python使用ai-benchmark 测试本机cpu、 gpu性能

 

• 安装ai-benchmark

pip install ai-benchmark。注意，需要安装tensorflow才能运行。

使用方法如下：

from ai_benchmark import AIBenchmark
results = AIBenchmark().run()

 

• 基本信息：

   *  TF Version: 1.13.1
*  Platform: Windows-10-10.0.17134-SP0
*  CPU: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz
*  CPU RAM: 8 GB
*  GPU/0: GeForce GTX 1050 Ti
*  GPU RAM: 2.9 GB
*  CUDA Version: 10.0
*  CUDA Build: V10.0.130

• 得分：

   

CPU得分	GPU得分
Device Inference Score: 220 Device Training Score: 244 Device AI Score: 464	Device Inference Score: 2987 Device Training Score: 3392 Device AI Score: 6379

• cpu测试详细信息：

>>   AI-Benchmark-v.0.1.1   
>>   Let the AI Games begin..

*  TF Version: 1.13.1
*  Platform: Windows-10-10.0.17134-SP0
*  CPU: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz
*  CPU RAM: 8 GB

The benchmark is running...
The tests might take up to 20 minutes
Please don't interrupt the script

1/19. MobileNet-V2

1.1 - inference | batch=50, size=224x224: 1017 ± 11 ms
1.2 - training  | batch=50, size=224x224: 8686 ± 81 ms

2/19. Inception-V3

2.1 - inference | batch=20, size=346x346: 2404 ± 23 ms
2.2 - training  | batch=20, size=346x346: 12050 ± 39 ms

3/19. Inception-V4

3.1 - inference | batch=10, size=346x346: 2300 ± 44 ms
3.2 - training  | batch=10, size=346x346: 11301 ± 266 ms

4/19. Inception-ResNet-V2

4.1 - inference | batch=10, size=346x346: 2830 ± 98 ms
4.2 - training  | batch=8, size=346x346: 10579 ± 248 ms

5/19. ResNet-V2-50

5.1 - inference | batch=10, size=346x346: 1562 ± 51 ms
5.2 - training  | batch=10, size=346x346: 10839 ± 425 ms

6/19. ResNet-V2-152

6.1 - inference | batch=10, size=256x256: 2221 ± 106 ms
6.2 - training  | batch=10, size=256x256: 12712 ± 154 ms

7/19. VGG-16

7.1 - inference | batch=20, size=224x224: 3623 ± 68 ms
7.2 - training  | batch=2, size=224x224: 2850 ± 38 ms

8/19. SRCNN 9-5-5

8.1 - inference | batch=10, size=512x512: 21744 ± 1096 ms
8.2 - inference | batch=1, size=1536x1536: 9254 ± 2078 ms
8.3 - training  | batch=10, size=512x512: 66310.0 ± 0.0 ms

9/19. VGG-19 Super-Res

9.1 - inference | batch=10, size=256x256: 11661 ± 45 ms
9.2 - inference | batch=1, size=1024x1024: 19044 ± 213 ms
9.3 - training  | batch=10, size=224x224: 26208 ± 75 ms

10/19. ResNet-SRGAN

10.1 - inference | batch=10, size=512x512: 11413 ± 65 ms
10.2 - inference | batch=1, size=1536x1536: 10331 ± 22 ms
10.3 - training  | batch=5, size=512x512: 14918 ± 175 ms

11/19. ResNet-DPED

11.1 - inference | batch=10, size=256x256: 15302 ± 286 ms
11.2 - inference | batch=1, size=1024x1024: 23054 ± 95 ms
11.3 - training  | batch=15, size=128x128: 16276 ± 97 ms

12/19. U-Net

12.1 - inference | batch=4, size=512x512: 17697 ± 296 ms
12.2 - inference | batch=1, size=1024x1024: 17718 ± 417 ms
12.3 - training  | batch=4, size=256x256: 14775 ± 373 ms

13/19. Nvidia-SPADE

13.1 - inference | batch=5, size=128x128: 4587 ± 100 ms
13.2 - training  | batch=1, size=128x128: 5842 ± 147 ms

14/19. ICNet

14.1 - inference | batch=5, size=1024x1536: 3811 ± 61 ms
14.2 - training  | batch=10, size=1024x1536: 8543 ± 142 ms

15/19. PSPNet

15.1 - inference | batch=5, size=720x720: 18942 ± 361 ms
15.2 - training  | batch=1, size=512x512: 11080 ± 206 ms

16/19. DeepLab

16.1 - inference | batch=2, size=512x512: 4855 ± 121 ms
16.2 - training  | batch=1, size=384x384: 8205 ± 275 ms

17/19. Pixel-RNN

17.1 - inference | batch=50, size=64x64: 4452 ± 125 ms
17.2 - training  | batch=10, size=64x64: 8015 ± 113 ms

18/19. LSTM-Sentiment

18.1 - inference | batch=100, size=1024x300: 12555 ± 167 ms
18.2 - training  | batch=10, size=1024x300: 29360 ± 105 ms

19/19. GNMT-Translation

19.1 - inference | batch=1, size=1x20: 2011 ± 17 ms

Device Inference Score: 220
Device Training Score: 244
Device AI Score: 464

• GPU测试详细信息：

AI-Benchmark-v.0.1.1   
>>   Let the AI Games begin..

