16. Performance Report

Performance statistics logging is turned on by setting a launch parameter, exportPerfStats to 1.

16.1. TDA4VM

16.1.1. ROSBAG, 15 FPS

Source: “rosbag play” and a demo ROS node are running in the ROS 1 Docker container on the target SK board. ROSBAG (zed1_2020-11-09-18-01-08.bag, 1280x720) is played back at 15 FPS.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.13

66.1128

NA

NA

18.38

694

528

1222

1.0

0.0

0.0

3.0

1.0

0

0

0

0

4.52

16.31

0

ti_sde (w/ pcl)

15.12

66.1162

NA

NA

33.25

1051

927

1978

0.0

16.0

0.0

4.0

3.0

6.74

0

0

0

4.59

21.10

0

ti_vision_cnn (semseg)

15.13

66.0911

3.0177

8.0303

11.97

525

309

834

11.0

0.0

0.0

2.0

1.0

3.27

0

0

0

2.17

0

0

ti_vision_cnn (objdet)

15.13

66.1066

3.0051

5.0025

11.47

491

278

769

8.0

1.0

0.0

2.0

1.0

3.27

0

0

0

2.16

0

0

ti_estop

15.15

66.0077

3.0026

8.0720

25.62

927

603

1530

11.0

0.0

0.0

4.0

2.0

3.29

0

0

0

4.52

16.17

0

ti_objdet_range

15.15

65.9896

3.0000

5.0415

40.89

1406

1029

2435

8.0

1.0

0.0

6.0

3.0

10.27

0

0

0

6.97

24.83

0

ti_vl

15.14

66.0387

3.0193

17.0386

30.67

1138

976

2114

18.0

68.0

2.0

2.0

1.0

1.25

0

0

0

0.79

0

0

Note: “A72 Load (%)” are for dual A72 cores in a scale of 100%. For example, 100% A72 loading means that two A72 cores are fully loaded.

16.1.2. Live ZED Camera, 15 FPS

Source: live ZED camera, 1280x720 on each of left and right image, at 15 FPS. “zed_capture” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.11

66.1753

NA

NA

25.18

923

860

1783

1.0

1.0

0.0

2.0

2.0

0

0

0

0

4.42

16.46

0

ti_sde (w/ pcl)

15.12

66.1377

NA

NA

40.75

1328

1311

2639

0.0

17.0

0.0

5.0

3.0

6.59

0

0

0

4.48

21.32

0

ti_vision_cnn (semseg)

15.12

66.1263

3.1748

8.0026

20.63

802

671

1473

11.0

0.0

0.0

2.0

1.0

3.23

0

0

0

2.15

0

0

ti_vision_cnn (objdet)

15.14

66.0699

3.0000

5.0080

20.19

761

639

1400

8.0

1.0

1.0

2.0

1.0

3.22

0

0

0

2.14

0

0

ti_estop

15.14

66.0397

3.1372

8.0871

33.33

1166

936

2102

11.0

0.0

0.0

4.0

2.0

3.34

0

0

0

4.59

16.39

0

ti_objdet_range

15.11

66.1702

3.0522

5.5039

49.50

1682

1358

3040

8.0

1.0

0.0

7.0

3.0

9.97

0

0

0

6.88

25.11

0

16.1.3. Live C920 Webcam, 30 FPS

Source: live C920 webcam, 1280x720 in MJPG mode, at 30 FPS. “gscam” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_vision_cnn (semseg)

30.48

32.8044

3.4448

8.0051

45.5

1247

772

2019

22.0

20.0

0.0

4.0

1.0

6.47

0

0

0

4.29

0

0

ti_vision_cnn (objdet)

30.46

32.8274

3.0379

5.0483

36.22

1092

664

1756

15.0

20.0

0.0

4.0

1.0

6.42

0

0

0

4.26

0

0

16.2. AM68A

16.2.1. ROSBAG, 15 FPS

Source: “rosbag play” and a demo ROS node are running in the ROS 1 Docker container on the target SK board. ROSBAG (zed1_2020-11-09-18-01-08.bag, 1280x720) is played back at 15 FPS.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.12

66.1523

NA

NA

18.25

1216

546

1762

0.0

0

0

2.0

1.0

0

0

0

0

4.27

16.10

0

ti_sde (w/ pcl)

15.12

66.1279

NA

NA

33.24

1606

912

2518

0.0

0

0

3.0

2.0

6.26

0

0

0

4.27

21.29

0

ti_vision_cnn (semseg)

15.12

66.1458

3.0102

10.0102

12.96

1836

1025

2861

15.0

0

0

2.0

1.0

3.16

0

0

0

2.10

0

0

ti_vision_cnn (objdet)

15.13

66.1079

3.0179

5.0026

10.57

1111

387

1498

7.0

0

0

1.0

1.0

3.9

0

0

0

2.6

0

0

ti_estop

15.15

66.0026

3.0182

10.0961

25.18

2267

1333

3600

15.0

0

0

2.0

1.0

3.10

0

0

0

4.24

16.44

0

ti_objdet_range

13.12

76.2331

3.0092

5.0245

40.35

2077

1180

3257

7.0

0

0

4.0

2.0

9.60

0

0

0

6.45

24.95

0

ti_vl

15.14

66.0396

3.0395

13.2006

30.15

1957

1272

3229

13.0

0

0

1.0

1.0

1.11

0

0

0

0.71

0

0

Note: “A72 Load (%)” are for dual A72 cores in a scale of 100%. For example, 100% A72 loading means that two A72 cores are fully loaded.

