:36] CPU #1: 0.30 kH/s
[2019-04-20 07:46:36] CPU #0: 0.29 kH/s
[2019-04-20 07:46:41] Total: 1.23 kH/s
[2019-04-20 07:46:46] CPU #1: 0.30 kH/s
[2019-04-20 07:46:46] CPU #2: 0.30 kH/s
[2019-04-20 07:46:46] CPU #3: 0.31 kH/s
[2019-04-20 07:46:46] Total: 1.21 kH/s

pi@raspberrypi:~/cpuminer-multi $ ./cpuminer --benchmark -a sha256d
** cpuminer-multi 1.3.6 by tpruvot@github **
BTC donation address: 1FhDPLPpw18X4srecguG3MxJYe4a1JsZnd (tpruvot)

[2019-04-20 07:47:08] 4 miner threads started, using 'sha256d' algorithm.
[2019-04-20 07:47:16] CPU #2: 285.96 kH/s
[2019-04-20 07:47:16] CPU #3: 285.28 kH/s
[2019-04-20 07:47:16] CPU #1: 283.61 kH/s
[2019-04-20 07:47:16] CPU #0: 275.14 kH/s
[2019-04-20 07:47:21] Total: 1124 kH/s
[2019-04-20 07:47:25] CPU #2: 290.39 kH/s
[2019-04-20 07:47:25] CPU #3: 290.31 kH/s
[2019-04-20 07:47:25] Total: 1133 kH/s
[2019-04-20 07:47:26] CPU #1: 287.98 kH/s
[2019-04-20 07:47:26] CPU #0: 279.75 kH/s
[2019-04-20 07:47:27] Total: 1118 kH/s
[2019-04-20 07:47:30] Total: 1137 kH/s
[2019-04-20 07:47:31] CPU #2: 275.05 kH/s
[2019-04-20 07:47:31] CPU #3: 275.09 kH/s
[2019-04-20 07:47:31] Total: 1102 kH/s
[2019-04-20 07:47:36] CPU #1: 280.75 kH/s
[2019-04-20 07:47:36] CPU #0: 279.76 kH/s
[2019-04-20 07:47:36] Total: 1142 kH/s

sudo apt-get -y install ninja-build

sudo pip install --upgrade git+https://github.com/Maratyszcza/PeachPy
sudo: pip: command not found

sudo apt-get -y install python-pip

 user@user-desktop:~$ python -V
# Python 2.7.15rc1

sudo pip install --upgrade git+https://github.com/Maratyszcza/PeachPy
sudo pip install --upgrade git+https://github.com/Maratyszcza/confu

cd
git clone https://github.com/Maratyszcza/NNPACK.git
cd NNPACK
confu setup
python configure.py

ninja

[87/198] LINK bin/convolution-inference-bench
FAILED: /home/user/NNPACK/bin/convolution-inference-bench
/home/user/NNPACK/deps/googlebenchmark/src/benchmark_register.cc:227: undefined reference to `benchmark::BenchmarkName::str[abi:cxx11]() const'
/home/user/NNPACK/lib/libgooglebenchmark.a(json_reporter.cc.o): In function `benchmark::JSONReporter::PrintRunData(benchmark::BenchmarkReporter::Run const&)':
/home/user/NNPACK/deps/googlebenchmark/src/json_reporter.cc:192: undefined reference to `benchmark::BenchmarkName::str[abi:cxx11]() const'
/home/user/NNPACK/lib/libgooglebenchmark.a(reporter.cc.o):/home/user/NNPACK/deps/googlebenchmark/src/reporter.cc:86: more undefined references to `benchmark::BenchmarkName::str[abi:cxx11]() const' follow
collect2: error: ld returned 1 exit status
[90/198] CXX bench/hxgemm.cc
ninja: build stopped: subcommand failed.

# python configure.pyの場合は Google Benchmark側のマイナーチェンジでビルドエラーが発生します

Maratyszcza commented Apr 28, 2019
confu recipe for Google Benchmark needs an update for recent Google Benchmark changes (mostly, add new files). One option is to fix the recipe here. Another option is to build NNPACK with CMake instead of Ninja.

sudo apt-get -y install ninja-build

cd
git clone https://github.com/Maratyszcza/NNPACK.git
cd NNPACK

mkdir build
cd build
cmake -G Ninja ..
ninja

# NNPACKのベンチマーク（動作テスト）
ninja test

user@user-desktop:~/NNPACK/build$ time ninja test
[0/1] Running tests...
Test project /home/user/NNPACK/build
      Start  1: convolution-inference-smoketest
 1/34 Test  #1: convolution-inference-smoketest .........   Passed   44.09 sec
      Start  2: convolution-inference-alexnet
 2/34 Test  #2: convolution-inference-alexnet ...........   Passed  259.91 sec
      Start  3: convolution-inference-overfeat
 3/34 Test  #3: convolution-inference-overfeat ..........   Passed  975.74 sec
      Start  4: convolution-inference-vgg
 4/34 Test  #4: convolution-inference-vgg ...............   Passed  1636.24 sec
      Start  5: convolution-output-smoketest
 5/34 Test  #5: convolution-output-smoketest ............   Passed    6.40 sec
      Start  6: convolution-output-alexnet
 6/34 Test  #6: convolution-output-alexnet ..............   Passed  3258.54 sec
      Start  7: convolution-output-overfeat
 7/34 Test  #7: convolution-output-overfeat .............   Passed  15647.77 sec
      Start  8: convolution-output-vgg