*  TF Version: 1.13.1
*  Platform: Windows-10-10.0.17134-SP0
*  CPU: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz
*  CPU RAM: 8 GB
*  GPU/0: GeForce GTX 1050 Ti
*  GPU RAM: 2.9 GB
*  CUDA Version: 10.0
*  CUDA Build: V10.0.130

The benchmark is running...
The tests might take up to 20 minutes
Please don't interrupt the script

1/19. MobileNet-V2

1.1 - inference | batch=50, size=224x224: 154 ± 4 ms
1.2 - training  | batch=50, size=224x224: 725 ± 20 ms

2/19. Inception-V3

2.1 - inference | batch=20, size=346x346: 241 ± 1 ms
2.2 - training  | batch=20, size=346x346: 895 ± 3 ms

3/19. Inception-V4

3.1 - inference | batch=10, size=346x346: 240 ± 1 ms
3.2 - training  | batch=10, size=346x346: 948 ± 2 ms

4/19. Inception-ResNet-V2

4.1 - inference | batch=10, size=346x346: 299 ± 1 ms
4.2 - training  | batch=8, size=346x346: 952 ± 2 ms

5/19. ResNet-V2-50

5.1 - inference | batch=10, size=346x346: 166 ± 1 ms
5.2 - training  | batch=10, size=346x346: 582 ± 1 ms

6/19. ResNet-V2-152

6.1 - inference | batch=10, size=256x256: 225 ± 1 ms
6.2 - training  | batch=10, size=256x256: 830 ± 1 ms

7/19. VGG-16

7.1 - inference | batch=20, size=224x224: 395 ± 3 ms
7.2 - training  | batch=2, size=224x224: 588 ± 4 ms

8/19. SRCNN 9-5-5

8.1 - inference | batch=10, size=512x512: 404 ± 10 ms
8.2 - inference | batch=1, size=1536x1536: 352 ± 4 ms
8.3 - training  | batch=10, size=512x512: 1050 ± 4 ms

9/19. VGG-19 Super-Res

9.1 - inference | batch=10, size=256x256: 460 ± 15 ms
9.2 - inference | batch=1, size=1024x1024: 693 ± 2 ms
9.3 - training  | batch=10, size=224x224: 1165 ± 8 ms

10/19. ResNet-SRGAN

10.1 - inference | batch=10, size=512x512: 432 ± 9 ms
10.2 - inference | batch=1, size=1536x1536: 369 ± 2 ms
10.3 - training  | batch=5, size=512x512: 674 ± 6 ms

11/19. ResNet-DPED

11.1 - inference | batch=10, size=256x256: 506 ± 5 ms
11.2 - inference | batch=1, size=1024x1024: 907 ± 2 ms
11.3 - training  | batch=15, size=128x128: 790 ± 6 ms

12/19. U-Net

12.1 - inference | batch=4, size=512x512: 930 ± 3 ms
12.2 - inference | batch=1, size=1024x1024: 926 ± 3 ms
12.3 - training  | batch=4, size=256x256: 921 ± 1 ms

13/19. Nvidia-SPADE

13.1 - inference | batch=5, size=128x128: 480 ± 2 ms
13.2 - training  | batch=1, size=128x128: 753 ± 2 ms

14/19. ICNet

14.1 - inference | batch=5, size=1024x1536: 451 ± 18 ms
14.2 - training  | batch=10, size=1024x1536: 1106 ± 12 ms

15/19. PSPNet

15.1 - inference | batch=5, size=720x720: 1914 ± 11 ms
15.2 - training  | batch=1, size=512x512: 745 ± 3 ms

16/19. DeepLab

16.1 - inference | batch=2, size=512x512: 521 ± 1 ms
16.2 - training  | batch=1, size=384x384: 602 ± 2 ms

17/19. Pixel-RNN

17.1 - inference | batch=50, size=64x64: 1041 ± 13 ms
17.2 - training  | batch=10, size=64x64: 1225 ± 76 ms

18/19. LSTM-Sentiment

18.1 - inference | batch=100, size=1024x300: 1455 ± 32 ms
18.2 - training  | batch=10, size=1024x300: 2028 ± 15 ms

19/19. GNMT-Translation

19.1 - inference | batch=1, size=1x20: 386.7 ± 0.9 ms

Device Inference Score: 2987
Device Training Score: 3392
Device AI Score: 6379