16.2.2. Live ZED Camera, 15 FPS

Source: live ZED camera, 1280x720 on each of left and right image, at 15 FPS. “zed_capture” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.11

66.1728

NA

NA

28.53

1464

881

2345

0.0

0

0

2.0

1.0

0

0

0

0

4.26

16.9

0

ti_sde (w/ pcl)

15.12

66.1395

NA

NA

39.84

1896

1300

3196

0.0

0

0

3.0

2.0

6.38

0

0

0

4.34

20.98

0

ti_vision_cnn (semseg)

15.12

66.1269

3.4677

10.0258

18.32

2119

1388

3507

15.0

0

0

1.0

1.0

3.12

0

0

0

2.8

0

0

ti_vision_cnn (objdet)

15.12

66.1273

3.0078

5.0052

17.67

1400

753

2153

7.0

0

0

2.0

1.0

3.15

0

0

0

2.9

0

0

ti_estop

15.15

66.0272

3.2636

10.1005

33.83

2506

1658

4164

15.0

0

0

2.0

1.0

3.17

0

0

0

4.35

16.27

0

ti_objdet_range

15.12

66.1250

3.1192

5.1518

48.73

2361

1501

3862

7.0

0

0

5.0

2.0

9.65

0

0

0

6.75

25.3

0

16.2.3. Live C920 Webcam, 30 FPS

Source: live C920 webcam, 1280x720 in MJPG mode, at 30 FPS. “gscam” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_vision_cnn (semseg)

30.50

32.7890

3.1703

10.1007

46.41

3362

2157

5519

30.0

0

0

2.0

1.0

6.14

0

0

0

4.11

0

0

ti_vision_cnn (objdet)

30.48

32.8050

3.0000

5.0039

42.43

1845

846

2691

13.0

0

0

2.0

1.0

6.15

0

0

0

4.11

0

0

16.3. AM69A

16.3.1. ROSBAG, 15 FPS

Source: “rosbag play” and a demo ROS node are running in the ROS 1 Docker container on the target SK board. ROSBAG (zed1_2020-11-09-18-01-08.bag, 1280x720) is played back at 15 FPS.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.12

66.1160

NA

NA

4.30

642

522

1164

0.0

0

0

2.0

1.0

0

0

0

0

4.28

16.11

0

ti_sde (w/ pcl)

15.13

66.0742

NA

NA

8.24

976

901

1877

0.0

0

0

3.0

2.0

6.27

0

0

0

4.27

21.30

0

ti_vision_cnn (semseg)

15.12

66.1432

3.8547

9.0025

2.74

1132

1162

2294

13.0

0

0

2.0

1.0

3.17

0

0

0

2.10

0

0

ti_vision_cnn (objdet)

15.13

66.1043

3.1114

4.0470

2.61

533

376

909

7.0

0

0

1.0

1.0

3.9

0

0

0

2.5

0

0

ti_estop

15.15

66.0227

3.7708

9.0730

6.10

1509

1456

2965

13.0

0

0

2.0

1.0

3.13

0

0

0

4.29

16.2

0

ti_objdet_range

15.15

66.0000

3.0150

4.0850

9.26

1372

1134

2506

7.0

0

0

4.0

2.0

9.62

0

0

0

6.54

24.60

0

ti_vl

15.20

65.7929

3.0136

16.3451

8.67

1344

1225

2569

12.0

0

0

2.0

1.0

1.10

0

0

0

0.70

0

0

Note: “A72 Load (%)” are for 8x A72 cores in a scale of 100%. For example, 100% A72 loading means that 8x A72 cores are fully loaded.

16.3.2. Live ZED Camera, 15 FPS

Source: live ZED camera, 1280x720 on each of left and right image, at 15 FPS. “zed_capture” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_sde

15.15

66.0203

NA

NA

5.8

1316

823

2139

0.0

0

0

2.0

1.0

0

0

0

0

4.30

16.13

0

ti_sde (w/ pcl)

15.14

66.0584

NA

NA

10.48

1717

1257

2974

0.0

0

0

3.0

3.0

6.29

0

0

0

4.31

20.68

0

ti_vision_cnn (semseg)

15.15

66.0000

4.0050

9.0025

4.36

1850

1493

3343

13.0

0

0

2.0

1.0

3.10

0

0

0

2.7

0

0

ti_vision_cnn (objdet)

15.16

65.9825

3.0000

4.9950

3.75

1250

708

1958

7.0

0

0

1.0

1.0

3.12

0

0

0

2.8

0

0

ti_estop

15.15

66.0000

4.0000

9.0799

7.76

2190

1758

3948

13.0

0

0

2.0

1.0

3.14

0

0

0

4.33

16.9

0

ti_objdet_range

15.16

65.9599

3.0000

5.0000

10.87

2075

1437

3512

7.0

0

0

5.0

2.0

9.57

0

0

0

6.53

24.44

0

16.3.3. Live C920 Webcam, 30 FPS

Source: live C920 webcam, 1280x720 in MJPG mode, at 30 FPS. “gscam” ROS node and a demo ROS node are running in the ROS 1 Docker container on the target SK board.

Demo

FPS

Total time (ms)

Preproc time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ti_vision_cnn (semseg)

30.47

32.8163

4.0168

9.0078

14.4

2926

2419

5345

25.0

0

0

3.0

1.0

6.24

0

0

0

4.32

0

0

ti_vision_cnn (objdet)

30.48

32.8105

3.0050

4.6974

9.54

1660

820

2480

13.0

0

0

2.0

1.0

6.23

0

0

0

4.37

0

0