 8/34 Test  #8: convolution-output-vgg ..................   Passed  24916.11 sec
      Start  9: convolution-input-gradient-smoketest
 9/34 Test  #9: convolution-input-gradient-smoketest ....   Passed    2.85 sec
      Start 10: convolution-input-gradient-alexnet
10/34 Test #10: convolution-input-gradient-alexnet ......   Passed  1644.86 sec
      Start 11: convolution-input-gradient-overfeat
11/34 Test #11: convolution-input-gradient-overfeat .....***Failed  10810.61 sec

[0/1] Running tests...
Test project /home/pi/NNPACK/build
      Start  1: convolution-inference-smoketest
 1/34 Test  #1: convolution-inference-smoketest .........   Passed   59.82 sec
      Start  2: convolution-inference-alexnet
 2/34 Test  #2: convolution-inference-alexnet ...........   Passed  480.32 sec
      Start  3: convolution-inference-overfeat
 3/34 Test  #3: convolution-inference-overfeat ..........***Failed  2103.42 sec
      Start  4: convolution-inference-vgg
 4/34 Test  #4: convolution-inference-vgg ...............   Passed  4407.35 sec
      Start  5: convolution-output-smoketest
 5/34 Test  #5: convolution-output-smoketest ............   Passed   15.74 sec

Raspberry Pi Raspbian Stretchで PhantomJS 2.1.1をビルドする方法、OpenSSL 1.1の罠

【2020年版】NVIDIA Jetson Nano、Jetson Xavier NXの便利スクリプト

NVIDIA Jetson Nano 開発者キットを買ってみた。メモリ容量 4GB LPDDR4 RAM

NVIDIA Jetson Nano 開発者キットを動かすのに最低限必要な物の一覧、冷却ファン、NGFF 無線カード

NVIDIA Jetson Nano 開発者キットで使用する microSDカードに Ubuntu OSイメージを焼く方法を説明

NVIDIA Jetson Nano 開発者キットで SDカードで起動したら一番最初にする事

NVIDIA Jetson Nano Ubuntuのパッケージがアップデート可能な場合にアップデートする方法

NVIDIA Jetson Nanoで nvcc command not found build CUDA app Errorの対応方法

NVIDIA Jetson Nanoの GUI環境を無効にして CUI環境で動く様にしてフリーメモリエリアを広げる

NVIDIA Jetson Nano 開発者キットの Tips一覧、冷却ファンが動かない、20Wモードで動かす、動作温度を知る、他

NVIDIA Jetson Nano 開発者キットに Raspberry Pi Camera Module V2 RaspiCamを接続する方法

Raspberry Piで TensorFlow Deep Learning Frameworkを自己ビルドする方法

Raspberry Piに PyTorch Deep Learning Frameworkをソースコードからビルドする方法、DeepDreamでキモイ絵を作成

Raspberry Piで darkflowを動かしてリアルタイムでカメラ映像を画像物体検出する方法

Raspberry Piで Caffe Deep Learning Frameworkで物体認識を行なってみるテスト

Raspberry Piで Deep Learningフレームワーク Chainerをインストールしてみる

Raspberry Piで DeepBeliefSDKをビルドして画像認識フレームワークを動かす方法

【成功版】最新版の Darknetに digitalbrain79版の Darknet with NNPACKの NNPACK処理を適用する

【成功版】Raspberry Piで Darknet Neural Network Frameworkをビルドする方法

【成功版】Raspberry Piに TensorFlow Deep Learning Frameworkをインストールする方法

Raspberry Piに Jupyter Notebookをインストールして拡張子 ipynb形式の IPythonを動かす

Raspberry Piで Caffe2 Deep Learning Frameworkをソースコードからビルドする方法

【ビルド版】Raspberry Piで DeepDreamを動かしてキモイ絵をモリモリ量産 Caffe Deep Learning Framework

Raspberry Pi 3 Model Bで動画処理アプリ FFmpegをコンパイルする方法

Raspberry Pi 3 Model Bで動画処理アプリ libavをコンパイルする方法

Raspberry Piで NNPACKをビルドする方法

BLAS、ベクトルと行列に関する基本線型代数操作を実行する演算ライブラリ APIのまとめ

Raspberry Pi 3の Linuxコンソール上で使用する各種コマンドまとめ

PIP機能付きの 4K対応の 4入力 1出力の HDMIセレクターを買ってみた、HDMI機器が複数有る場合に便利

EDID保持機能付きの 4K対応の 4入力 2出力の マトリックス切り替え HDMIセレクター、液晶画面 2台と使用で最強